2023-06-04 22:34:30 +02:00
|
|
|
#include <metal_stdlib>
|
|
|
|
|
|
|
|
using namespace metal;
|
|
|
|
|
|
|
|
#define MAX(x, y) ((x) > (y) ? (x) : (y))
|
2023-12-07 12:03:17 +01:00
|
|
|
#define MIN(x, y) ((x) < (y) ? (x) : (y))
|
2023-12-07 21:26:54 +01:00
|
|
|
#define SWAP(x, y) { auto tmp = (x); (x) = (y); (y) = tmp; }
|
2023-06-04 22:34:30 +02:00
|
|
|
|
|
|
|
#define QK4_0 32
|
|
|
|
#define QR4_0 2
|
|
|
|
typedef struct {
|
|
|
|
half d; // delta
|
|
|
|
uint8_t qs[QK4_0 / 2]; // nibbles / quants
|
|
|
|
} block_q4_0;
|
|
|
|
|
2023-06-10 10:28:11 +02:00
|
|
|
#define QK4_1 32
|
|
|
|
typedef struct {
|
2023-10-08 09:01:53 +02:00
|
|
|
half d; // delta
|
|
|
|
half m; // min
|
2023-06-10 10:28:11 +02:00
|
|
|
uint8_t qs[QK4_1 / 2]; // nibbles / quants
|
|
|
|
} block_q4_1;
|
|
|
|
|
2023-10-18 14:21:48 +02:00
|
|
|
#define QK5_0 32
|
|
|
|
typedef struct {
|
|
|
|
half d; // delta
|
|
|
|
uint8_t qh[4]; // 5-th bit of quants
|
|
|
|
uint8_t qs[QK5_0 / 2]; // nibbles / quants
|
|
|
|
} block_q5_0;
|
|
|
|
|
|
|
|
#define QK5_1 32
|
|
|
|
typedef struct {
|
|
|
|
half d; // delta
|
|
|
|
half m; // min
|
|
|
|
uint8_t qh[4]; // 5-th bit of quants
|
|
|
|
uint8_t qs[QK5_1 / 2]; // nibbles / quants
|
|
|
|
} block_q5_1;
|
|
|
|
|
2023-08-24 15:19:57 +02:00
|
|
|
#define QK8_0 32
|
|
|
|
typedef struct {
|
|
|
|
half d; // delta
|
|
|
|
int8_t qs[QK8_0]; // quants
|
|
|
|
} block_q8_0;
|
|
|
|
|
2023-12-01 09:51:24 +01:00
|
|
|
#define N_SIMDWIDTH 32 // assuming SIMD group size is 32
|
|
|
|
|
2023-12-07 21:26:54 +01:00
|
|
|
enum ggml_sort_order {
|
|
|
|
GGML_SORT_ASC,
|
|
|
|
GGML_SORT_DESC,
|
|
|
|
};
|
|
|
|
|
|
|
|
// general-purpose kernel for addition, multiplication and division of two tensors
|
|
|
|
// pros: works for non-contiguous tensors, supports broadcast across all dims
|
2023-09-28 18:04:36 +02:00
|
|
|
// cons: not very efficient
|
2023-06-04 22:34:30 +02:00
|
|
|
kernel void kernel_add(
|
2023-09-28 18:04:36 +02:00
|
|
|
device const char * src0,
|
|
|
|
device const char * src1,
|
|
|
|
device char * dst,
|
|
|
|
constant int64_t & ne00,
|
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant int64_t & ne03,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant uint64_t & nb00,
|
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant uint64_t & nb02,
|
|
|
|
constant uint64_t & nb03,
|
2023-09-28 18:04:36 +02:00
|
|
|
constant int64_t & ne10,
|
|
|
|
constant int64_t & ne11,
|
|
|
|
constant int64_t & ne12,
|
|
|
|
constant int64_t & ne13,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant uint64_t & nb10,
|
|
|
|
constant uint64_t & nb11,
|
|
|
|
constant uint64_t & nb12,
|
|
|
|
constant uint64_t & nb13,
|
2023-09-28 18:04:36 +02:00
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
|
|
|
constant int64_t & ne2,
|
|
|
|
constant int64_t & ne3,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant uint64_t & nb0,
|
|
|
|
constant uint64_t & nb1,
|
|
|
|
constant uint64_t & nb2,
|
|
|
|
constant uint64_t & nb3,
|
2023-12-13 20:54:54 +01:00
|
|
|
constant int64_t & offs,
|
2023-09-28 18:04:36 +02:00
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint3 tpitg[[thread_position_in_threadgroup]],
|
|
|
|
uint3 ntg[[threads_per_threadgroup]]) {
|
|
|
|
const int64_t i03 = tgpig.z;
|
|
|
|
const int64_t i02 = tgpig.y;
|
|
|
|
const int64_t i01 = tgpig.x;
|
|
|
|
|
|
|
|
const int64_t i13 = i03 % ne13;
|
|
|
|
const int64_t i12 = i02 % ne12;
|
|
|
|
const int64_t i11 = i01 % ne11;
|
|
|
|
|
2023-12-13 20:54:54 +01:00
|
|
|
device const char * src0_ptr = src0 + i03*nb03 + i02*nb02 + i01*nb01 + offs;
|
2023-12-07 21:26:54 +01:00
|
|
|
device const char * src1_ptr = src1 + i13*nb13 + i12*nb12 + i11*nb11;
|
2023-12-13 20:54:54 +01:00
|
|
|
device char * dst_ptr = dst + i03*nb3 + i02*nb2 + i01*nb1 + offs;
|
2023-09-28 18:04:36 +02:00
|
|
|
|
|
|
|
for (int i0 = tpitg.x; i0 < ne0; i0 += ntg.x) {
|
2023-12-07 21:26:54 +01:00
|
|
|
const int i10 = i0 % ne10;
|
|
|
|
*((device float *)(dst_ptr + i0*nb0)) = *((device float *)(src0_ptr + i0*nb00)) + *((device float *)(src1_ptr + i10*nb10));
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
kernel void kernel_mul(
|
|
|
|
device const char * src0,
|
|
|
|
device const char * src1,
|
|
|
|
device char * dst,
|
|
|
|
constant int64_t & ne00,
|
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant int64_t & ne03,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant uint64_t & nb00,
|
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant uint64_t & nb02,
|
|
|
|
constant uint64_t & nb03,
|
2023-12-07 21:26:54 +01:00
|
|
|
constant int64_t & ne10,
|
|
|
|
constant int64_t & ne11,
|
|
|
|
constant int64_t & ne12,
|
|
|
|
constant int64_t & ne13,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant uint64_t & nb10,
|
|
|
|
constant uint64_t & nb11,
|
|
|
|
constant uint64_t & nb12,
|
|
|
|
constant uint64_t & nb13,
|
2023-12-07 21:26:54 +01:00
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
|
|
|
constant int64_t & ne2,
|
|
|
|
constant int64_t & ne3,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant uint64_t & nb0,
|
|
|
|
constant uint64_t & nb1,
|
|
|
|
constant uint64_t & nb2,
|
|
|
|
constant uint64_t & nb3,
|
2023-12-07 21:26:54 +01:00
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint3 tpitg[[thread_position_in_threadgroup]],
|
|
|
|
uint3 ntg[[threads_per_threadgroup]]) {
|
|
|
|
const int64_t i03 = tgpig.z;
|
|
|
|
const int64_t i02 = tgpig.y;
|
|
|
|
const int64_t i01 = tgpig.x;
|
2023-09-28 18:04:36 +02:00
|
|
|
|
2023-12-07 21:26:54 +01:00
|
|
|
const int64_t i13 = i03 % ne13;
|
|
|
|
const int64_t i12 = i02 % ne12;
|
|
|
|
const int64_t i11 = i01 % ne11;
|
|
|
|
|
|
|
|
device const char * src0_ptr = src0 + i03*nb03 + i02*nb02 + i01*nb01;
|
|
|
|
device const char * src1_ptr = src1 + i13*nb13 + i12*nb12 + i11*nb11;
|
|
|
|
device char * dst_ptr = dst + i03*nb3 + i02*nb2 + i01*nb1;
|
|
|
|
|
|
|
|
for (int i0 = tpitg.x; i0 < ne0; i0 += ntg.x) {
|
|
|
|
const int i10 = i0 % ne10;
|
|
|
|
*((device float *)(dst_ptr + i0*nb0)) = *((device float *)(src0_ptr + i0*nb00)) * *((device float *)(src1_ptr + i10*nb10));
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
kernel void kernel_div(
|
|
|
|
device const char * src0,
|
|
|
|
device const char * src1,
|
|
|
|
device char * dst,
|
|
|
|
constant int64_t & ne00,
|
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant int64_t & ne03,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant uint64_t & nb00,
|
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant uint64_t & nb02,
|
|
|
|
constant uint64_t & nb03,
|
2023-12-07 21:26:54 +01:00
|
|
|
constant int64_t & ne10,
|
|
|
|
constant int64_t & ne11,
|
|
|
|
constant int64_t & ne12,
|
|
|
|
constant int64_t & ne13,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant uint64_t & nb10,
|
|
|
|
constant uint64_t & nb11,
|
|
|
|
constant uint64_t & nb12,
|
|
|
|
constant uint64_t & nb13,
|
2023-12-07 21:26:54 +01:00
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
|
|
|
constant int64_t & ne2,
|
|
|
|
constant int64_t & ne3,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant uint64_t & nb0,
|
|
|
|
constant uint64_t & nb1,
|
|
|
|
constant uint64_t & nb2,
|
|
|
|
constant uint64_t & nb3,
|
2023-12-07 21:26:54 +01:00
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint3 tpitg[[thread_position_in_threadgroup]],
|
|
|
|
uint3 ntg[[threads_per_threadgroup]]) {
|
|
|
|
const int64_t i03 = tgpig.z;
|
|
|
|
const int64_t i02 = tgpig.y;
|
|
|
|
const int64_t i01 = tgpig.x;
|
|
|
|
|
|
|
|
const int64_t i13 = i03 % ne13;
|
|
|
|
const int64_t i12 = i02 % ne12;
|
|
|
|
const int64_t i11 = i01 % ne11;
|
|
|
|
|
|
|
|
device const char * src0_ptr = src0 + i03*nb03 + i02*nb02 + i01*nb01;
|
|
|
|
device const char * src1_ptr = src1 + i13*nb13 + i12*nb12 + i11*nb11;
|
|
|
|
device char * dst_ptr = dst + i03*nb3 + i02*nb2 + i01*nb1;
|
|
|
|
|
|
|
|
for (int i0 = tpitg.x; i0 < ne0; i0 += ntg.x) {
|
|
|
|
const int i10 = i0 % ne10;
|
|
|
|
*((device float *)(dst_ptr + i0*nb0)) = *((device float *)(src0_ptr + i0*nb00)) / *((device float *)(src1_ptr + i10*nb10));
|
2023-09-28 18:04:36 +02:00
|
|
|
}
|
2023-06-04 22:34:30 +02:00
|
|
|
}
|
|
|
|
|
2023-07-23 13:00:37 +02:00
|
|
|
// assumption: src1 is a row
|
|
|
|
// broadcast src1 into src0
|
|
|
|
kernel void kernel_add_row(
|
2023-09-01 12:42:41 +02:00
|
|
|
device const float4 * src0,
|
|
|
|
device const float4 * src1,
|
|
|
|
device float4 * dst,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant uint64_t & nb [[buffer(28)]],
|
2023-07-23 13:00:37 +02:00
|
|
|
uint tpig[[thread_position_in_grid]]) {
|
2023-09-01 12:42:41 +02:00
|
|
|
dst[tpig] = src0[tpig] + src1[tpig % nb];
|
2023-07-23 13:00:37 +02:00
|
|
|
}
|
|
|
|
|
2023-12-07 21:26:54 +01:00
|
|
|
kernel void kernel_mul_row(
|
2023-09-01 12:42:41 +02:00
|
|
|
device const float4 * src0,
|
|
|
|
device const float4 * src1,
|
|
|
|
device float4 * dst,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant uint64_t & nb [[buffer(28)]],
|
2023-06-04 22:34:30 +02:00
|
|
|
uint tpig[[thread_position_in_grid]]) {
|
2023-12-07 21:26:54 +01:00
|
|
|
dst[tpig] = src0[tpig] * src1[tpig % nb];
|
2023-06-04 22:34:30 +02:00
|
|
|
}
|
|
|
|
|
2023-12-07 21:26:54 +01:00
|
|
|
kernel void kernel_div_row(
|
2023-09-01 12:42:41 +02:00
|
|
|
device const float4 * src0,
|
|
|
|
device const float4 * src1,
|
|
|
|
device float4 * dst,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant uint64_t & nb [[buffer(28)]],
|
2023-06-04 22:34:30 +02:00
|
|
|
uint tpig[[thread_position_in_grid]]) {
|
2023-12-07 21:26:54 +01:00
|
|
|
dst[tpig] = src0[tpig] / src1[tpig % nb];
|
2023-06-04 22:34:30 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
kernel void kernel_scale(
|
2023-10-24 08:46:50 +02:00
|
|
|
device const float * src0,
|
|
|
|
device float * dst,
|
|
|
|
constant float & scale,
|
|
|
|
uint tpig[[thread_position_in_grid]]) {
|
|
|
|
dst[tpig] = src0[tpig] * scale;
|
|
|
|
}
|
|
|
|
|
|
|
|
kernel void kernel_scale_4(
|
2023-09-11 09:30:11 +02:00
|
|
|
device const float4 * src0,
|
|
|
|
device float4 * dst,
|
2023-10-24 08:46:50 +02:00
|
|
|
constant float & scale,
|
2023-06-04 22:34:30 +02:00
|
|
|
uint tpig[[thread_position_in_grid]]) {
|
|
|
|
dst[tpig] = src0[tpig] * scale;
|
|
|
|
}
|
|
|
|
|
2023-12-13 20:54:54 +01:00
|
|
|
kernel void kernel_relu(
|
|
|
|
device const float * src0,
|
|
|
|
device float * dst,
|
2023-06-04 22:34:30 +02:00
|
|
|
uint tpig[[thread_position_in_grid]]) {
|
2023-12-13 20:54:54 +01:00
|
|
|
dst[tpig] = max(0.0f, src0[tpig]);
|
2023-06-04 22:34:30 +02:00
|
|
|
}
|
|
|
|
|
2023-12-13 20:54:54 +01:00
|
|
|
kernel void kernel_tanh(
|
2023-06-04 22:34:30 +02:00
|
|
|
device const float * src0,
|
|
|
|
device float * dst,
|
|
|
|
uint tpig[[thread_position_in_grid]]) {
|
2023-12-13 20:54:54 +01:00
|
|
|
device const float & x = src0[tpig];
|
|
|
|
dst[tpig] = precise::tanh(x);
|
|
|
|
}
|
|
|
|
|
|
|
|
constant float GELU_COEF_A = 0.044715f;
|
|
|
|
constant float GELU_QUICK_COEF = -1.702f;
|
|
|
|
constant float SQRT_2_OVER_PI = 0.79788456080286535587989211986876f;
|
|
|
|
|
|
|
|
kernel void kernel_gelu(
|
|
|
|
device const float4 * src0,
|
|
|
|
device float4 * dst,
|
|
|
|
uint tpig[[thread_position_in_grid]]) {
|
|
|
|
device const float4 & x = src0[tpig];
|
|
|
|
|
|
|
|
// BEWARE !!!
|
|
|
|
// Simply using "tanh" instead of "precise::tanh" will sometimes results in NaNs!
|
|
|
|
// This was observed with Falcon 7B and 40B models
|
|
|
|
//
|
|
|
|
dst[tpig] = 0.5f*x*(1.0f + precise::tanh(SQRT_2_OVER_PI*x*(1.0f + GELU_COEF_A*x*x)));
|
|
|
|
}
|
|
|
|
|
|
|
|
kernel void kernel_gelu_quick(
|
|
|
|
device const float4 * src0,
|
|
|
|
device float4 * dst,
|
|
|
|
uint tpig[[thread_position_in_grid]]) {
|
|
|
|
device const float4 & x = src0[tpig];
|
|
|
|
|
|
|
|
dst[tpig] = x*(1.0f/(1.0f+exp(GELU_QUICK_COEF*x)));
|
|
|
|
}
|
|
|
|
|
|
|
|
kernel void kernel_silu(
|
|
|
|
device const float4 * src0,
|
|
|
|
device float4 * dst,
|
|
|
|
uint tpig[[thread_position_in_grid]]) {
|
|
|
|
device const float4 & x = src0[tpig];
|
|
|
|
dst[tpig] = x / (1.0f + exp(-x));
|
2023-06-04 22:34:30 +02:00
|
|
|
}
|
|
|
|
|
2023-10-07 09:12:43 +02:00
|
|
|
kernel void kernel_sqr(
|
|
|
|
device const float * src0,
|
|
|
|
device float * dst,
|
|
|
|
uint tpig[[thread_position_in_grid]]) {
|
|
|
|
dst[tpig] = src0[tpig] * src0[tpig];
|
|
|
|
}
|
|
|
|
|
2023-12-07 21:26:54 +01:00
|
|
|
kernel void kernel_sum_rows(
|
|
|
|
device const float * src0,
|
|
|
|
device float * dst,
|
|
|
|
constant int64_t & ne00,
|
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant int64_t & ne03,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant uint64_t & nb00,
|
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant uint64_t & nb02,
|
|
|
|
constant uint64_t & nb03,
|
2023-12-07 21:26:54 +01:00
|
|
|
constant int64_t & ne10,
|
|
|
|
constant int64_t & ne11,
|
|
|
|
constant int64_t & ne12,
|
|
|
|
constant int64_t & ne13,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant uint64_t & nb10,
|
|
|
|
constant uint64_t & nb11,
|
|
|
|
constant uint64_t & nb12,
|
|
|
|
constant uint64_t & nb13,
|
2023-12-07 21:26:54 +01:00
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
|
|
|
constant int64_t & ne2,
|
|
|
|
constant int64_t & ne3,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant uint64_t & nb0,
|
|
|
|
constant uint64_t & nb1,
|
|
|
|
constant uint64_t & nb2,
|
|
|
|
constant uint64_t & nb3,
|
2023-12-07 21:26:54 +01:00
|
|
|
uint3 tpig[[thread_position_in_grid]]) {
|
|
|
|
int64_t i3 = tpig.z;
|
|
|
|
int64_t i2 = tpig.y;
|
|
|
|
int64_t i1 = tpig.x;
|
|
|
|
|
|
|
|
if (i3 >= ne03 || i2 >= ne02 || i1 >= ne01) {
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
device const float * src_row = (device const float *) ((device const char *) src0 + i1*nb01 + i2*nb02 + i3*nb03);
|
|
|
|
device float * dst_row = (device float *) ((device char *) dst + i1*nb1 + i2*nb2 + i3*nb3);
|
|
|
|
|
|
|
|
float row_sum = 0;
|
|
|
|
|
|
|
|
for (int64_t i0 = 0; i0 < ne00; i0++) {
|
|
|
|
row_sum += src_row[i0];
|
|
|
|
}
|
|
|
|
|
|
|
|
dst_row[0] = row_sum;
|
|
|
|
}
|
|
|
|
|
2023-06-04 22:34:30 +02:00
|
|
|
kernel void kernel_soft_max(
|
|
|
|
device const float * src0,
|
2023-12-01 09:51:24 +01:00
|
|
|
device const float * src1,
|
2023-06-04 22:34:30 +02:00
|
|
|
device float * dst,
|
|
|
|
constant int64_t & ne00,
|
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
2023-12-01 09:51:24 +01:00
|
|
|
constant float & scale,
|
2023-11-01 20:25:00 +01:00
|
|
|
threadgroup float * buf [[threadgroup(0)]],
|
|
|
|
uint tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint tpitg[[thread_position_in_threadgroup]],
|
|
|
|
uint sgitg[[simdgroup_index_in_threadgroup]],
|
|
|
|
uint tiisg[[thread_index_in_simdgroup]],
|
|
|
|
uint ntg[[threads_per_threadgroup]]) {
|
|
|
|
const int64_t i03 = (tgpig) / (ne02*ne01);
|
|
|
|
const int64_t i02 = (tgpig - i03*ne02*ne01) / ne01;
|
|
|
|
const int64_t i01 = (tgpig - i03*ne02*ne01 - i02*ne01);
|
2023-06-04 22:34:30 +02:00
|
|
|
|
2023-12-13 13:04:25 +01:00
|
|
|
device const float * psrc0 = src0 + i03*ne02*ne01*ne00 + i02*ne01*ne00 + i01*ne00;
|
|
|
|
device const float * pmask = src1 != src0 ? src1 + i01*ne00 : nullptr;
|
|
|
|
device float * pdst = dst + i03*ne02*ne01*ne00 + i02*ne01*ne00 + i01*ne00;
|
2023-06-04 22:34:30 +02:00
|
|
|
|
|
|
|
// parallel max
|
2023-12-01 09:51:24 +01:00
|
|
|
float lmax = -INFINITY;
|
2023-11-01 20:25:00 +01:00
|
|
|
|
2023-12-01 09:51:24 +01:00
|
|
|
for (int i00 = tpitg; i00 < ne00; i00 += ntg) {
|
|
|
|
lmax = MAX(lmax, psrc0[i00]*scale + (pmask ? pmask[i00] : 0.0f));
|
2023-06-04 22:34:30 +02:00
|
|
|
}
|
2023-11-01 20:25:00 +01:00
|
|
|
|
2023-12-01 09:51:24 +01:00
|
|
|
// find the max value in the block
|
|
|
|
float max_val = simd_max(lmax);
|
|
|
|
if (ntg > N_SIMDWIDTH) {
|
|
|
|
if (sgitg == 0) {
|
|
|
|
buf[tiisg] = -INFINITY;
|
|
|
|
}
|
2023-11-01 20:25:00 +01:00
|
|
|
|
2023-12-01 09:51:24 +01:00
|
|
|
threadgroup_barrier(mem_flags::mem_threadgroup);
|
2023-11-01 20:25:00 +01:00
|
|
|
|
2023-12-01 09:51:24 +01:00
|
|
|
if (tiisg == 0) {
|
|
|
|
buf[sgitg] = max_val;
|
|
|
|
}
|
2023-11-01 20:25:00 +01:00
|
|
|
|
2023-12-01 09:51:24 +01:00
|
|
|
threadgroup_barrier(mem_flags::mem_threadgroup);
|
2023-11-01 20:25:00 +01:00
|
|
|
|
2023-12-01 09:51:24 +01:00
|
|
|
max_val = buf[tiisg];
|
|
|
|
max_val = simd_max(max_val);
|
|
|
|
}
|
2023-06-04 22:34:30 +02:00
|
|
|
|
|
|
|
// parallel sum
|
2023-09-11 09:30:11 +02:00
|
|
|
float lsum = 0.0f;
|
2023-11-01 20:25:00 +01:00
|
|
|
for (int i00 = tpitg; i00 < ne00; i00 += ntg) {
|
2023-12-01 09:51:24 +01:00
|
|
|
const float exp_psrc0 = exp((psrc0[i00]*scale + (pmask ? pmask[i00] : 0.0f)) - max_val);
|
2023-09-11 09:30:11 +02:00
|
|
|
lsum += exp_psrc0;
|
2023-09-03 10:06:22 +02:00
|
|
|
pdst[i00] = exp_psrc0;
|
2023-06-04 22:34:30 +02:00
|
|
|
}
|
|
|
|
|
2023-12-13 13:04:25 +01:00
|
|
|
// This barrier fixes a failing test
|
|
|
|
// ref: https://github.com/ggerganov/ggml/pull/621#discussion_r1425156335
|
|
|
|
threadgroup_barrier(mem_flags::mem_none);
|
|
|
|
|
2023-11-01 20:25:00 +01:00
|
|
|
float sum = simd_sum(lsum);
|
2023-12-13 13:04:25 +01:00
|
|
|
|
2023-12-01 09:51:24 +01:00
|
|
|
if (ntg > N_SIMDWIDTH) {
|
|
|
|
if (sgitg == 0) {
|
|
|
|
buf[tiisg] = 0.0f;
|
|
|
|
}
|
2023-11-01 20:25:00 +01:00
|
|
|
|
2023-12-01 09:51:24 +01:00
|
|
|
threadgroup_barrier(mem_flags::mem_threadgroup);
|
2023-11-01 20:25:00 +01:00
|
|
|
|
2023-12-01 09:51:24 +01:00
|
|
|
if (tiisg == 0) {
|
|
|
|
buf[sgitg] = sum;
|
|
|
|
}
|
2023-11-01 20:25:00 +01:00
|
|
|
|
2023-12-01 09:51:24 +01:00
|
|
|
threadgroup_barrier(mem_flags::mem_threadgroup);
|
|
|
|
|
|
|
|
sum = buf[tiisg];
|
|
|
|
sum = simd_sum(sum);
|
|
|
|
}
|
2023-11-01 20:25:00 +01:00
|
|
|
|
2023-12-01 09:51:24 +01:00
|
|
|
const float inv_sum = 1.0f/sum;
|
2023-09-11 09:30:11 +02:00
|
|
|
|
2023-11-01 20:25:00 +01:00
|
|
|
for (int i00 = tpitg; i00 < ne00; i00 += ntg) {
|
2023-12-01 09:51:24 +01:00
|
|
|
pdst[i00] *= inv_sum;
|
2023-06-04 22:34:30 +02:00
|
|
|
}
|
2023-09-11 09:30:11 +02:00
|
|
|
}
|
2023-06-04 22:34:30 +02:00
|
|
|
|
2023-09-11 09:30:11 +02:00
|
|
|
kernel void kernel_soft_max_4(
|
|
|
|
device const float * src0,
|
2023-12-01 09:51:24 +01:00
|
|
|
device const float * src1,
|
2023-09-11 09:30:11 +02:00
|
|
|
device float * dst,
|
|
|
|
constant int64_t & ne00,
|
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
2023-12-01 09:51:24 +01:00
|
|
|
constant float & scale,
|
2023-11-01 20:25:00 +01:00
|
|
|
threadgroup float * buf [[threadgroup(0)]],
|
|
|
|
uint tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint tpitg[[thread_position_in_threadgroup]],
|
|
|
|
uint sgitg[[simdgroup_index_in_threadgroup]],
|
|
|
|
uint tiisg[[thread_index_in_simdgroup]],
|
|
|
|
uint ntg[[threads_per_threadgroup]]) {
|
|
|
|
const int64_t i03 = (tgpig) / (ne02*ne01);
|
|
|
|
const int64_t i02 = (tgpig - i03*ne02*ne01) / ne01;
|
|
|
|
const int64_t i01 = (tgpig - i03*ne02*ne01 - i02*ne01);
|
2023-06-04 22:34:30 +02:00
|
|
|
|
2023-12-13 13:04:25 +01:00
|
|
|
device const float4 * psrc4 = (device const float4 *)(src0 + i03*ne02*ne01*ne00 + i02*ne01*ne00 + i01*ne00);
|
|
|
|
device const float4 * pmask = src1 != src0 ? (device const float4 *)(src1 + i01*ne00) : nullptr;
|
|
|
|
device float4 * pdst4 = (device float4 *)(dst + i03*ne02*ne01*ne00 + i02*ne01*ne00 + i01*ne00);
|
2023-06-04 22:34:30 +02:00
|
|
|
|
2023-09-11 09:30:11 +02:00
|
|
|
// parallel max
|
2023-12-01 09:51:24 +01:00
|
|
|
float4 lmax4 = -INFINITY;
|
2023-11-01 20:25:00 +01:00
|
|
|
|
2023-12-01 09:51:24 +01:00
|
|
|
for (int i00 = tpitg; i00 < ne00/4; i00 += ntg) {
|
|
|
|
lmax4 = fmax(lmax4, psrc4[i00]*scale + (pmask ? pmask[i00] : 0.0f));
|
2023-09-11 09:30:11 +02:00
|
|
|
}
|
2023-06-04 22:34:30 +02:00
|
|
|
|
2023-11-01 20:25:00 +01:00
|
|
|
const float lmax = MAX(MAX(lmax4[0], lmax4[1]), MAX(lmax4[2], lmax4[3]));
|
|
|
|
|
2023-12-01 09:51:24 +01:00
|
|
|
float max_val = simd_max(lmax);
|
|
|
|
if (ntg > N_SIMDWIDTH) {
|
|
|
|
if (sgitg == 0) {
|
|
|
|
buf[tiisg] = -INFINITY;
|
|
|
|
}
|
2023-11-01 20:25:00 +01:00
|
|
|
|
2023-12-01 09:51:24 +01:00
|
|
|
threadgroup_barrier(mem_flags::mem_threadgroup);
|
2023-11-01 20:25:00 +01:00
|
|
|
|
2023-12-01 09:51:24 +01:00
|
|
|
if (tiisg == 0) {
|
|
|
|
buf[sgitg] = max_val;
|
|
|
|
}
|
|
|
|
|
|
|
|
threadgroup_barrier(mem_flags::mem_threadgroup);
|
2023-11-01 20:25:00 +01:00
|
|
|
|
2023-12-01 09:51:24 +01:00
|
|
|
max_val = buf[tiisg];
|
|
|
|
max_val = simd_max(max_val);
|
|
|
|
}
|
2023-09-11 09:30:11 +02:00
|
|
|
|
|
|
|
// parallel sum
|
|
|
|
float4 lsum4 = 0.0f;
|
2023-11-01 20:25:00 +01:00
|
|
|
for (int i00 = tpitg; i00 < ne00/4; i00 += ntg) {
|
2023-12-01 09:51:24 +01:00
|
|
|
const float4 exp_psrc4 = exp((psrc4[i00]*scale + (pmask ? pmask[i00] : 0.0f)) - max_val);
|
2023-09-11 09:30:11 +02:00
|
|
|
lsum4 += exp_psrc4;
|
|
|
|
pdst4[i00] = exp_psrc4;
|
|
|
|
}
|
|
|
|
|
2023-11-01 20:25:00 +01:00
|
|
|
const float lsum = lsum4[0] + lsum4[1] + lsum4[2] + lsum4[3];
|
2023-12-13 13:04:25 +01:00
|
|
|
|
|
|
|
// This barrier fixes a failing test
|
|
|
|
// ref: https://github.com/ggerganov/ggml/pull/621#discussion_r1425156335
|
|
|
|
threadgroup_barrier(mem_flags::mem_none);
|
|
|
|
|
2023-11-01 20:25:00 +01:00
|
|
|
float sum = simd_sum(lsum);
|
2023-12-13 13:04:25 +01:00
|
|
|
|
2023-12-01 09:51:24 +01:00
|
|
|
if (ntg > N_SIMDWIDTH) {
|
|
|
|
if (sgitg == 0) {
|
|
|
|
buf[tiisg] = 0.0f;
|
|
|
|
}
|
2023-11-01 20:25:00 +01:00
|
|
|
|
2023-12-01 09:51:24 +01:00
|
|
|
threadgroup_barrier(mem_flags::mem_threadgroup);
|
2023-11-01 20:25:00 +01:00
|
|
|
|
2023-12-01 09:51:24 +01:00
|
|
|
if (tiisg == 0) {
|
|
|
|
buf[sgitg] = sum;
|
|
|
|
}
|
2023-11-01 20:25:00 +01:00
|
|
|
|
2023-12-01 09:51:24 +01:00
|
|
|
threadgroup_barrier(mem_flags::mem_threadgroup);
|
|
|
|
|
|
|
|
sum = buf[tiisg];
|
|
|
|
sum = simd_sum(sum);
|
|
|
|
}
|
2023-09-11 09:30:11 +02:00
|
|
|
|
2023-12-01 09:51:24 +01:00
|
|
|
const float inv_sum = 1.0f/sum;
|
2023-11-01 20:25:00 +01:00
|
|
|
|
|
|
|
for (int i00 = tpitg; i00 < ne00/4; i00 += ntg) {
|
2023-12-01 09:51:24 +01:00
|
|
|
pdst4[i00] *= inv_sum;
|
2023-06-04 22:34:30 +02:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
kernel void kernel_diag_mask_inf(
|
|
|
|
device const float * src0,
|
|
|
|
device float * dst,
|
|
|
|
constant int64_t & ne00,
|
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int & n_past,
|
|
|
|
uint3 tpig[[thread_position_in_grid]]) {
|
|
|
|
const int64_t i02 = tpig[2];
|
|
|
|
const int64_t i01 = tpig[1];
|
|
|
|
const int64_t i00 = tpig[0];
|
|
|
|
|
|
|
|
if (i00 > n_past + i01) {
|
|
|
|
dst[i02*ne01*ne00 + i01*ne00 + i00] = -INFINITY;
|
|
|
|
} else {
|
|
|
|
dst[i02*ne01*ne00 + i01*ne00 + i00] = src0[i02*ne01*ne00 + i01*ne00 + i00];
|
2023-11-01 20:25:00 +01:00
|
|
|
}
|
2023-09-11 09:30:11 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
kernel void kernel_diag_mask_inf_8(
|
|
|
|
device const float4 * src0,
|
|
|
|
device float4 * dst,
|
|
|
|
constant int64_t & ne00,
|
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int & n_past,
|
|
|
|
uint3 tpig[[thread_position_in_grid]]) {
|
|
|
|
|
|
|
|
const int64_t i = 2*tpig[0];
|
|
|
|
|
|
|
|
dst[i+0] = src0[i+0];
|
|
|
|
dst[i+1] = src0[i+1];
|
|
|
|
int64_t i4 = 4*i;
|
|
|
|
const int64_t i02 = i4/(ne00*ne01); i4 -= i02*ne00*ne01;
|
|
|
|
const int64_t i01 = i4/(ne00); i4 -= i01*ne00;
|
|
|
|
const int64_t i00 = i4;
|
|
|
|
for (int k = 3; k >= 0; --k) {
|
|
|
|
if (i00 + 4 + k <= n_past + i01) {
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
dst[i+1][k] = -INFINITY;
|
|
|
|
if (i00 + k > n_past + i01) {
|
|
|
|
dst[i][k] = -INFINITY;
|
|
|
|
}
|
2023-06-04 22:34:30 +02:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2023-06-17 16:37:49 +02:00
|
|
|
kernel void kernel_norm(
|
|
|
|
device const void * src0,
|
|
|
|
device float * dst,
|
|
|
|
constant int64_t & ne00,
|
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant float & eps,
|
|
|
|
threadgroup float * sum [[threadgroup(0)]],
|
|
|
|
uint tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint tpitg[[thread_position_in_threadgroup]],
|
|
|
|
uint ntg[[threads_per_threadgroup]]) {
|
|
|
|
device const float * x = (device const float *) ((device const char *) src0 + tgpig*nb01);
|
|
|
|
// MEAN
|
|
|
|
// parallel sum
|
|
|
|
sum[tpitg] = 0.0f;
|
|
|
|
for (int i00 = tpitg; i00 < ne00; i00 += ntg) {
|
|
|
|
sum[tpitg] += x[i00];
|
|
|
|
}
|
|
|
|
// reduce
|
|
|
|
threadgroup_barrier(mem_flags::mem_threadgroup);
|
|
|
|
for (uint i = ntg/2; i > 0; i /= 2) {
|
|
|
|
if (tpitg < i) {
|
|
|
|
sum[tpitg] += sum[tpitg + i];
|
|
|
|
}
|
|
|
|
threadgroup_barrier(mem_flags::mem_threadgroup);
|
|
|
|
}
|
2023-09-07 15:42:42 +02:00
|
|
|
const float mean = sum[0] / ne00;
|
2023-06-17 16:37:49 +02:00
|
|
|
|
2023-09-07 15:42:42 +02:00
|
|
|
// recenter and VARIANCE
|
|
|
|
threadgroup_barrier(mem_flags::mem_threadgroup);
|
2023-06-17 16:37:49 +02:00
|
|
|
device float * y = dst + tgpig*ne00;
|
2023-09-07 14:49:09 +02:00
|
|
|
sum[tpitg] = 0.0f;
|
|
|
|
for (int i00 = tpitg; i00 < ne00; i00 += ntg) {
|
2023-09-07 15:42:42 +02:00
|
|
|
y[i00] = x[i00] - mean;
|
2023-06-17 16:37:49 +02:00
|
|
|
sum[tpitg] += y[i00] * y[i00];
|
|
|
|
}
|
2023-09-03 10:06:22 +02:00
|
|
|
|
2023-06-17 16:37:49 +02:00
|
|
|
// reduce
|
|
|
|
threadgroup_barrier(mem_flags::mem_threadgroup);
|
|
|
|
for (uint i = ntg/2; i > 0; i /= 2) {
|
|
|
|
if (tpitg < i) {
|
|
|
|
sum[tpitg] += sum[tpitg + i];
|
|
|
|
}
|
|
|
|
threadgroup_barrier(mem_flags::mem_threadgroup);
|
|
|
|
}
|
2023-09-07 15:42:42 +02:00
|
|
|
const float variance = sum[0] / ne00;
|
2023-06-17 16:37:49 +02:00
|
|
|
|
|
|
|
const float scale = 1.0f/sqrt(variance + eps);
|
|
|
|
for (int i00 = tpitg; i00 < ne00; i00 += ntg) {
|
|
|
|
y[i00] = y[i00] * scale;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2023-06-04 22:34:30 +02:00
|
|
|
kernel void kernel_rms_norm(
|
|
|
|
device const void * src0,
|
|
|
|
device float * dst,
|
|
|
|
constant int64_t & ne00,
|
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant float & eps,
|
2023-12-01 09:51:24 +01:00
|
|
|
threadgroup float * buf [[threadgroup(0)]],
|
2023-06-04 22:34:30 +02:00
|
|
|
uint tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint tpitg[[thread_position_in_threadgroup]],
|
2023-07-20 12:32:22 +02:00
|
|
|
uint sgitg[[simdgroup_index_in_threadgroup]],
|
|
|
|
uint tiisg[[thread_index_in_simdgroup]],
|
2023-06-04 22:34:30 +02:00
|
|
|
uint ntg[[threads_per_threadgroup]]) {
|
2023-12-01 09:51:24 +01:00
|
|
|
device const float4 * x = (device const float4 *) ((device const char *) src0 + tgpig*nb01);
|
2023-10-09 13:32:17 +02:00
|
|
|
|
|
|
|
float4 sumf = 0;
|
|
|
|
float all_sum = 0;
|
2023-06-04 22:34:30 +02:00
|
|
|
|
|
|
|
// parallel sum
|
2023-07-20 12:32:22 +02:00
|
|
|
for (int i00 = tpitg; i00 < ne00/4; i00 += ntg) {
|
|
|
|
sumf += x[i00] * x[i00];
|
|
|
|
}
|
|
|
|
all_sum = sumf[0] + sumf[1] + sumf[2] + sumf[3];
|
|
|
|
all_sum = simd_sum(all_sum);
|
2023-12-01 09:51:24 +01:00
|
|
|
if (ntg > N_SIMDWIDTH) {
|
|
|
|
if (sgitg == 0) {
|
|
|
|
buf[tiisg] = 0.0f;
|
|
|
|
}
|
2023-06-04 22:34:30 +02:00
|
|
|
|
2023-12-01 09:51:24 +01:00
|
|
|
threadgroup_barrier(mem_flags::mem_threadgroup);
|
2023-10-09 13:32:17 +02:00
|
|
|
|
2023-12-01 09:51:24 +01:00
|
|
|
if (tiisg == 0) {
|
|
|
|
buf[sgitg] = all_sum;
|
2023-10-09 13:32:17 +02:00
|
|
|
}
|
2023-06-04 22:34:30 +02:00
|
|
|
|
2023-12-01 09:51:24 +01:00
|
|
|
threadgroup_barrier(mem_flags::mem_threadgroup);
|
|
|
|
|
|
|
|
all_sum = buf[tiisg];
|
|
|
|
all_sum = simd_sum(all_sum);
|
|
|
|
}
|
2023-06-04 22:34:30 +02:00
|
|
|
|
2023-12-01 09:51:24 +01:00
|
|
|
const float mean = all_sum/ne00;
|
2023-06-04 22:34:30 +02:00
|
|
|
const float scale = 1.0f/sqrt(mean + eps);
|
|
|
|
|
2023-07-20 12:32:22 +02:00
|
|
|
device float4 * y = (device float4 *) (dst + tgpig*ne00);
|
|
|
|
for (int i00 = tpitg; i00 < ne00/4; i00 += ntg) {
|
2023-06-04 22:34:30 +02:00
|
|
|
y[i00] = x[i00] * scale;
|
|
|
|
}
|
2023-07-20 12:32:22 +02:00
|
|
|
}
|
|
|
|
|
2023-12-13 20:54:54 +01:00
|
|
|
kernel void kernel_group_norm(
|
|
|
|
device const float * src0,
|
|
|
|
device float * dst,
|
|
|
|
constant int64_t & ne00,
|
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant uint64_t & nb00,
|
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant uint64_t & nb02,
|
|
|
|
constant int32_t & n_groups,
|
|
|
|
constant float & eps,
|
|
|
|
threadgroup float * buf [[threadgroup(0)]],
|
|
|
|
uint tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint tpitg[[thread_position_in_threadgroup]],
|
|
|
|
uint sgitg[[simdgroup_index_in_threadgroup]],
|
|
|
|
uint tiisg[[thread_index_in_simdgroup]],
|
|
|
|
uint ntg[[threads_per_threadgroup]]) {
|
|
|
|
const int64_t ne = ne00*ne01*ne02;
|
|
|
|
const int64_t gs = ne00*ne01*((ne02 + n_groups - 1) / n_groups);
|
|
|
|
|
|
|
|
int start = tgpig * gs;
|
|
|
|
int end = start + gs;
|
|
|
|
|
|
|
|
start += tpitg;
|
|
|
|
|
|
|
|
if (end >= ne) {
|
|
|
|
end = ne;
|
|
|
|
}
|
|
|
|
|
|
|
|
float tmp = 0.0f; // partial sum for thread in warp
|
|
|
|
|
|
|
|
for (int j = start; j < end; j += ntg) {
|
|
|
|
tmp += src0[j];
|
|
|
|
}
|
|
|
|
|
|
|
|
threadgroup_barrier(mem_flags::mem_threadgroup);
|
|
|
|
tmp = simd_sum(tmp);
|
|
|
|
if (ntg > N_SIMDWIDTH) {
|
|
|
|
if (sgitg == 0) {
|
|
|
|
buf[tiisg] = 0.0f;
|
|
|
|
}
|
|
|
|
|
|
|
|
threadgroup_barrier(mem_flags::mem_threadgroup);
|
|
|
|
|
|
|
|
if (tiisg == 0) {
|
|
|
|
buf[sgitg] = tmp;
|
|
|
|
}
|
|
|
|
|
|
|
|
threadgroup_barrier(mem_flags::mem_threadgroup);
|
|
|
|
|
|
|
|
tmp = buf[tiisg];
|
|
|
|
tmp = simd_sum(tmp);
|
|
|
|
}
|
|
|
|
|
|
|
|
const float mean = tmp / gs;
|
|
|
|
tmp = 0.0f;
|
|
|
|
|
|
|
|
for (int j = start; j < end; j += ntg) {
|
|
|
|
float xi = src0[j] - mean;
|
|
|
|
dst[j] = xi;
|
|
|
|
tmp += xi * xi;
|
|
|
|
}
|
|
|
|
|
|
|
|
tmp = simd_sum(tmp);
|
|
|
|
if (ntg > N_SIMDWIDTH) {
|
|
|
|
if (sgitg == 0) {
|
|
|
|
buf[tiisg] = 0.0f;
|
|
|
|
}
|
|
|
|
|
|
|
|
threadgroup_barrier(mem_flags::mem_threadgroup);
|
|
|
|
|
|
|
|
if (tiisg == 0) {
|
|
|
|
buf[sgitg] = tmp;
|
|
|
|
}
|
|
|
|
|
|
|
|
threadgroup_barrier(mem_flags::mem_threadgroup);
|
|
|
|
|
|
|
|
tmp = buf[tiisg];
|
|
|
|
tmp = simd_sum(tmp);
|
|
|
|
}
|
|
|
|
|
|
|
|
const float variance = tmp / gs;
|
|
|
|
const float scale = 1.0f/sqrt(variance + eps);
|
|
|
|
for (int j = start; j < end; j += ntg) {
|
|
|
|
dst[j] *= scale;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2023-07-25 12:48:29 +02:00
|
|
|
// function for calculate inner product between half a q4_0 block and 16 floats (yl), sumy is SUM(yl[i])
|
|
|
|
// il indicates where the q4 quants begin (0 or QK4_0/4)
|
|
|
|
// we assume that the yl's have been multiplied with the appropriate scale factor
|
|
|
|
// that corresponds to the missing bit shifts (1, 1/16, 1/256, 1/4096)
|
|
|
|
inline float block_q_n_dot_y(device const block_q4_0 * qb_curr, float sumy, thread float * yl, int il) {
|
2023-07-20 12:32:22 +02:00
|
|
|
float d = qb_curr->d;
|
2023-10-18 14:21:48 +02:00
|
|
|
|
2023-07-25 12:48:29 +02:00
|
|
|
float2 acc = 0.f;
|
2023-10-18 14:21:48 +02:00
|
|
|
|
2023-07-25 12:48:29 +02:00
|
|
|
device const uint16_t * qs = ((device const uint16_t *)qb_curr + 1 + il/2);
|
2023-10-18 14:21:48 +02:00
|
|
|
|
2023-07-25 12:48:29 +02:00
|
|
|
for (int i = 0; i < 8; i+=2) {
|
|
|
|
acc[0] += yl[i + 0] * (qs[i / 2] & 0x000F)
|
|
|
|
+ yl[i + 1] * (qs[i / 2] & 0x0F00);
|
|
|
|
acc[1] += yl[i + 8] * (qs[i / 2] & 0x00F0)
|
|
|
|
+ yl[i + 9] * (qs[i / 2] & 0xF000);
|
2023-07-20 12:32:22 +02:00
|
|
|
}
|
2023-07-25 12:48:29 +02:00
|
|
|
return d * (sumy * -8.f + acc[0] + acc[1]);
|
2023-07-20 12:32:22 +02:00
|
|
|
}
|
|
|
|
|
2023-07-25 12:48:29 +02:00
|
|
|
// function for calculate inner product between half a q4_1 block and 16 floats (yl), sumy is SUM(yl[i])
|
|
|
|
// il indicates where the q4 quants begin (0 or QK4_0/4)
|
|
|
|
// we assume that the yl's have been multiplied with the appropriate scale factor
|
|
|
|
// that corresponds to the missing bit shifts (1, 1/16, 1/256, 1/4096)
|
|
|
|
inline float block_q_n_dot_y(device const block_q4_1 * qb_curr, float sumy, thread float * yl, int il) {
|
2023-07-20 12:32:22 +02:00
|
|
|
float d = qb_curr->d;
|
|
|
|
float m = qb_curr->m;
|
2023-10-18 14:21:48 +02:00
|
|
|
|
2023-07-25 12:48:29 +02:00
|
|
|
float2 acc = 0.f;
|
2023-10-18 14:21:48 +02:00
|
|
|
|
|
|
|
device const uint16_t * qs = ((device const uint16_t *)qb_curr + 2 + il/2);
|
|
|
|
|
2023-07-25 12:48:29 +02:00
|
|
|
for (int i = 0; i < 8; i+=2) {
|
|
|
|
acc[0] += yl[i + 0] * (qs[i / 2] & 0x000F)
|
|
|
|
+ yl[i + 1] * (qs[i / 2] & 0x0F00);
|
|
|
|
acc[1] += yl[i + 8] * (qs[i / 2] & 0x00F0)
|
|
|
|
+ yl[i + 9] * (qs[i / 2] & 0xF000);
|
2023-07-20 12:32:22 +02:00
|
|
|
}
|
2023-07-25 12:48:29 +02:00
|
|
|
return d * (acc[0] + acc[1]) + sumy * m;
|
2023-06-04 22:34:30 +02:00
|
|
|
}
|
|
|
|
|
2023-10-18 14:21:48 +02:00
|
|
|
// function for calculate inner product between half a q5_0 block and 16 floats (yl), sumy is SUM(yl[i])
|
|
|
|
// il indicates where the q5 quants begin (0 or QK5_0/4)
|
|
|
|
// we assume that the yl's have been multiplied with the appropriate scale factor
|
|
|
|
// that corresponds to the missing bit shifts (1, 1/16, 1/256, 1/4096)
|
|
|
|
inline float block_q_n_dot_y(device const block_q5_0 * qb_curr, float sumy, thread float * yl, int il) {
|
|
|
|
float d = qb_curr->d;
|
|
|
|
|
|
|
|
float2 acc = 0.f;
|
|
|
|
|
|
|
|
device const uint16_t * qs = ((device const uint16_t *)qb_curr + 3 + il/2);
|
|
|
|
const uint32_t qh = *((device const uint32_t *)qb_curr->qh);
|
|
|
|
|
|
|
|
for (int i = 0; i < 8; i+=2) {
|
|
|
|
acc[0] += yl[i + 0] * ((qs[i / 2] & 0x000F) | ((qh >> (i+0+il ) << 4 ) & 0x00010))
|
|
|
|
+ yl[i + 1] * ((qs[i / 2] & 0x0F00) | ((qh >> (i+1+il ) << 12) & 0x01000));
|
|
|
|
acc[1] += yl[i + 8] * ((qs[i / 2] & 0x00F0) | ((qh >> (i+0+il+QK5_0/2) << 8 ) & 0x00100))
|
|
|
|
+ yl[i + 9] * ((qs[i / 2] & 0xF000) | ((qh >> (i+1+il+QK5_0/2) << 16) & 0x10000));
|
|
|
|
}
|
|
|
|
return d * (sumy * -16.f + acc[0] + acc[1]);
|
|
|
|
}
|
|
|
|
|
|
|
|
// function for calculate inner product between half a q5_1 block and 16 floats (yl), sumy is SUM(yl[i])
|
|
|
|
// il indicates where the q5 quants begin (0 or QK5_1/4)
|
|
|
|
// we assume that the yl's have been multiplied with the appropriate scale factor
|
|
|
|
// that corresponds to the missing bit shifts (1, 1/16, 1/256, 1/4096)
|
|
|
|
inline float block_q_n_dot_y(device const block_q5_1 * qb_curr, float sumy, thread float * yl, int il) {
|
|
|
|
float d = qb_curr->d;
|
|
|
|
float m = qb_curr->m;
|
|
|
|
|
|
|
|
float2 acc = 0.f;
|
|
|
|
|
|
|
|
device const uint16_t * qs = ((device const uint16_t *)qb_curr + 4 + il/2);
|
|
|
|
const uint32_t qh = *((device const uint32_t *)qb_curr->qh);
|
|
|
|
|
|
|
|
for (int i = 0; i < 8; i+=2) {
|
|
|
|
acc[0] += yl[i + 0] * ((qs[i / 2] & 0x000F) | ((qh >> (i+0+il ) << 4 ) & 0x00010))
|
|
|
|
+ yl[i + 1] * ((qs[i / 2] & 0x0F00) | ((qh >> (i+1+il ) << 12) & 0x01000));
|
|
|
|
acc[1] += yl[i + 8] * ((qs[i / 2] & 0x00F0) | ((qh >> (i+0+il+QK5_0/2) << 8 ) & 0x00100))
|
|
|
|
+ yl[i + 9] * ((qs[i / 2] & 0xF000) | ((qh >> (i+1+il+QK5_0/2) << 16) & 0x10000));
|
|
|
|
}
|
|
|
|
return d * (acc[0] + acc[1]) + sumy * m;
|
|
|
|
}
|
|
|
|
|
2023-07-12 22:10:55 +02:00
|
|
|
// putting them in the kernel cause a significant performance penalty
|
2023-10-08 09:01:53 +02:00
|
|
|
#define N_DST 4 // each SIMD group works on 4 rows
|
|
|
|
#define N_SIMDGROUP 2 // number of SIMD groups in a thread group
|
2023-07-25 12:48:29 +02:00
|
|
|
//Note: This is a template, but strictly speaking it only applies to
|
|
|
|
// quantizations where the block size is 32. It also does not
|
2024-01-02 20:07:47 +01:00
|
|
|
// guard against the number of rows not being divisible by
|
2023-07-25 12:48:29 +02:00
|
|
|
// N_DST, so this is another explicit assumption of the implementation.
|
|
|
|
template<typename block_q_type, int nr, int nsg, int nw>
|
2023-12-13 13:04:25 +01:00
|
|
|
void mul_vec_q_n_f32_impl(
|
2023-12-07 21:26:54 +01:00
|
|
|
device const void * src0,
|
|
|
|
device const float * src1,
|
|
|
|
device float * dst,
|
|
|
|
int64_t ne00,
|
|
|
|
int64_t ne01,
|
|
|
|
int64_t ne02,
|
|
|
|
int64_t ne10,
|
|
|
|
int64_t ne12,
|
|
|
|
int64_t ne0,
|
|
|
|
int64_t ne1,
|
|
|
|
uint r2,
|
|
|
|
uint r3,
|
|
|
|
uint3 tgpig, uint tiisg, uint sgitg) {
|
2023-06-04 22:34:30 +02:00
|
|
|
const int nb = ne00/QK4_0;
|
2023-10-08 09:01:53 +02:00
|
|
|
|
2023-07-12 22:10:55 +02:00
|
|
|
const int r0 = tgpig.x;
|
|
|
|
const int r1 = tgpig.y;
|
2023-08-16 22:07:04 +02:00
|
|
|
const int im = tgpig.z;
|
2023-10-08 09:01:53 +02:00
|
|
|
|
2023-07-25 12:48:29 +02:00
|
|
|
const int first_row = (r0 * nsg + sgitg) * nr;
|
2023-10-08 09:01:53 +02:00
|
|
|
|
2023-12-07 21:26:54 +01:00
|
|
|
const uint i12 = im%ne12;
|
|
|
|
const uint i13 = im/ne12;
|
|
|
|
|
|
|
|
const uint offset0 = first_row * nb + (i12/r2)*(nb*ne01) + (i13/r3)*(nb*ne01*ne02);
|
2023-10-08 09:01:53 +02:00
|
|
|
|
2023-08-16 22:07:04 +02:00
|
|
|
device const block_q_type * x = (device const block_q_type *) src0 + offset0;
|
2023-08-24 15:19:57 +02:00
|
|
|
device const float * y = (device const float *) src1 + r1*ne10 + im*ne00*ne1;
|
2023-07-12 22:10:55 +02:00
|
|
|
|
2023-10-08 09:01:53 +02:00
|
|
|
float yl[16]; // src1 vector cache
|
|
|
|
float sumf[nr] = {0.f};
|
|
|
|
|
|
|
|
const int ix = (tiisg/2);
|
|
|
|
const int il = (tiisg%2)*8;
|
2023-06-04 22:34:30 +02:00
|
|
|
|
2023-07-25 12:48:29 +02:00
|
|
|
device const float * yb = y + ix * QK4_0 + il;
|
2023-06-04 22:34:30 +02:00
|
|
|
|
2023-07-25 12:48:29 +02:00
|
|
|
// each thread in a SIMD group deals with half a block.
|
|
|
|
for (int ib = ix; ib < nb; ib += nw/2) {
|
2023-07-14 11:46:21 +02:00
|
|
|
float sumy = 0;
|
2023-07-25 12:48:29 +02:00
|
|
|
for (int i = 0; i < 8; i += 2) {
|
|
|
|
sumy += yb[i] + yb[i+1];
|
|
|
|
yl[i+0] = yb[i+ 0];
|
|
|
|
yl[i+1] = yb[i+ 1]/256.f;
|
2023-10-08 09:01:53 +02:00
|
|
|
|
2023-07-25 12:48:29 +02:00
|
|
|
sumy += yb[i+16] + yb[i+17];
|
|
|
|
yl[i+8] = yb[i+16]/16.f;
|
|
|
|
yl[i+9] = yb[i+17]/4096.f;
|
2023-07-12 22:10:55 +02:00
|
|
|
}
|
2023-07-14 11:46:21 +02:00
|
|
|
|
2023-07-25 12:48:29 +02:00
|
|
|
for (int row = 0; row < nr; row++) {
|
|
|
|
sumf[row] += block_q_n_dot_y(x+ib+row*nb, sumy, yl, il);
|
2023-07-20 12:32:22 +02:00
|
|
|
}
|
2023-07-25 12:48:29 +02:00
|
|
|
|
|
|
|
yb += QK4_0 * 16;
|
2023-07-20 12:32:22 +02:00
|
|
|
}
|
2023-07-14 11:46:21 +02:00
|
|
|
|
2023-07-25 12:48:29 +02:00
|
|
|
for (int row = 0; row < nr; ++row) {
|
|
|
|
const float tot = simd_sum(sumf[row]);
|
|
|
|
if (tiisg == 0 && first_row + row < ne01) {
|
2023-10-08 09:01:53 +02:00
|
|
|
dst[im*ne0*ne1 + r1*ne0 + first_row + row] = tot;
|
2023-07-12 22:10:55 +02:00
|
|
|
}
|
2023-06-04 22:34:30 +02:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2023-10-08 09:01:53 +02:00
|
|
|
kernel void kernel_mul_mv_q4_0_f32(
|
2023-06-10 10:28:11 +02:00
|
|
|
device const void * src0,
|
|
|
|
device const float * src1,
|
|
|
|
device float * dst,
|
|
|
|
constant int64_t & ne00,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant uint64_t & nb00,
|
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant uint64_t & nb02,
|
|
|
|
constant int64_t & ne10,
|
|
|
|
constant int64_t & ne11,
|
|
|
|
constant int64_t & ne12,
|
|
|
|
constant uint64_t & nb10,
|
|
|
|
constant uint64_t & nb11,
|
|
|
|
constant uint64_t & nb12,
|
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
|
|
|
constant uint & r2,
|
|
|
|
constant uint & r3,
|
2023-08-16 22:07:04 +02:00
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
2023-10-08 09:01:53 +02:00
|
|
|
uint tiisg[[thread_index_in_simdgroup]],
|
|
|
|
uint sgitg[[simdgroup_index_in_threadgroup]]) {
|
2023-12-13 13:04:25 +01:00
|
|
|
mul_vec_q_n_f32_impl<block_q4_0, N_DST, N_SIMDGROUP, N_SIMDWIDTH>(src0,src1,dst,ne00,ne01,ne02,ne10,ne12,ne0,ne1,r2,r3,tgpig,tiisg,sgitg);
|
2023-07-20 12:32:22 +02:00
|
|
|
}
|
2023-06-10 10:28:11 +02:00
|
|
|
|
2023-10-08 09:01:53 +02:00
|
|
|
kernel void kernel_mul_mv_q4_1_f32(
|
2023-07-20 12:32:22 +02:00
|
|
|
device const void * src0,
|
|
|
|
device const float * src1,
|
|
|
|
device float * dst,
|
|
|
|
constant int64_t & ne00,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant uint64_t & nb00,
|
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant uint64_t & nb02,
|
|
|
|
constant int64_t & ne10,
|
|
|
|
constant int64_t & ne11,
|
|
|
|
constant int64_t & ne12,
|
|
|
|
constant uint64_t & nb10,
|
|
|
|
constant uint64_t & nb11,
|
|
|
|
constant uint64_t & nb12,
|
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
|
|
|
constant uint & r2,
|
|
|
|
constant uint & r3,
|
2023-08-16 22:07:04 +02:00
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
2023-07-20 12:32:22 +02:00
|
|
|
uint tiisg[[thread_index_in_simdgroup]],
|
|
|
|
uint sgitg[[simdgroup_index_in_threadgroup]]) {
|
2023-12-13 13:04:25 +01:00
|
|
|
mul_vec_q_n_f32_impl<block_q4_1, N_DST, N_SIMDGROUP, N_SIMDWIDTH>(src0,src1,dst,ne00,ne01,ne02,ne10,ne12,ne0,ne1,r2,r3,tgpig,tiisg,sgitg);
|
2023-06-10 10:28:11 +02:00
|
|
|
}
|
|
|
|
|
2023-10-18 14:21:48 +02:00
|
|
|
kernel void kernel_mul_mv_q5_0_f32(
|
|
|
|
device const void * src0,
|
|
|
|
device const float * src1,
|
|
|
|
device float * dst,
|
|
|
|
constant int64_t & ne00,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant uint64_t & nb00,
|
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant uint64_t & nb02,
|
|
|
|
constant int64_t & ne10,
|
|
|
|
constant int64_t & ne11,
|
|
|
|
constant int64_t & ne12,
|
|
|
|
constant uint64_t & nb10,
|
|
|
|
constant uint64_t & nb11,
|
|
|
|
constant uint64_t & nb12,
|
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
|
|
|
constant uint & r2,
|
|
|
|
constant uint & r3,
|
2023-10-18 14:21:48 +02:00
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint tiisg[[thread_index_in_simdgroup]],
|
|
|
|
uint sgitg[[simdgroup_index_in_threadgroup]]) {
|
2023-12-13 13:04:25 +01:00
|
|
|
mul_vec_q_n_f32_impl<block_q5_0, N_DST, N_SIMDGROUP, N_SIMDWIDTH>(src0,src1,dst,ne00,ne01,ne02,ne10,ne12,ne0,ne1,r2,r3,tgpig,tiisg,sgitg);
|
2023-10-18 14:21:48 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
kernel void kernel_mul_mv_q5_1_f32(
|
|
|
|
device const void * src0,
|
|
|
|
device const float * src1,
|
|
|
|
device float * dst,
|
|
|
|
constant int64_t & ne00,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant uint64_t & nb00,
|
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant uint64_t & nb02,
|
|
|
|
constant int64_t & ne10,
|
|
|
|
constant int64_t & ne11,
|
|
|
|
constant int64_t & ne12,
|
|
|
|
constant uint64_t & nb10,
|
|
|
|
constant uint64_t & nb11,
|
|
|
|
constant uint64_t & nb12,
|
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
|
|
|
constant uint & r2,
|
|
|
|
constant uint & r3,
|
2023-10-18 14:21:48 +02:00
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint tiisg[[thread_index_in_simdgroup]],
|
|
|
|
uint sgitg[[simdgroup_index_in_threadgroup]]) {
|
2023-12-13 13:04:25 +01:00
|
|
|
mul_vec_q_n_f32_impl<block_q5_1, N_DST, N_SIMDGROUP, N_SIMDWIDTH>(src0,src1,dst,ne00,ne01,ne02,ne10,ne12,ne0,ne1,r2,r3,tgpig,tiisg,sgitg);
|
2023-10-18 14:21:48 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
|
2023-09-03 10:06:22 +02:00
|
|
|
#define NB_Q8_0 8
|
|
|
|
|
2023-12-13 13:04:25 +01:00
|
|
|
void kernel_mul_mv_q8_0_f32_impl(
|
2023-08-24 15:19:57 +02:00
|
|
|
device const void * src0,
|
|
|
|
device const float * src1,
|
|
|
|
device float * dst,
|
|
|
|
constant int64_t & ne00,
|
2023-12-13 13:04:25 +01:00
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant int64_t & ne10,
|
|
|
|
constant int64_t & ne12,
|
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
|
|
|
constant uint & r2,
|
|
|
|
constant uint & r3,
|
2023-08-24 15:19:57 +02:00
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
2023-12-13 13:04:25 +01:00
|
|
|
uint tiisg[[thread_index_in_simdgroup]],
|
|
|
|
uint sgitg[[simdgroup_index_in_threadgroup]]) {
|
2023-08-24 15:19:57 +02:00
|
|
|
const int nr = N_DST;
|
|
|
|
const int nsg = N_SIMDGROUP;
|
|
|
|
const int nw = N_SIMDWIDTH;
|
|
|
|
|
|
|
|
const int nb = ne00/QK8_0;
|
|
|
|
const int r0 = tgpig.x;
|
|
|
|
const int r1 = tgpig.y;
|
|
|
|
const int im = tgpig.z;
|
2023-12-07 21:26:54 +01:00
|
|
|
|
2023-08-24 15:19:57 +02:00
|
|
|
const int first_row = (r0 * nsg + sgitg) * nr;
|
2023-12-07 21:26:54 +01:00
|
|
|
|
|
|
|
const uint i12 = im%ne12;
|
|
|
|
const uint i13 = im/ne12;
|
|
|
|
|
|
|
|
const uint offset0 = first_row * nb + (i12/r2)*(nb*ne01) + (i13/r3)*(nb*ne01*ne02);
|
|
|
|
|
2023-08-24 15:19:57 +02:00
|
|
|
device const block_q8_0 * x = (device const block_q8_0 *) src0 + offset0;
|
|
|
|
device const float * y = (device const float *) src1 + r1*ne10 + im*ne00*ne1;
|
|
|
|
|
2023-09-03 10:06:22 +02:00
|
|
|
float yl[NB_Q8_0];
|
2023-08-24 15:19:57 +02:00
|
|
|
float sumf[nr]={0.f};
|
|
|
|
|
2023-09-03 10:06:22 +02:00
|
|
|
const int ix = tiisg/4;
|
|
|
|
const int il = tiisg%4;
|
2023-08-24 15:19:57 +02:00
|
|
|
|
2023-09-03 10:06:22 +02:00
|
|
|
device const float * yb = y + ix * QK8_0 + NB_Q8_0*il;
|
2023-08-24 15:19:57 +02:00
|
|
|
|
2023-09-03 10:06:22 +02:00
|
|
|
// each thread in a SIMD group deals with NB_Q8_0 quants at a time
|
|
|
|
for (int ib = ix; ib < nb; ib += nw/4) {
|
|
|
|
for (int i = 0; i < NB_Q8_0; ++i) {
|
2023-08-24 15:19:57 +02:00
|
|
|
yl[i] = yb[i];
|
|
|
|
}
|
|
|
|
|
|
|
|
for (int row = 0; row < nr; row++) {
|
2023-09-03 10:06:22 +02:00
|
|
|
device const int8_t * qs = x[ib+row*nb].qs + NB_Q8_0*il;
|
2023-08-24 15:19:57 +02:00
|
|
|
float sumq = 0.f;
|
2023-09-03 10:06:22 +02:00
|
|
|
for (int iq = 0; iq < NB_Q8_0; ++iq) {
|
2023-08-24 15:19:57 +02:00
|
|
|
sumq += qs[iq] * yl[iq];
|
|
|
|
}
|
|
|
|
sumf[row] += sumq*x[ib+row*nb].d;
|
|
|
|
}
|
|
|
|
|
2023-09-03 10:06:22 +02:00
|
|
|
yb += NB_Q8_0 * nw;
|
2023-08-24 15:19:57 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
for (int row = 0; row < nr; ++row) {
|
|
|
|
const float tot = simd_sum(sumf[row]);
|
|
|
|
if (tiisg == 0 && first_row + row < ne01) {
|
|
|
|
dst[r1*ne0 + im*ne0*ne1 + first_row + row] = tot;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2023-12-13 13:04:25 +01:00
|
|
|
[[host_name("kernel_mul_mv_q8_0_f32")]]
|
|
|
|
kernel void kernel_mul_mv_q8_0_f32(
|
|
|
|
device const void * src0,
|
|
|
|
device const float * src1,
|
|
|
|
device float * dst,
|
|
|
|
constant int64_t & ne00,
|
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant uint64_t & nb00,
|
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant uint64_t & nb02,
|
2023-12-13 13:04:25 +01:00
|
|
|
constant int64_t & ne10,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant int64_t & ne11,
|
2023-12-13 13:04:25 +01:00
|
|
|
constant int64_t & ne12,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant uint64_t & nb10,
|
|
|
|
constant uint64_t & nb11,
|
|
|
|
constant uint64_t & nb12,
|
2023-12-13 13:04:25 +01:00
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant uint & r2,
|
|
|
|
constant uint & r3,
|
2023-12-13 13:04:25 +01:00
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint tiisg[[thread_index_in_simdgroup]],
|
|
|
|
uint sgitg[[simdgroup_index_in_threadgroup]]) {
|
|
|
|
kernel_mul_mv_q8_0_f32_impl(src0,src1,dst,ne00,ne01,ne02,ne10,ne12,ne0,ne1,r2,r3,tgpig,tiisg,sgitg);
|
|
|
|
}
|
|
|
|
|
2023-09-15 18:06:03 +02:00
|
|
|
#define N_F32_F32 4
|
|
|
|
|
2023-12-13 13:04:25 +01:00
|
|
|
void kernel_mul_mv_f32_f32_impl(
|
2023-09-15 18:06:03 +02:00
|
|
|
device const char * src0,
|
|
|
|
device const char * src1,
|
|
|
|
device float * dst,
|
|
|
|
constant int64_t & ne00,
|
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant uint64_t & nb00,
|
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant uint64_t & nb02,
|
|
|
|
constant int64_t & ne10,
|
|
|
|
constant int64_t & ne11,
|
|
|
|
constant int64_t & ne12,
|
|
|
|
constant uint64_t & nb10,
|
|
|
|
constant uint64_t & nb11,
|
|
|
|
constant uint64_t & nb12,
|
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
2023-12-13 13:04:25 +01:00
|
|
|
constant uint & r2,
|
|
|
|
constant uint & r3,
|
2023-09-15 18:06:03 +02:00
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
2023-11-13 15:55:52 +01:00
|
|
|
uint tiisg[[thread_index_in_simdgroup]]) {
|
2023-09-15 18:06:03 +02:00
|
|
|
|
|
|
|
const int64_t r0 = tgpig.x;
|
|
|
|
const int64_t rb = tgpig.y*N_F32_F32;
|
|
|
|
const int64_t im = tgpig.z;
|
|
|
|
|
2023-12-07 21:26:54 +01:00
|
|
|
const uint i12 = im%ne12;
|
|
|
|
const uint i13 = im/ne12;
|
|
|
|
|
|
|
|
const uint offset0 = r0*nb01 + (i12/r2)*nb02 + (i13/r3)*nb02*ne02;
|
|
|
|
|
|
|
|
device const float * x = (device const float *) (src0 + offset0);
|
2023-09-15 18:06:03 +02:00
|
|
|
|
|
|
|
if (ne00 < 128) {
|
|
|
|
for (int row = 0; row < N_F32_F32; ++row) {
|
|
|
|
int r1 = rb + row;
|
|
|
|
if (r1 >= ne11) {
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
|
|
|
device const float * y = (device const float *) (src1 + r1*nb11 + im*nb12);
|
|
|
|
|
|
|
|
float sumf = 0;
|
|
|
|
for (int i = tiisg; i < ne00; i += 32) {
|
|
|
|
sumf += (float) x[i] * (float) y[i];
|
|
|
|
}
|
|
|
|
|
|
|
|
float all_sum = simd_sum(sumf);
|
|
|
|
if (tiisg == 0) {
|
|
|
|
dst[im*ne1*ne0 + r1*ne0 + r0] = all_sum;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
device const float4 * x4 = (device const float4 *)x;
|
|
|
|
for (int row = 0; row < N_F32_F32; ++row) {
|
|
|
|
int r1 = rb + row;
|
|
|
|
if (r1 >= ne11) {
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
|
|
|
device const float * y = (device const float *) (src1 + r1*nb11 + im*nb12);
|
|
|
|
device const float4 * y4 = (device const float4 *) y;
|
|
|
|
|
|
|
|
float sumf = 0;
|
|
|
|
for (int i = tiisg; i < ne00/4; i += 32) {
|
|
|
|
for (int k = 0; k < 4; ++k) sumf += (float) x4[i][k] * y4[i][k];
|
|
|
|
}
|
|
|
|
|
|
|
|
float all_sum = simd_sum(sumf);
|
|
|
|
if (tiisg == 0) {
|
|
|
|
for (int i = 4*(ne00/4); i < ne00; ++i) all_sum += (float) x[i] * y[i];
|
|
|
|
dst[im*ne1*ne0 + r1*ne0 + r0] = all_sum;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2023-12-13 13:04:25 +01:00
|
|
|
[[host_name("kernel_mul_mv_f32_f32")]]
|
|
|
|
kernel void kernel_mul_mv_f32_f32(
|
|
|
|
device const char * src0,
|
|
|
|
device const char * src1,
|
|
|
|
device float * dst,
|
|
|
|
constant int64_t & ne00,
|
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant uint64_t & nb00,
|
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant uint64_t & nb02,
|
|
|
|
constant int64_t & ne10,
|
|
|
|
constant int64_t & ne11,
|
|
|
|
constant int64_t & ne12,
|
|
|
|
constant uint64_t & nb10,
|
|
|
|
constant uint64_t & nb11,
|
|
|
|
constant uint64_t & nb12,
|
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant uint & r2,
|
|
|
|
constant uint & r3,
|
2023-12-13 13:04:25 +01:00
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint tiisg[[thread_index_in_simdgroup]]) {
|
|
|
|
kernel_mul_mv_f32_f32_impl(src0, src1, dst, ne00, ne01, ne02, nb00, nb01, nb02, ne10, ne11, ne12, nb10, nb11, nb12, ne0, ne1, r2, r3, tgpig, tiisg);
|
|
|
|
}
|
|
|
|
|
2023-11-13 15:55:52 +01:00
|
|
|
#define N_F16_F16 4
|
|
|
|
|
|
|
|
kernel void kernel_mul_mv_f16_f16(
|
|
|
|
device const char * src0,
|
|
|
|
device const char * src1,
|
|
|
|
device float * dst,
|
|
|
|
constant int64_t & ne00,
|
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant uint64_t & nb00,
|
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant uint64_t & nb02,
|
|
|
|
constant int64_t & ne10,
|
|
|
|
constant int64_t & ne11,
|
|
|
|
constant int64_t & ne12,
|
|
|
|
constant uint64_t & nb10,
|
|
|
|
constant uint64_t & nb11,
|
|
|
|
constant uint64_t & nb12,
|
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant uint & r2,
|
|
|
|
constant uint & r3,
|
2023-11-13 15:55:52 +01:00
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint tiisg[[thread_index_in_simdgroup]]) {
|
|
|
|
|
|
|
|
const int64_t r0 = tgpig.x;
|
|
|
|
const int64_t rb = tgpig.y*N_F16_F16;
|
|
|
|
const int64_t im = tgpig.z;
|
|
|
|
|
2023-12-07 21:26:54 +01:00
|
|
|
const uint i12 = im%ne12;
|
|
|
|
const uint i13 = im/ne12;
|
|
|
|
|
|
|
|
const uint offset0 = r0*nb01 + (i12/r2)*nb02 + (i13/r3)*nb02*ne02;
|
|
|
|
|
|
|
|
device const half * x = (device const half *) (src0 + offset0);
|
2023-11-13 15:55:52 +01:00
|
|
|
|
|
|
|
if (ne00 < 128) {
|
|
|
|
for (int row = 0; row < N_F16_F16; ++row) {
|
|
|
|
int r1 = rb + row;
|
|
|
|
if (r1 >= ne11) {
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
|
|
|
device const half * y = (device const half *) (src1 + r1*nb11 + im*nb12);
|
|
|
|
|
|
|
|
float sumf = 0;
|
|
|
|
for (int i = tiisg; i < ne00; i += 32) {
|
|
|
|
sumf += (half) x[i] * (half) y[i];
|
|
|
|
}
|
|
|
|
|
|
|
|
float all_sum = simd_sum(sumf);
|
|
|
|
if (tiisg == 0) {
|
|
|
|
dst[im*ne1*ne0 + r1*ne0 + r0] = all_sum;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
device const half4 * x4 = (device const half4 *)x;
|
|
|
|
for (int row = 0; row < N_F16_F16; ++row) {
|
|
|
|
int r1 = rb + row;
|
|
|
|
if (r1 >= ne11) {
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
|
|
|
device const half * y = (device const half *) (src1 + r1*nb11 + im*nb12);
|
|
|
|
device const half4 * y4 = (device const half4 *) y;
|
|
|
|
|
|
|
|
float sumf = 0;
|
|
|
|
for (int i = tiisg; i < ne00/4; i += 32) {
|
|
|
|
for (int k = 0; k < 4; ++k) sumf += (half) x4[i][k] * y4[i][k];
|
|
|
|
}
|
|
|
|
|
|
|
|
float all_sum = simd_sum(sumf);
|
|
|
|
if (tiisg == 0) {
|
|
|
|
for (int i = 4*(ne00/4); i < ne00; ++i) all_sum += (half) x[i] * y[i];
|
|
|
|
dst[im*ne1*ne0 + r1*ne0 + r0] = all_sum;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2023-12-13 13:04:25 +01:00
|
|
|
void kernel_mul_mv_f16_f32_1row_impl(
|
2023-06-04 22:34:30 +02:00
|
|
|
device const char * src0,
|
|
|
|
device const char * src1,
|
|
|
|
device float * dst,
|
|
|
|
constant int64_t & ne00,
|
|
|
|
constant int64_t & ne01,
|
2023-08-01 09:43:12 +02:00
|
|
|
constant int64_t & ne02,
|
2023-06-04 22:34:30 +02:00
|
|
|
constant uint64_t & nb00,
|
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant uint64_t & nb02,
|
|
|
|
constant int64_t & ne10,
|
|
|
|
constant int64_t & ne11,
|
2023-08-01 09:43:12 +02:00
|
|
|
constant int64_t & ne12,
|
2023-06-04 22:34:30 +02:00
|
|
|
constant uint64_t & nb10,
|
|
|
|
constant uint64_t & nb11,
|
|
|
|
constant uint64_t & nb12,
|
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
2023-12-13 13:04:25 +01:00
|
|
|
constant uint & r2,
|
|
|
|
constant uint & r3,
|
2023-06-04 22:34:30 +02:00
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
2023-10-08 09:01:53 +02:00
|
|
|
uint tiisg[[thread_index_in_simdgroup]]) {
|
2023-06-09 09:39:59 +02:00
|
|
|
|
2023-06-04 22:34:30 +02:00
|
|
|
const int64_t r0 = tgpig.x;
|
|
|
|
const int64_t r1 = tgpig.y;
|
|
|
|
const int64_t im = tgpig.z;
|
|
|
|
|
2023-12-07 21:26:54 +01:00
|
|
|
const uint i12 = im%ne12;
|
|
|
|
const uint i13 = im/ne12;
|
|
|
|
|
|
|
|
const uint offset0 = r0*nb01 + (i12/r2)*nb02 + (i13/r3)*nb02*ne02;
|
|
|
|
|
|
|
|
device const half * x = (device const half *) (src0 + offset0);
|
2023-06-04 22:34:30 +02:00
|
|
|
device const float * y = (device const float *) (src1 + r1*nb11 + im*nb12);
|
|
|
|
|
2023-09-03 10:06:22 +02:00
|
|
|
float sumf = 0;
|
2023-09-03 12:23:33 +02:00
|
|
|
if (ne00 < 128) {
|
|
|
|
for (int i = tiisg; i < ne00; i += 32) {
|
|
|
|
sumf += (float) x[i] * (float) y[i];
|
|
|
|
}
|
|
|
|
float all_sum = simd_sum(sumf);
|
|
|
|
if (tiisg == 0) {
|
|
|
|
dst[im*ne1*ne0 + r1*ne0 + r0] = all_sum;
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
device const half4 * x4 = (device const half4 *) x;
|
|
|
|
device const float4 * y4 = (device const float4 *) y;
|
|
|
|
for (int i = tiisg; i < ne00/4; i += 32) {
|
|
|
|
for (int k = 0; k < 4; ++k) sumf += (float)x4[i][k] * y4[i][k];
|
|
|
|
}
|
|
|
|
float all_sum = simd_sum(sumf);
|
|
|
|
if (tiisg == 0) {
|
|
|
|
for (int i = 4*(ne00/4); i < ne00; ++i) all_sum += (float) x[i] * y[i];
|
|
|
|
dst[im*ne1*ne0 + r1*ne0 + r0] = all_sum;
|
|
|
|
}
|
2023-09-03 10:06:22 +02:00
|
|
|
}
|
2023-12-13 13:04:25 +01:00
|
|
|
}
|
2023-06-04 22:34:30 +02:00
|
|
|
|
2023-12-13 13:04:25 +01:00
|
|
|
[[host_name("kernel_mul_mv_f16_f32_1row")]]
|
|
|
|
kernel void kernel_mul_mv_f16_f32_1row(
|
|
|
|
device const char * src0,
|
|
|
|
device const char * src1,
|
|
|
|
device float * dst,
|
|
|
|
constant int64_t & ne00,
|
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant uint64_t & nb00,
|
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant uint64_t & nb02,
|
|
|
|
constant int64_t & ne10,
|
|
|
|
constant int64_t & ne11,
|
|
|
|
constant int64_t & ne12,
|
|
|
|
constant uint64_t & nb10,
|
|
|
|
constant uint64_t & nb11,
|
|
|
|
constant uint64_t & nb12,
|
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant uint & r2,
|
|
|
|
constant uint & r3,
|
2023-12-13 13:04:25 +01:00
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint tiisg[[thread_index_in_simdgroup]]) {
|
|
|
|
kernel_mul_mv_f16_f32_1row_impl(src0, src1, dst, ne00, ne01, ne02, nb00, nb01, nb02, ne10, ne11, ne12, nb10, nb11, nb12, ne0, ne1, r2, r3, tgpig, tiisg);
|
2023-09-03 10:06:22 +02:00
|
|
|
}
|
2023-09-01 10:15:57 +02:00
|
|
|
|
2023-09-03 10:06:22 +02:00
|
|
|
#define N_F16_F32 4
|
2023-06-04 22:34:30 +02:00
|
|
|
|
2023-12-13 13:04:25 +01:00
|
|
|
void kernel_mul_mv_f16_f32_impl(
|
2023-09-03 10:06:22 +02:00
|
|
|
device const char * src0,
|
|
|
|
device const char * src1,
|
|
|
|
device float * dst,
|
|
|
|
constant int64_t & ne00,
|
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant uint64_t & nb00,
|
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant uint64_t & nb02,
|
|
|
|
constant int64_t & ne10,
|
|
|
|
constant int64_t & ne11,
|
|
|
|
constant int64_t & ne12,
|
|
|
|
constant uint64_t & nb10,
|
|
|
|
constant uint64_t & nb11,
|
|
|
|
constant uint64_t & nb12,
|
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
2023-12-13 13:04:25 +01:00
|
|
|
constant uint & r2,
|
|
|
|
constant uint & r3,
|
2023-09-03 10:06:22 +02:00
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint tiisg[[thread_index_in_simdgroup]]) {
|
|
|
|
|
|
|
|
const int64_t r0 = tgpig.x;
|
2023-09-03 12:23:33 +02:00
|
|
|
const int64_t rb = tgpig.y*N_F16_F32;
|
2023-09-03 10:06:22 +02:00
|
|
|
const int64_t im = tgpig.z;
|
|
|
|
|
2023-12-07 21:26:54 +01:00
|
|
|
const uint i12 = im%ne12;
|
|
|
|
const uint i13 = im/ne12;
|
|
|
|
|
|
|
|
const uint offset0 = r0*nb01 + (i12/r2)*nb02 + (i13/r3)*nb02*ne02;
|
|
|
|
|
|
|
|
device const half * x = (device const half *) (src0 + offset0);
|
2023-09-03 10:06:22 +02:00
|
|
|
|
2023-09-03 12:23:33 +02:00
|
|
|
if (ne00 < 128) {
|
|
|
|
for (int row = 0; row < N_F16_F32; ++row) {
|
|
|
|
int r1 = rb + row;
|
|
|
|
if (r1 >= ne11) {
|
|
|
|
break;
|
|
|
|
}
|
2023-09-03 10:06:22 +02:00
|
|
|
|
2023-09-03 12:23:33 +02:00
|
|
|
device const float * y = (device const float *) (src1 + r1*nb11 + im*nb12);
|
2023-09-03 10:06:22 +02:00
|
|
|
|
2023-09-03 12:23:33 +02:00
|
|
|
float sumf = 0;
|
|
|
|
for (int i = tiisg; i < ne00; i += 32) {
|
|
|
|
sumf += (float) x[i] * (float) y[i];
|
|
|
|
}
|
|
|
|
|
|
|
|
float all_sum = simd_sum(sumf);
|
|
|
|
if (tiisg == 0) {
|
|
|
|
dst[im*ne1*ne0 + r1*ne0 + r0] = all_sum;
|
|
|
|
}
|
2023-09-03 11:40:56 +02:00
|
|
|
}
|
2023-09-03 12:23:33 +02:00
|
|
|
} else {
|
|
|
|
device const half4 * x4 = (device const half4 *)x;
|
|
|
|
for (int row = 0; row < N_F16_F32; ++row) {
|
|
|
|
int r1 = rb + row;
|
|
|
|
if (r1 >= ne11) {
|
|
|
|
break;
|
|
|
|
}
|
2023-09-03 10:06:22 +02:00
|
|
|
|
2023-09-03 12:23:33 +02:00
|
|
|
device const float * y = (device const float *) (src1 + r1*nb11 + im*nb12);
|
|
|
|
device const float4 * y4 = (device const float4 *) y;
|
|
|
|
|
|
|
|
float sumf = 0;
|
|
|
|
for (int i = tiisg; i < ne00/4; i += 32) {
|
|
|
|
for (int k = 0; k < 4; ++k) sumf += (float) x4[i][k] * y4[i][k];
|
|
|
|
}
|
|
|
|
|
|
|
|
float all_sum = simd_sum(sumf);
|
|
|
|
if (tiisg == 0) {
|
|
|
|
for (int i = 4*(ne00/4); i < ne00; ++i) all_sum += (float) x[i] * y[i];
|
|
|
|
dst[im*ne1*ne0 + r1*ne0 + r0] = all_sum;
|
|
|
|
}
|
2023-09-03 10:06:22 +02:00
|
|
|
}
|
2023-06-04 22:34:30 +02:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2023-12-13 13:04:25 +01:00
|
|
|
[[host_name("kernel_mul_mv_f16_f32")]]
|
|
|
|
kernel void kernel_mul_mv_f16_f32(
|
|
|
|
device const char * src0,
|
|
|
|
device const char * src1,
|
|
|
|
device float * dst,
|
|
|
|
constant int64_t & ne00,
|
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant uint64_t & nb00,
|
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant uint64_t & nb02,
|
|
|
|
constant int64_t & ne10,
|
|
|
|
constant int64_t & ne11,
|
|
|
|
constant int64_t & ne12,
|
|
|
|
constant uint64_t & nb10,
|
|
|
|
constant uint64_t & nb11,
|
|
|
|
constant uint64_t & nb12,
|
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant uint & r2,
|
|
|
|
constant uint & r3,
|
2023-12-13 13:04:25 +01:00
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint tiisg[[thread_index_in_simdgroup]]) {
|
|
|
|
kernel_mul_mv_f16_f32_impl(src0, src1, dst, ne00, ne01, ne02, nb00, nb01, nb02, ne10, ne11, ne12, nb10, nb11, nb12, ne0, ne1, r2, r3, tgpig, tiisg);
|
|
|
|
}
|
|
|
|
|
2023-09-11 09:30:11 +02:00
|
|
|
// Assumes row size (ne00) is a multiple of 4
|
2023-10-08 09:01:53 +02:00
|
|
|
kernel void kernel_mul_mv_f16_f32_l4(
|
2023-09-11 09:30:11 +02:00
|
|
|
device const char * src0,
|
|
|
|
device const char * src1,
|
|
|
|
device float * dst,
|
|
|
|
constant int64_t & ne00,
|
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant uint64_t & nb00,
|
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant uint64_t & nb02,
|
|
|
|
constant int64_t & ne10,
|
|
|
|
constant int64_t & ne11,
|
|
|
|
constant int64_t & ne12,
|
|
|
|
constant uint64_t & nb10,
|
|
|
|
constant uint64_t & nb11,
|
|
|
|
constant uint64_t & nb12,
|
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant uint & r2,
|
|
|
|
constant uint & r3,
|
2023-09-11 09:30:11 +02:00
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint tiisg[[thread_index_in_simdgroup]]) {
|
|
|
|
|
|
|
|
const int nrows = ne11;
|
|
|
|
const int64_t r0 = tgpig.x;
|
|
|
|
const int64_t im = tgpig.z;
|
|
|
|
|
2023-12-07 21:26:54 +01:00
|
|
|
const uint i12 = im%ne12;
|
|
|
|
const uint i13 = im/ne12;
|
|
|
|
|
|
|
|
const uint offset0 = r0*nb01 + (i12/r2)*nb02 + (i13/r3)*nb02*ne02;
|
|
|
|
|
|
|
|
device const half4 * x4 = (device const half4 *) (src0 + offset0);
|
2023-09-11 09:30:11 +02:00
|
|
|
|
|
|
|
for (int r1 = 0; r1 < nrows; ++r1) {
|
|
|
|
device const float4 * y4 = (device const float4 *) (src1 + r1*nb11 + im*nb12);
|
|
|
|
|
|
|
|
float sumf = 0;
|
|
|
|
for (int i = tiisg; i < ne00/4; i += 32) {
|
|
|
|
for (int k = 0; k < 4; ++k) sumf += (float) x4[i][k] * y4[i][k];
|
|
|
|
}
|
|
|
|
|
|
|
|
float all_sum = simd_sum(sumf);
|
|
|
|
if (tiisg == 0) {
|
|
|
|
dst[im*ne1*ne0 + r1*ne0 + r0] = all_sum;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2023-06-17 16:37:49 +02:00
|
|
|
kernel void kernel_alibi_f32(
|
|
|
|
device const float * src0,
|
|
|
|
device float * dst,
|
|
|
|
constant int64_t & ne00,
|
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant int64_t & ne03,
|
|
|
|
constant uint64_t & nb00,
|
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant uint64_t & nb02,
|
|
|
|
constant uint64_t & nb03,
|
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
|
|
|
constant int64_t & ne2,
|
|
|
|
constant int64_t & ne3,
|
|
|
|
constant uint64_t & nb0,
|
|
|
|
constant uint64_t & nb1,
|
|
|
|
constant uint64_t & nb2,
|
|
|
|
constant uint64_t & nb3,
|
2023-10-03 18:55:21 +02:00
|
|
|
constant float & m0,
|
|
|
|
constant float & m1,
|
|
|
|
constant int & n_heads_log2_floor,
|
2023-06-17 16:37:49 +02:00
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint3 tpitg[[thread_position_in_threadgroup]],
|
|
|
|
uint3 ntg[[threads_per_threadgroup]]) {
|
|
|
|
const int64_t i03 = tgpig[2];
|
|
|
|
const int64_t i02 = tgpig[1];
|
|
|
|
const int64_t i01 = tgpig[0];
|
|
|
|
|
|
|
|
const int64_t n = i03*ne02*ne01*ne00 + i02*ne01*ne00 + i01*ne00;
|
|
|
|
|
|
|
|
const int64_t i3 = n / (ne2*ne1*ne0);
|
|
|
|
const int64_t i2 = (n - i3*ne2*ne1*ne0) / (ne1*ne0);
|
|
|
|
const int64_t i1 = (n - i3*ne2*ne1*ne0 - i2*ne1*ne0) / ne0;
|
2024-01-02 09:57:44 +01:00
|
|
|
//const int64_t i0 = (n - i3*ne2*ne1*ne0 - i2*ne1*ne0 - i1*ne0);
|
|
|
|
|
2023-12-07 21:26:54 +01:00
|
|
|
const int64_t k = i3*ne3 + i2;
|
2023-06-17 16:37:49 +02:00
|
|
|
|
2023-10-03 18:55:21 +02:00
|
|
|
float m_k;
|
2023-12-07 21:26:54 +01:00
|
|
|
if (k < n_heads_log2_floor) {
|
|
|
|
m_k = pow(m0, k + 1);
|
2023-10-03 18:55:21 +02:00
|
|
|
} else {
|
2023-12-07 21:26:54 +01:00
|
|
|
m_k = pow(m1, 2 * (k - n_heads_log2_floor) + 1);
|
2023-10-03 18:55:21 +02:00
|
|
|
}
|
2023-12-07 21:26:54 +01:00
|
|
|
|
|
|
|
device char * dst_row = (device char *) dst + i3*nb3 + i2*nb2 + i1*nb1;
|
|
|
|
device const char * src_row = (device char *) src0 + i03*nb03 + i02*nb02 + i01*nb01;
|
2023-06-17 16:37:49 +02:00
|
|
|
for (int64_t i00 = tpitg.x; i00 < ne00; i00 += ntg.x) {
|
2023-12-07 21:26:54 +01:00
|
|
|
const float src_v = *(device float *)(src_row + i00*nb00);
|
|
|
|
device float * dst_v = (device float *)(dst_row + i00*nb0);
|
|
|
|
*dst_v = i00 * m_k + src_v;
|
2023-06-17 16:37:49 +02:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2023-11-01 23:04:33 +01:00
|
|
|
static float rope_yarn_ramp(const float low, const float high, const int i0) {
|
|
|
|
const float y = (i0 / 2 - low) / max(0.001f, high - low);
|
|
|
|
return 1.0f - min(1.0f, max(0.0f, y));
|
|
|
|
}
|
|
|
|
|
|
|
|
// YaRN algorithm based on LlamaYaRNScaledRotaryEmbedding.py from https://github.com/jquesnelle/yarn
|
|
|
|
// MIT licensed. Copyright (c) 2023 Jeffrey Quesnelle and Bowen Peng.
|
|
|
|
static void rope_yarn(
|
|
|
|
float theta_extrap, float freq_scale, float corr_dims[2], int64_t i0, float ext_factor, float mscale,
|
2023-11-02 07:33:37 +01:00
|
|
|
thread float * cos_theta, thread float * sin_theta
|
2023-11-01 23:04:33 +01:00
|
|
|
) {
|
|
|
|
// Get n-d rotational scaling corrected for extrapolation
|
|
|
|
float theta_interp = freq_scale * theta_extrap;
|
|
|
|
float theta = theta_interp;
|
|
|
|
if (ext_factor != 0.0f) {
|
2023-11-02 07:33:37 +01:00
|
|
|
float ramp_mix = rope_yarn_ramp(corr_dims[0], corr_dims[1], i0) * ext_factor;
|
2023-11-01 23:04:33 +01:00
|
|
|
theta = theta_interp * (1 - ramp_mix) + theta_extrap * ramp_mix;
|
|
|
|
|
|
|
|
// Get n-d magnitude scaling corrected for interpolation
|
2023-11-02 07:33:37 +01:00
|
|
|
mscale *= 1.0f + 0.1f * log(1.0f / freq_scale);
|
2023-11-01 23:04:33 +01:00
|
|
|
}
|
2023-11-02 07:33:37 +01:00
|
|
|
*cos_theta = cos(theta) * mscale;
|
|
|
|
*sin_theta = sin(theta) * mscale;
|
2023-11-01 23:04:33 +01:00
|
|
|
}
|
|
|
|
|
|
|
|
// Apparently solving `n_rot = 2pi * x * base^((2 * max_pos_emb) / n_dims)` for x, we get
|
|
|
|
// `corr_fac(n_rot) = n_dims * log(max_pos_emb / (n_rot * 2pi)) / (2 * log(base))`
|
|
|
|
static float rope_yarn_corr_factor(int n_dims, int n_orig_ctx, float n_rot, float base) {
|
|
|
|
return n_dims * log(n_orig_ctx / (n_rot * 2 * M_PI_F)) / (2 * log(base));
|
|
|
|
}
|
|
|
|
|
|
|
|
static void rope_yarn_corr_dims(
|
|
|
|
int n_dims, int n_orig_ctx, float freq_base, float beta_fast, float beta_slow, float dims[2]
|
|
|
|
) {
|
|
|
|
// start and end correction dims
|
|
|
|
dims[0] = max(0.0f, floor(rope_yarn_corr_factor(n_dims, n_orig_ctx, beta_fast, freq_base)));
|
|
|
|
dims[1] = min(n_dims - 1.0f, ceil(rope_yarn_corr_factor(n_dims, n_orig_ctx, beta_slow, freq_base)));
|
|
|
|
}
|
|
|
|
|
2023-09-28 18:04:36 +02:00
|
|
|
typedef void (rope_t)(
|
|
|
|
device const void * src0,
|
|
|
|
device const int32_t * src1,
|
|
|
|
device float * dst,
|
|
|
|
constant int64_t & ne00,
|
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant int64_t & ne03,
|
|
|
|
constant uint64_t & nb00,
|
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant uint64_t & nb02,
|
|
|
|
constant uint64_t & nb03,
|
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
|
|
|
constant int64_t & ne2,
|
|
|
|
constant int64_t & ne3,
|
|
|
|
constant uint64_t & nb0,
|
|
|
|
constant uint64_t & nb1,
|
|
|
|
constant uint64_t & nb2,
|
|
|
|
constant uint64_t & nb3,
|
|
|
|
constant int & n_past,
|
|
|
|
constant int & n_dims,
|
|
|
|
constant int & mode,
|
2023-11-02 07:33:37 +01:00
|
|
|
constant int & n_orig_ctx,
|
2023-09-28 18:04:36 +02:00
|
|
|
constant float & freq_base,
|
|
|
|
constant float & freq_scale,
|
2023-11-02 07:33:37 +01:00
|
|
|
constant float & ext_factor,
|
|
|
|
constant float & attn_factor,
|
|
|
|
constant float & beta_fast,
|
|
|
|
constant float & beta_slow,
|
2023-09-28 18:04:36 +02:00
|
|
|
uint tiitg[[thread_index_in_threadgroup]],
|
|
|
|
uint3 tptg[[threads_per_threadgroup]],
|
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]]);
|
|
|
|
|
|
|
|
template<typename T>
|
2023-06-04 22:34:30 +02:00
|
|
|
kernel void kernel_rope(
|
2023-09-28 18:04:36 +02:00
|
|
|
device const void * src0,
|
|
|
|
device const int32_t * src1,
|
|
|
|
device float * dst,
|
|
|
|
constant int64_t & ne00,
|
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant int64_t & ne03,
|
|
|
|
constant uint64_t & nb00,
|
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant uint64_t & nb02,
|
|
|
|
constant uint64_t & nb03,
|
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
|
|
|
constant int64_t & ne2,
|
|
|
|
constant int64_t & ne3,
|
|
|
|
constant uint64_t & nb0,
|
|
|
|
constant uint64_t & nb1,
|
|
|
|
constant uint64_t & nb2,
|
|
|
|
constant uint64_t & nb3,
|
|
|
|
constant int & n_past,
|
|
|
|
constant int & n_dims,
|
|
|
|
constant int & mode,
|
2023-11-02 07:33:37 +01:00
|
|
|
constant int & n_orig_ctx,
|
2023-09-28 18:04:36 +02:00
|
|
|
constant float & freq_base,
|
|
|
|
constant float & freq_scale,
|
2023-11-01 23:04:33 +01:00
|
|
|
constant float & ext_factor,
|
|
|
|
constant float & attn_factor,
|
|
|
|
constant float & beta_fast,
|
|
|
|
constant float & beta_slow,
|
2023-09-07 15:45:01 +02:00
|
|
|
uint tiitg[[thread_index_in_threadgroup]],
|
|
|
|
uint3 tptg[[threads_per_threadgroup]],
|
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]]) {
|
|
|
|
const int64_t i3 = tgpig[2];
|
|
|
|
const int64_t i2 = tgpig[1];
|
|
|
|
const int64_t i1 = tgpig[0];
|
2023-06-04 22:34:30 +02:00
|
|
|
|
|
|
|
const bool is_neox = mode & 2;
|
|
|
|
|
2023-11-01 23:04:33 +01:00
|
|
|
float corr_dims[2];
|
|
|
|
rope_yarn_corr_dims(n_dims, n_orig_ctx, freq_base, beta_fast, beta_slow, corr_dims);
|
|
|
|
|
2023-09-28 18:04:36 +02:00
|
|
|
device const int32_t * pos = src1;
|
|
|
|
|
|
|
|
const int64_t p = pos[i2];
|
2023-06-04 22:34:30 +02:00
|
|
|
|
2023-11-01 23:04:33 +01:00
|
|
|
const float theta_0 = (float)p;
|
2023-09-07 15:45:01 +02:00
|
|
|
const float inv_ndims = -1.f/n_dims;
|
2023-06-04 22:34:30 +02:00
|
|
|
|
|
|
|
if (!is_neox) {
|
2023-09-07 15:45:01 +02:00
|
|
|
for (int64_t i0 = 2*tiitg; i0 < ne0; i0 += 2*tptg.x) {
|
|
|
|
|
|
|
|
const float theta = theta_0 * pow(freq_base, inv_ndims*i0);
|
2023-11-01 23:04:33 +01:00
|
|
|
float cos_theta, sin_theta;
|
|
|
|
rope_yarn(theta, freq_scale, corr_dims, i0, ext_factor, attn_factor, &cos_theta, &sin_theta);
|
2023-06-04 22:34:30 +02:00
|
|
|
|
2023-09-28 18:04:36 +02:00
|
|
|
device const T * const src = (device T *)((device char *) src0 + i3*nb03 + i2*nb02 + i1*nb01 + i0*nb00);
|
|
|
|
device T * dst_data = (device T *)((device char *) dst + i3*nb3 + i2*nb2 + i1*nb1 + i0*nb0);
|
2023-06-04 22:34:30 +02:00
|
|
|
|
2023-09-28 18:04:36 +02:00
|
|
|
const T x0 = src[0];
|
|
|
|
const T x1 = src[1];
|
2023-06-04 22:34:30 +02:00
|
|
|
|
|
|
|
dst_data[0] = x0*cos_theta - x1*sin_theta;
|
|
|
|
dst_data[1] = x0*sin_theta + x1*cos_theta;
|
|
|
|
}
|
|
|
|
} else {
|
2023-12-18 18:27:47 +01:00
|
|
|
for (int64_t ic = 2*tiitg; ic < ne0; ic += 2*tptg.x) {
|
|
|
|
if (ic < n_dims) {
|
|
|
|
const int64_t ib = 0;
|
2023-09-07 15:45:01 +02:00
|
|
|
|
2023-11-01 23:04:33 +01:00
|
|
|
// simplified from `(ib * n_dims + ic) * inv_ndims`
|
|
|
|
const float cur_rot = inv_ndims*ic - ib;
|
|
|
|
|
|
|
|
const float theta = theta_0 * pow(freq_base, cur_rot);
|
|
|
|
float cos_theta, sin_theta;
|
|
|
|
rope_yarn(theta, freq_scale, corr_dims, cur_rot, ext_factor, attn_factor, &cos_theta, &sin_theta);
|
2023-08-23 22:08:04 +02:00
|
|
|
|
|
|
|
const int64_t i0 = ib*n_dims + ic/2;
|
|
|
|
|
2023-09-28 18:04:36 +02:00
|
|
|
device const T * const src = (device T *)((device char *) src0 + i3*nb03 + i2*nb02 + i1*nb01 + i0*nb00);
|
|
|
|
device T * dst_data = (device T *)((device char *) dst + i3*nb3 + i2*nb2 + i1*nb1 + i0*nb0);
|
2023-08-23 22:08:04 +02:00
|
|
|
|
|
|
|
const float x0 = src[0];
|
|
|
|
const float x1 = src[n_dims/2];
|
|
|
|
|
|
|
|
dst_data[0] = x0*cos_theta - x1*sin_theta;
|
|
|
|
dst_data[n_dims/2] = x0*sin_theta + x1*cos_theta;
|
2023-12-18 18:27:47 +01:00
|
|
|
} else {
|
|
|
|
const int64_t i0 = ic;
|
|
|
|
|
|
|
|
device const T * const src = (device T *)((device char *) src0 + i3*nb03 + i2*nb02 + i1*nb01 + i0*nb00);
|
|
|
|
device T * dst_data = (device T *)((device char *) dst + i3*nb3 + i2*nb2 + i1*nb1 + i0*nb0);
|
|
|
|
|
|
|
|
dst_data[0] = src[0];
|
|
|
|
dst_data[1] = src[1];
|
2023-08-23 22:08:04 +02:00
|
|
|
}
|
|
|
|
}
|
2023-06-04 22:34:30 +02:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2023-09-28 18:04:36 +02:00
|
|
|
template [[host_name("kernel_rope_f32")]] kernel rope_t kernel_rope<float>;
|
|
|
|
template [[host_name("kernel_rope_f16")]] kernel rope_t kernel_rope<half>;
|
|
|
|
|
2023-11-13 15:55:52 +01:00
|
|
|
kernel void kernel_im2col_f16(
|
|
|
|
device const float * x,
|
|
|
|
device half * dst,
|
|
|
|
constant int32_t & ofs0,
|
|
|
|
constant int32_t & ofs1,
|
|
|
|
constant int32_t & IW,
|
|
|
|
constant int32_t & IH,
|
|
|
|
constant int32_t & CHW,
|
|
|
|
constant int32_t & s0,
|
|
|
|
constant int32_t & s1,
|
|
|
|
constant int32_t & p0,
|
|
|
|
constant int32_t & p1,
|
|
|
|
constant int32_t & d0,
|
|
|
|
constant int32_t & d1,
|
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint3 tgpg[[threadgroups_per_grid]],
|
|
|
|
uint3 tpitg[[thread_position_in_threadgroup]],
|
|
|
|
uint3 ntg[[threads_per_threadgroup]]) {
|
|
|
|
const int32_t iiw = tgpig[2] * s0 + tpitg[2] * d0 - p0;
|
|
|
|
const int32_t iih = tgpig[1] * s1 + tpitg[1] * d1 - p1;
|
|
|
|
|
|
|
|
const int32_t offset_dst =
|
|
|
|
(tpitg[0] * tgpg[1] * tgpg[2] + tgpig[1] * tgpg[2] + tgpig[2]) * CHW +
|
|
|
|
(tgpig[0] * (ntg[1] * ntg[2]) + tpitg[1] * ntg[2] + tpitg[2]);
|
|
|
|
|
|
|
|
if (iih < 0 || iih >= IH || iiw < 0 || iiw >= IW) {
|
|
|
|
dst[offset_dst] = 0.0f;
|
|
|
|
} else {
|
|
|
|
const int32_t offset_src = tpitg[0] * ofs0 + tgpig[0] * ofs1;
|
|
|
|
dst[offset_dst] = x[offset_src + iih * IW + iiw];
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2023-12-13 20:54:54 +01:00
|
|
|
kernel void kernel_upscale_f32(
|
|
|
|
device const char * src0,
|
|
|
|
device char * dst,
|
|
|
|
constant int64_t & ne00,
|
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant int64_t & ne03,
|
|
|
|
constant uint64_t & nb00,
|
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant uint64_t & nb02,
|
|
|
|
constant uint64_t & nb03,
|
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
|
|
|
constant int64_t & ne2,
|
|
|
|
constant int64_t & ne3,
|
|
|
|
constant uint64_t & nb0,
|
|
|
|
constant uint64_t & nb1,
|
|
|
|
constant uint64_t & nb2,
|
|
|
|
constant uint64_t & nb3,
|
|
|
|
constant int32_t & sf,
|
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint3 tpitg[[thread_position_in_threadgroup]],
|
|
|
|
uint3 ntg[[threads_per_threadgroup]]) {
|
|
|
|
|
|
|
|
const int64_t i3 = tgpig.z;
|
|
|
|
const int64_t i2 = tgpig.y;
|
|
|
|
const int64_t i1 = tgpig.x;
|
|
|
|
|
|
|
|
const int64_t i03 = i3;
|
|
|
|
const int64_t i02 = i2;
|
|
|
|
const int64_t i01 = i1/sf;
|
|
|
|
|
|
|
|
device const float * src0_ptr = (device const float *) (src0 + i03*nb03 + i02*nb02 + i01*nb01);
|
|
|
|
device float * dst_ptr = (device float *) (dst + i3*nb3 + i2*nb2 + i1*nb1);
|
|
|
|
|
|
|
|
for (int i0 = tpitg.x; i0 < ne0; i0 += ntg.x) {
|
|
|
|
dst_ptr[i0] = src0_ptr[i0/sf];
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
kernel void kernel_pad_f32(
|
|
|
|
device const char * src0,
|
|
|
|
device char * dst,
|
|
|
|
constant int64_t & ne00,
|
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant int64_t & ne03,
|
|
|
|
constant uint64_t & nb00,
|
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant uint64_t & nb02,
|
|
|
|
constant uint64_t & nb03,
|
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
|
|
|
constant int64_t & ne2,
|
|
|
|
constant int64_t & ne3,
|
|
|
|
constant uint64_t & nb0,
|
|
|
|
constant uint64_t & nb1,
|
|
|
|
constant uint64_t & nb2,
|
|
|
|
constant uint64_t & nb3,
|
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint3 tpitg[[thread_position_in_threadgroup]],
|
|
|
|
uint3 ntg[[threads_per_threadgroup]]) {
|
|
|
|
|
|
|
|
const int64_t i3 = tgpig.z;
|
|
|
|
const int64_t i2 = tgpig.y;
|
|
|
|
const int64_t i1 = tgpig.x;
|
|
|
|
|
|
|
|
const int64_t i03 = i3;
|
|
|
|
const int64_t i02 = i2;
|
|
|
|
const int64_t i01 = i1;
|
|
|
|
|
|
|
|
device const float * src0_ptr = (device const float *) (src0 + i03*nb03 + i02*nb02 + i01*nb01);
|
|
|
|
device float * dst_ptr = (device float *) (dst + i3*nb3 + i2*nb2 + i1*nb1);
|
|
|
|
|
|
|
|
if (i1 < ne01 && i2 < ne02 && i3 < ne03) {
|
|
|
|
for (int i0 = tpitg.x; i0 < ne0; i0 += ntg.x) {
|
|
|
|
if (i0 < ne00) {
|
|
|
|
dst_ptr[i0] = src0_ptr[i0];
|
|
|
|
} else {
|
|
|
|
dst_ptr[i0] = 0.0f;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
for (int i0 = tpitg.x; i0 < ne0; i0 += ntg.x) {
|
|
|
|
dst_ptr[i0] = 0.0f;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2023-12-07 21:26:54 +01:00
|
|
|
// bitonic sort implementation following the CUDA kernels as reference
|
|
|
|
typedef void (argsort_t)(
|
|
|
|
device const float * x,
|
|
|
|
device int32_t * dst,
|
|
|
|
constant int64_t & ncols,
|
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint3 tpitg[[thread_position_in_threadgroup]]);
|
|
|
|
|
|
|
|
template<ggml_sort_order order>
|
|
|
|
kernel void kernel_argsort_f32_i32(
|
|
|
|
device const float * x,
|
|
|
|
device int32_t * dst,
|
|
|
|
constant int64_t & ncols,
|
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint3 tpitg[[thread_position_in_threadgroup]]) {
|
|
|
|
// bitonic sort
|
|
|
|
int col = tpitg[0];
|
|
|
|
int row = tgpig[1];
|
|
|
|
|
|
|
|
if (col >= ncols) return;
|
|
|
|
|
|
|
|
device const float * x_row = x + row * ncols;
|
|
|
|
device int32_t * dst_row = dst + row * ncols;
|
|
|
|
|
|
|
|
// initialize indices
|
|
|
|
if (col < ncols) {
|
|
|
|
dst_row[col] = col;
|
|
|
|
}
|
|
|
|
threadgroup_barrier(mem_flags::mem_threadgroup);
|
|
|
|
|
|
|
|
for (int k = 2; k <= ncols; k *= 2) {
|
|
|
|
for (int j = k / 2; j > 0; j /= 2) {
|
|
|
|
int ixj = col ^ j;
|
|
|
|
if (ixj > col) {
|
|
|
|
if ((col & k) == 0) {
|
|
|
|
if (order == GGML_SORT_ASC ? x_row[dst_row[col]] > x_row[dst_row[ixj]] : x_row[dst_row[col]] < x_row[dst_row[ixj]]) {
|
|
|
|
SWAP(dst_row[col], dst_row[ixj]);
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
if (order == GGML_SORT_ASC ? x_row[dst_row[col]] < x_row[dst_row[ixj]] : x_row[dst_row[col]] > x_row[dst_row[ixj]]) {
|
|
|
|
SWAP(dst_row[col], dst_row[ixj]);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
threadgroup_barrier(mem_flags::mem_threadgroup);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
template [[host_name("kernel_argsort_f32_i32_asc")]] kernel argsort_t kernel_argsort_f32_i32<GGML_SORT_ASC>;
|
|
|
|
template [[host_name("kernel_argsort_f32_i32_desc")]] kernel argsort_t kernel_argsort_f32_i32<GGML_SORT_DESC>;
|
|
|
|
|
2023-12-13 20:54:54 +01:00
|
|
|
kernel void kernel_leaky_relu_f32(
|
|
|
|
device const float * src0,
|
|
|
|
device float * dst,
|
|
|
|
constant float & slope,
|
|
|
|
uint tpig[[thread_position_in_grid]]) {
|
|
|
|
dst[tpig] = src0[tpig] > 0.0f ? src0[tpig] : src0[tpig] * slope;
|
|
|
|
}
|
|
|
|
|
2023-06-17 16:37:49 +02:00
|
|
|
kernel void kernel_cpy_f16_f16(
|
2023-12-13 13:04:25 +01:00
|
|
|
device const half * src0,
|
|
|
|
device half * dst,
|
2023-06-17 16:37:49 +02:00
|
|
|
constant int64_t & ne00,
|
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant int64_t & ne03,
|
|
|
|
constant uint64_t & nb00,
|
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant uint64_t & nb02,
|
|
|
|
constant uint64_t & nb03,
|
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
|
|
|
constant int64_t & ne2,
|
|
|
|
constant int64_t & ne3,
|
|
|
|
constant uint64_t & nb0,
|
|
|
|
constant uint64_t & nb1,
|
|
|
|
constant uint64_t & nb2,
|
|
|
|
constant uint64_t & nb3,
|
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint3 tpitg[[thread_position_in_threadgroup]],
|
|
|
|
uint3 ntg[[threads_per_threadgroup]]) {
|
|
|
|
const int64_t i03 = tgpig[2];
|
|
|
|
const int64_t i02 = tgpig[1];
|
|
|
|
const int64_t i01 = tgpig[0];
|
|
|
|
|
|
|
|
const int64_t n = i03*ne02*ne01*ne00 + i02*ne01*ne00 + i01*ne00;
|
|
|
|
|
|
|
|
const int64_t i3 = n / (ne2*ne1*ne0);
|
|
|
|
const int64_t i2 = (n - i3*ne2*ne1*ne0) / (ne1*ne0);
|
|
|
|
const int64_t i1 = (n - i3*ne2*ne1*ne0 - i2*ne1*ne0) / ne0;
|
|
|
|
const int64_t i0 = (n - i3*ne2*ne1*ne0 - i2*ne1*ne0 - i1*ne0);
|
|
|
|
|
|
|
|
device half * dst_data = (device half *) ((device char *) dst + i3*nb3 + i2*nb2 + i1*nb1 + i0*nb0);
|
|
|
|
|
|
|
|
for (int64_t i00 = tpitg.x; i00 < ne00; i00 += ntg.x) {
|
|
|
|
device const half * src = (device half *)((device char *) src0 + i03*nb03 + i02*nb02 + i01*nb01 + i00*nb00);
|
|
|
|
dst_data[i00] = src[0];
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2023-12-13 13:04:25 +01:00
|
|
|
kernel void kernel_cpy_f16_f32(
|
|
|
|
device const half * src0,
|
|
|
|
device float * dst,
|
|
|
|
constant int64_t & ne00,
|
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant int64_t & ne03,
|
|
|
|
constant uint64_t & nb00,
|
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant uint64_t & nb02,
|
|
|
|
constant uint64_t & nb03,
|
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
|
|
|
constant int64_t & ne2,
|
|
|
|
constant int64_t & ne3,
|
|
|
|
constant uint64_t & nb0,
|
|
|
|
constant uint64_t & nb1,
|
|
|
|
constant uint64_t & nb2,
|
|
|
|
constant uint64_t & nb3,
|
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint3 tpitg[[thread_position_in_threadgroup]],
|
|
|
|
uint3 ntg[[threads_per_threadgroup]]) {
|
|
|
|
const int64_t i03 = tgpig[2];
|
|
|
|
const int64_t i02 = tgpig[1];
|
|
|
|
const int64_t i01 = tgpig[0];
|
|
|
|
|
|
|
|
const int64_t n = i03*ne02*ne01*ne00 + i02*ne01*ne00 + i01*ne00;
|
|
|
|
|
|
|
|
const int64_t i3 = n / (ne2*ne1*ne0);
|
|
|
|
const int64_t i2 = (n - i3*ne2*ne1*ne0) / (ne1*ne0);
|
|
|
|
const int64_t i1 = (n - i3*ne2*ne1*ne0 - i2*ne1*ne0) / ne0;
|
|
|
|
const int64_t i0 = (n - i3*ne2*ne1*ne0 - i2*ne1*ne0 - i1*ne0);
|
|
|
|
|
|
|
|
device float * dst_data = (device float *) ((device char *) dst + i3*nb3 + i2*nb2 + i1*nb1 + i0*nb0);
|
|
|
|
|
|
|
|
for (int64_t i00 = tpitg.x; i00 < ne00; i00 += ntg.x) {
|
|
|
|
device const half * src = (device half *)((device char *) src0 + i03*nb03 + i02*nb02 + i01*nb01 + i00*nb00);
|
|
|
|
dst_data[i00] = src[0];
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2023-06-04 22:34:30 +02:00
|
|
|
kernel void kernel_cpy_f32_f16(
|
|
|
|
device const float * src0,
|
|
|
|
device half * dst,
|
|
|
|
constant int64_t & ne00,
|
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant int64_t & ne03,
|
|
|
|
constant uint64_t & nb00,
|
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant uint64_t & nb02,
|
|
|
|
constant uint64_t & nb03,
|
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
|
|
|
constant int64_t & ne2,
|
|
|
|
constant int64_t & ne3,
|
|
|
|
constant uint64_t & nb0,
|
|
|
|
constant uint64_t & nb1,
|
|
|
|
constant uint64_t & nb2,
|
|
|
|
constant uint64_t & nb3,
|
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint3 tpitg[[thread_position_in_threadgroup]],
|
|
|
|
uint3 ntg[[threads_per_threadgroup]]) {
|
|
|
|
const int64_t i03 = tgpig[2];
|
|
|
|
const int64_t i02 = tgpig[1];
|
|
|
|
const int64_t i01 = tgpig[0];
|
|
|
|
|
|
|
|
const int64_t n = i03*ne02*ne01*ne00 + i02*ne01*ne00 + i01*ne00;
|
|
|
|
|
|
|
|
const int64_t i3 = n / (ne2*ne1*ne0);
|
|
|
|
const int64_t i2 = (n - i3*ne2*ne1*ne0) / (ne1*ne0);
|
|
|
|
const int64_t i1 = (n - i3*ne2*ne1*ne0 - i2*ne1*ne0) / ne0;
|
|
|
|
const int64_t i0 = (n - i3*ne2*ne1*ne0 - i2*ne1*ne0 - i1*ne0);
|
|
|
|
|
|
|
|
device half * dst_data = (device half *) ((device char *) dst + i3*nb3 + i2*nb2 + i1*nb1 + i0*nb0);
|
|
|
|
|
|
|
|
for (int64_t i00 = tpitg.x; i00 < ne00; i00 += ntg.x) {
|
|
|
|
device const float * src = (device float *)((device char *) src0 + i03*nb03 + i02*nb02 + i01*nb01 + i00*nb00);
|
|
|
|
|
|
|
|
dst_data[i00] = src[0];
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
kernel void kernel_cpy_f32_f32(
|
|
|
|
device const float * src0,
|
|
|
|
device float * dst,
|
|
|
|
constant int64_t & ne00,
|
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant int64_t & ne03,
|
|
|
|
constant uint64_t & nb00,
|
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant uint64_t & nb02,
|
|
|
|
constant uint64_t & nb03,
|
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
|
|
|
constant int64_t & ne2,
|
|
|
|
constant int64_t & ne3,
|
|
|
|
constant uint64_t & nb0,
|
|
|
|
constant uint64_t & nb1,
|
|
|
|
constant uint64_t & nb2,
|
|
|
|
constant uint64_t & nb3,
|
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint3 tpitg[[thread_position_in_threadgroup]],
|
|
|
|
uint3 ntg[[threads_per_threadgroup]]) {
|
|
|
|
const int64_t i03 = tgpig[2];
|
|
|
|
const int64_t i02 = tgpig[1];
|
|
|
|
const int64_t i01 = tgpig[0];
|
|
|
|
|
|
|
|
const int64_t n = i03*ne02*ne01*ne00 + i02*ne01*ne00 + i01*ne00;
|
|
|
|
|
|
|
|
const int64_t i3 = n / (ne2*ne1*ne0);
|
|
|
|
const int64_t i2 = (n - i3*ne2*ne1*ne0) / (ne1*ne0);
|
|
|
|
const int64_t i1 = (n - i3*ne2*ne1*ne0 - i2*ne1*ne0) / ne0;
|
|
|
|
const int64_t i0 = (n - i3*ne2*ne1*ne0 - i2*ne1*ne0 - i1*ne0);
|
|
|
|
|
|
|
|
device float * dst_data = (device float *) ((device char *) dst + i3*nb3 + i2*nb2 + i1*nb1 + i0*nb0);
|
|
|
|
|
|
|
|
for (int64_t i00 = tpitg.x; i00 < ne00; i00 += ntg.x) {
|
|
|
|
device const float * src = (device float *)((device char *) src0 + i03*nb03 + i02*nb02 + i01*nb01 + i00*nb00);
|
|
|
|
|
|
|
|
dst_data[i00] = src[0];
|
|
|
|
}
|
|
|
|
}
|
2023-06-08 09:08:23 +02:00
|
|
|
|
2023-12-07 12:03:17 +01:00
|
|
|
kernel void kernel_cpy_f32_q8_0(
|
|
|
|
device const float * src0,
|
|
|
|
device void * dst,
|
|
|
|
constant int64_t & ne00,
|
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant int64_t & ne03,
|
|
|
|
constant uint64_t & nb00,
|
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant uint64_t & nb02,
|
|
|
|
constant uint64_t & nb03,
|
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
|
|
|
constant int64_t & ne2,
|
|
|
|
constant int64_t & ne3,
|
|
|
|
constant uint64_t & nb0,
|
|
|
|
constant uint64_t & nb1,
|
|
|
|
constant uint64_t & nb2,
|
|
|
|
constant uint64_t & nb3,
|
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint3 tpitg[[thread_position_in_threadgroup]],
|
|
|
|
uint3 ntg[[threads_per_threadgroup]]) {
|
|
|
|
const int64_t i03 = tgpig[2];
|
|
|
|
const int64_t i02 = tgpig[1];
|
|
|
|
const int64_t i01 = tgpig[0];
|
|
|
|
|
|
|
|
const int64_t n = i03*ne02*ne01*ne00 + i02*ne01*ne00 + i01*ne00;
|
|
|
|
|
|
|
|
const int64_t i3 = n / (ne2*ne1*ne0);
|
|
|
|
const int64_t i2 = (n - i3*ne2*ne1*ne0) / (ne1*ne0);
|
|
|
|
const int64_t i1 = (n - i3*ne2*ne1*ne0 - i2*ne1*ne0) / ne0;
|
|
|
|
const int64_t i0 = (n - i3*ne2*ne1*ne0 - i2*ne1*ne0 - i1*ne0)/QK8_0;
|
|
|
|
|
|
|
|
device block_q8_0 * dst_data = (device block_q8_0 *) ((device char *) dst + i3*nb3 + i2*nb2 + i1*nb1 + i0*nb0);
|
|
|
|
|
|
|
|
for (int64_t i00 = tpitg.x*QK8_0; i00 < ne00; i00 += ntg.x*QK8_0) {
|
|
|
|
device const float * src = (device float *)((device char *) src0 + i03*nb03 + i02*nb02 + i01*nb01 + i00*nb00);
|
|
|
|
|
|
|
|
float amax = 0.0f; // absolute max
|
|
|
|
|
|
|
|
for (int j = 0; j < QK8_0; j++) {
|
|
|
|
const float v = src[j];
|
|
|
|
amax = MAX(amax, fabs(v));
|
|
|
|
}
|
|
|
|
|
|
|
|
const float d = amax / ((1 << 7) - 1);
|
|
|
|
const float id = d ? 1.0f/d : 0.0f;
|
|
|
|
|
|
|
|
dst_data[i00/QK8_0].d = d;
|
|
|
|
|
|
|
|
for (int j = 0; j < QK8_0; ++j) {
|
|
|
|
const float x0 = src[j]*id;
|
|
|
|
|
|
|
|
dst_data[i00/QK8_0].qs[j] = round(x0);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
kernel void kernel_cpy_f32_q4_0(
|
|
|
|
device const float * src0,
|
|
|
|
device void * dst,
|
|
|
|
constant int64_t & ne00,
|
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant int64_t & ne03,
|
|
|
|
constant uint64_t & nb00,
|
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant uint64_t & nb02,
|
|
|
|
constant uint64_t & nb03,
|
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
|
|
|
constant int64_t & ne2,
|
|
|
|
constant int64_t & ne3,
|
|
|
|
constant uint64_t & nb0,
|
|
|
|
constant uint64_t & nb1,
|
|
|
|
constant uint64_t & nb2,
|
|
|
|
constant uint64_t & nb3,
|
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint3 tpitg[[thread_position_in_threadgroup]],
|
|
|
|
uint3 ntg[[threads_per_threadgroup]]) {
|
|
|
|
const int64_t i03 = tgpig[2];
|
|
|
|
const int64_t i02 = tgpig[1];
|
|
|
|
const int64_t i01 = tgpig[0];
|
|
|
|
|
|
|
|
const int64_t n = i03*ne02*ne01*ne00 + i02*ne01*ne00 + i01*ne00;
|
|
|
|
|
|
|
|
const int64_t i3 = n / (ne2*ne1*ne0);
|
|
|
|
const int64_t i2 = (n - i3*ne2*ne1*ne0) / (ne1*ne0);
|
|
|
|
const int64_t i1 = (n - i3*ne2*ne1*ne0 - i2*ne1*ne0) / ne0;
|
|
|
|
const int64_t i0 = (n - i3*ne2*ne1*ne0 - i2*ne1*ne0 - i1*ne0)/QK4_0;
|
|
|
|
|
|
|
|
device block_q4_0 * dst_data = (device block_q4_0 *) ((device char *) dst + i3*nb3 + i2*nb2 + i1*nb1 + i0*nb0);
|
|
|
|
|
|
|
|
for (int64_t i00 = tpitg.x*QK4_0; i00 < ne00; i00 += ntg.x*QK4_0) {
|
|
|
|
device const float * src = (device float *)((device char *) src0 + i03*nb03 + i02*nb02 + i01*nb01 + i00*nb00);
|
|
|
|
|
|
|
|
float amax = 0.0f; // absolute max
|
|
|
|
float max = 0.0f;
|
|
|
|
|
|
|
|
for (int j = 0; j < QK4_0; j++) {
|
|
|
|
const float v = src[j];
|
|
|
|
if (amax < fabs(v)) {
|
|
|
|
amax = fabs(v);
|
|
|
|
max = v;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
const float d = max / -8;
|
|
|
|
const float id = d ? 1.0f/d : 0.0f;
|
|
|
|
|
|
|
|
dst_data[i00/QK4_0].d = d;
|
|
|
|
|
|
|
|
for (int j = 0; j < QK4_0/2; ++j) {
|
|
|
|
const float x0 = src[0 + j]*id;
|
|
|
|
const float x1 = src[QK4_0/2 + j]*id;
|
|
|
|
|
|
|
|
const uint8_t xi0 = MIN(15, (int8_t)(x0 + 8.5f));
|
|
|
|
const uint8_t xi1 = MIN(15, (int8_t)(x1 + 8.5f));
|
|
|
|
|
|
|
|
dst_data[i00/QK4_0].qs[j] = xi0;
|
|
|
|
dst_data[i00/QK4_0].qs[j] |= xi1 << 4;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
kernel void kernel_cpy_f32_q4_1(
|
|
|
|
device const float * src0,
|
|
|
|
device void * dst,
|
|
|
|
constant int64_t & ne00,
|
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant int64_t & ne03,
|
|
|
|
constant uint64_t & nb00,
|
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant uint64_t & nb02,
|
|
|
|
constant uint64_t & nb03,
|
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
|
|
|
constant int64_t & ne2,
|
|
|
|
constant int64_t & ne3,
|
|
|
|
constant uint64_t & nb0,
|
|
|
|
constant uint64_t & nb1,
|
|
|
|
constant uint64_t & nb2,
|
|
|
|
constant uint64_t & nb3,
|
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint3 tpitg[[thread_position_in_threadgroup]],
|
|
|
|
uint3 ntg[[threads_per_threadgroup]]) {
|
|
|
|
const int64_t i03 = tgpig[2];
|
|
|
|
const int64_t i02 = tgpig[1];
|
|
|
|
const int64_t i01 = tgpig[0];
|
|
|
|
|
|
|
|
const int64_t n = i03*ne02*ne01*ne00 + i02*ne01*ne00 + i01*ne00;
|
|
|
|
|
|
|
|
const int64_t i3 = n / (ne2*ne1*ne0);
|
|
|
|
const int64_t i2 = (n - i3*ne2*ne1*ne0) / (ne1*ne0);
|
|
|
|
const int64_t i1 = (n - i3*ne2*ne1*ne0 - i2*ne1*ne0) / ne0;
|
|
|
|
const int64_t i0 = (n - i3*ne2*ne1*ne0 - i2*ne1*ne0 - i1*ne0)/QK4_1;
|
|
|
|
|
|
|
|
device block_q4_1 * dst_data = (device block_q4_1 *) ((device char *) dst + i3*nb3 + i2*nb2 + i1*nb1 + i0*nb0);
|
|
|
|
|
|
|
|
for (int64_t i00 = tpitg.x*QK4_1; i00 < ne00; i00 += ntg.x*QK4_1) {
|
|
|
|
device const float * src = (device float *)((device char *) src0 + i03*nb03 + i02*nb02 + i01*nb01 + i00*nb00);
|
|
|
|
|
|
|
|
float min = FLT_MAX;
|
|
|
|
float max = -FLT_MAX;
|
|
|
|
|
|
|
|
for (int j = 0; j < QK4_1; j++) {
|
|
|
|
const float v = src[j];
|
|
|
|
if (min > v) min = v;
|
|
|
|
if (max < v) max = v;
|
|
|
|
}
|
|
|
|
|
|
|
|
const float d = (max - min) / ((1 << 4) - 1);
|
|
|
|
const float id = d ? 1.0f/d : 0.0f;
|
|
|
|
|
|
|
|
dst_data[i00/QK4_1].d = d;
|
|
|
|
dst_data[i00/QK4_1].m = min;
|
|
|
|
|
|
|
|
for (int j = 0; j < QK4_1/2; ++j) {
|
|
|
|
const float x0 = (src[0 + j] - min)*id;
|
|
|
|
const float x1 = (src[QK4_1/2 + j] - min)*id;
|
|
|
|
|
|
|
|
const uint8_t xi0 = MIN(15, (int8_t)(x0 + 0.5f));
|
|
|
|
const uint8_t xi1 = MIN(15, (int8_t)(x1 + 0.5f));
|
|
|
|
|
|
|
|
dst_data[i00/QK4_1].qs[j] = xi0;
|
|
|
|
dst_data[i00/QK4_1].qs[j] |= xi1 << 4;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2023-10-07 09:12:43 +02:00
|
|
|
kernel void kernel_concat(
|
2023-12-13 20:54:54 +01:00
|
|
|
device const char * src0,
|
|
|
|
device const char * src1,
|
|
|
|
device char * dst,
|
2023-10-07 09:12:43 +02:00
|
|
|
constant int64_t & ne00,
|
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant int64_t & ne03,
|
|
|
|
constant uint64_t & nb00,
|
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant uint64_t & nb02,
|
|
|
|
constant uint64_t & nb03,
|
|
|
|
constant int64_t & ne10,
|
|
|
|
constant int64_t & ne11,
|
|
|
|
constant int64_t & ne12,
|
|
|
|
constant int64_t & ne13,
|
|
|
|
constant uint64_t & nb10,
|
|
|
|
constant uint64_t & nb11,
|
|
|
|
constant uint64_t & nb12,
|
|
|
|
constant uint64_t & nb13,
|
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
|
|
|
constant int64_t & ne2,
|
|
|
|
constant int64_t & ne3,
|
|
|
|
constant uint64_t & nb0,
|
|
|
|
constant uint64_t & nb1,
|
|
|
|
constant uint64_t & nb2,
|
|
|
|
constant uint64_t & nb3,
|
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint3 tpitg[[thread_position_in_threadgroup]],
|
|
|
|
uint3 ntg[[threads_per_threadgroup]]) {
|
|
|
|
|
|
|
|
const int64_t i03 = tgpig.z;
|
|
|
|
const int64_t i02 = tgpig.y;
|
|
|
|
const int64_t i01 = tgpig.x;
|
|
|
|
|
|
|
|
const int64_t i13 = i03 % ne13;
|
|
|
|
const int64_t i12 = i02 % ne12;
|
|
|
|
const int64_t i11 = i01 % ne11;
|
|
|
|
|
2023-12-13 20:54:54 +01:00
|
|
|
device const char * src0_ptr = src0 + i03*nb03 + i02*nb02 + i01*nb01 + tpitg.x*nb00;
|
2023-10-07 09:12:43 +02:00
|
|
|
device const char * src1_ptr = src1 + i13*nb13 + i12*nb12 + i11*nb11 + tpitg.x*nb10;
|
|
|
|
device char * dst_ptr = dst + i03*nb3 + i02*nb2 + i01*nb1 + tpitg.x*nb0;
|
|
|
|
|
|
|
|
for (int i0 = tpitg.x; i0 < ne0; i0 += ntg.x) {
|
|
|
|
if (i02 < ne02) {
|
|
|
|
((device float *)dst_ptr)[0] = ((device float *)src0_ptr)[0];
|
|
|
|
src0_ptr += ntg.x*nb00;
|
|
|
|
} else {
|
|
|
|
((device float *)dst_ptr)[0] = ((device float *)src1_ptr)[0];
|
|
|
|
src1_ptr += ntg.x*nb10;
|
|
|
|
}
|
|
|
|
dst_ptr += ntg.x*nb0;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2023-06-08 09:08:23 +02:00
|
|
|
//============================================ k-quants ======================================================
|
|
|
|
|
k-quants : support for super-block size of 64 (#2001)
* k_quants: WIP super-blocks with 64 weights
* k_quants: WIP super-blocks with 64 weights
Q6_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q4_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q2_K scalar and AVX2 works. Q2_K is way too slow (it is actually slower
than the scalar implementation)
* k_quants: WIP super-blocks with 64 weights
Q3_K scalar and AVX2 works.
* k_quants: WIP super-blocks with 64 weights
Q5_K scalar and AVX2 works, and with that all
k_quants are done on AVX2 and scalar
* k_quants: WIP super-blocks with 64 weights
Q6_K working on CUDA. Cannot make it run quite as gast as
with super-blocks with 256 weigths: 8% slower on 4080,
20% slower on the 1660 (but there we fit 1 less layer on the
GPU because pf the larger model size), so some fraction of
these 20% is due to that,
* k_quants: WIP super-blocks with 64 weights
Q4_K working on CUDA. ~10% slower on GTX-1660,
16% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q2_K working on CUDA. ~3% slower on GTX-1660,
10% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q3_K working on CUDA.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on CUDA, and with this CUDA is done.
* k_quants: WIP super-blocks with 64 weights
Q6_K working on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Q4_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q2_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q3_K working on ARM_NEON, but quite a bit slower than 256 weights.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on ARM_NEON, but quite a bit slower than 256 weights.
With that, we have full support for ARM_NEON, although
performance is not quite there.
* k_quants: WIP super-blocks with 64 weights
Slightly more efficient Q3_K and Q5_K
* k_quants: WIP super-blocks with 64 weights
Another small improvement for Q3_K and Q5_K on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Yet another speedup for Q5_K on ARM_NEON.
We are now within 10% of the QK_K = 256 version.
* k_quants: WIP super-blocks with 64 weights
* We are able to pass preprocessor macros to the Metal
compiler
* Q6_K works and is actually slightly more efficient than
the QK_K = 256 version (25.2 ms vs 25.8 ms)
* k_quants: WIP super-blocks with 64 weights
Q4_K works on Metal and is actually slightly faster
than QK_K = 256 (21.95 ms vs 24.0 ms).
* k_quants: WIP super-blocks with 64 weights
Q2_K works on Metal and is very slightly faster
than QK_K = 256 (23.8 ms vs 24.2 ms).
* k_quants: WIP super-blocks with 64 weights
Q3_K works on Metal and is slightly faster
than QK_K = 256 (26.6 ms vs 28.3 ms).
* k_quants: WIP super-blocks with 64 weights
Q5_K works on Metal and is slightly faster
than QK_K = 256 (23.7 ms vs 26.3 ms).
* k_quants: call them _K, not _k, also on Metal
* k_quants: correctly define QK_K in llama.cpp
* Fixed bug in q4_K quantization added with the 64-block addition
* Simplify via lambda
* k_quants: swicth Q3_K to 4-bit scales when QK_K = 64
Otherwise there isn't much benefit from this
quantization type. There is some very slight loss
in accuracy, but we reduce size by ~7%.
E.g., for OpenLLaMA-3B, Q3_K_S perplexity is
8.6131 with 8-bit scales and 8.6352 with 4-bit,
while file size decreases from 1.53G to 1.44G.
* k_quants: switch Q4_K to 4-bit scales when QK_K = 64
Here the loss in accuracy is greater than for Q3_K,
but the Q4_K points still move further to the left on
the perplexity vs size curve.
* k_quants: forgot to add the Metal changes in last commit
* k_quants: change Q5_K to be type 0 when QK_K = 64
Still needs AVX2 implementation
* k_quants: AVX2 implementation for new 64-weight Q5_K
* k_quants: 10% faster ARM_NEON Q5_K dot product
* k_quants: fixed issue caused by merging with master
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-26 18:43:07 +02:00
|
|
|
#ifndef QK_K
|
2023-06-08 09:08:23 +02:00
|
|
|
#define QK_K 256
|
k-quants : support for super-block size of 64 (#2001)
* k_quants: WIP super-blocks with 64 weights
* k_quants: WIP super-blocks with 64 weights
Q6_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q4_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q2_K scalar and AVX2 works. Q2_K is way too slow (it is actually slower
than the scalar implementation)
* k_quants: WIP super-blocks with 64 weights
Q3_K scalar and AVX2 works.
* k_quants: WIP super-blocks with 64 weights
Q5_K scalar and AVX2 works, and with that all
k_quants are done on AVX2 and scalar
* k_quants: WIP super-blocks with 64 weights
Q6_K working on CUDA. Cannot make it run quite as gast as
with super-blocks with 256 weigths: 8% slower on 4080,
20% slower on the 1660 (but there we fit 1 less layer on the
GPU because pf the larger model size), so some fraction of
these 20% is due to that,
* k_quants: WIP super-blocks with 64 weights
Q4_K working on CUDA. ~10% slower on GTX-1660,
16% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q2_K working on CUDA. ~3% slower on GTX-1660,
10% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q3_K working on CUDA.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on CUDA, and with this CUDA is done.
* k_quants: WIP super-blocks with 64 weights
Q6_K working on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Q4_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q2_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q3_K working on ARM_NEON, but quite a bit slower than 256 weights.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on ARM_NEON, but quite a bit slower than 256 weights.
With that, we have full support for ARM_NEON, although
performance is not quite there.
* k_quants: WIP super-blocks with 64 weights
Slightly more efficient Q3_K and Q5_K
* k_quants: WIP super-blocks with 64 weights
Another small improvement for Q3_K and Q5_K on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Yet another speedup for Q5_K on ARM_NEON.
We are now within 10% of the QK_K = 256 version.
* k_quants: WIP super-blocks with 64 weights
* We are able to pass preprocessor macros to the Metal
compiler
* Q6_K works and is actually slightly more efficient than
the QK_K = 256 version (25.2 ms vs 25.8 ms)
* k_quants: WIP super-blocks with 64 weights
Q4_K works on Metal and is actually slightly faster
than QK_K = 256 (21.95 ms vs 24.0 ms).
* k_quants: WIP super-blocks with 64 weights
Q2_K works on Metal and is very slightly faster
than QK_K = 256 (23.8 ms vs 24.2 ms).
* k_quants: WIP super-blocks with 64 weights
Q3_K works on Metal and is slightly faster
than QK_K = 256 (26.6 ms vs 28.3 ms).
* k_quants: WIP super-blocks with 64 weights
Q5_K works on Metal and is slightly faster
than QK_K = 256 (23.7 ms vs 26.3 ms).
* k_quants: call them _K, not _k, also on Metal
* k_quants: correctly define QK_K in llama.cpp
* Fixed bug in q4_K quantization added with the 64-block addition
* Simplify via lambda
* k_quants: swicth Q3_K to 4-bit scales when QK_K = 64
Otherwise there isn't much benefit from this
quantization type. There is some very slight loss
in accuracy, but we reduce size by ~7%.
E.g., for OpenLLaMA-3B, Q3_K_S perplexity is
8.6131 with 8-bit scales and 8.6352 with 4-bit,
while file size decreases from 1.53G to 1.44G.
* k_quants: switch Q4_K to 4-bit scales when QK_K = 64
Here the loss in accuracy is greater than for Q3_K,
but the Q4_K points still move further to the left on
the perplexity vs size curve.
* k_quants: forgot to add the Metal changes in last commit
* k_quants: change Q5_K to be type 0 when QK_K = 64
Still needs AVX2 implementation
* k_quants: AVX2 implementation for new 64-weight Q5_K
* k_quants: 10% faster ARM_NEON Q5_K dot product
* k_quants: fixed issue caused by merging with master
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-26 18:43:07 +02:00
|
|
|
#else
|
|
|
|
static_assert(QK_K == 256 || QK_K == 64, "QK_K must be 256 or 64");
|
|
|
|
#endif
|
|
|
|
|
|
|
|
#if QK_K == 256
|
|
|
|
#define K_SCALE_SIZE 12
|
|
|
|
#else
|
|
|
|
#define K_SCALE_SIZE 4
|
|
|
|
#endif
|
2023-06-08 09:08:23 +02:00
|
|
|
|
2023-06-08 21:28:21 +02:00
|
|
|
typedef struct {
|
|
|
|
uint8_t scales[QK_K/16]; // scales and mins, quantized with 4 bits
|
|
|
|
uint8_t qs[QK_K/4]; // quants
|
|
|
|
half d; // super-block scale for quantized scales
|
|
|
|
half dmin; // super-block scale for quantized mins
|
k-quants : support for super-block size of 64 (#2001)
* k_quants: WIP super-blocks with 64 weights
* k_quants: WIP super-blocks with 64 weights
Q6_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q4_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q2_K scalar and AVX2 works. Q2_K is way too slow (it is actually slower
than the scalar implementation)
* k_quants: WIP super-blocks with 64 weights
Q3_K scalar and AVX2 works.
* k_quants: WIP super-blocks with 64 weights
Q5_K scalar and AVX2 works, and with that all
k_quants are done on AVX2 and scalar
* k_quants: WIP super-blocks with 64 weights
Q6_K working on CUDA. Cannot make it run quite as gast as
with super-blocks with 256 weigths: 8% slower on 4080,
20% slower on the 1660 (but there we fit 1 less layer on the
GPU because pf the larger model size), so some fraction of
these 20% is due to that,
* k_quants: WIP super-blocks with 64 weights
Q4_K working on CUDA. ~10% slower on GTX-1660,
16% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q2_K working on CUDA. ~3% slower on GTX-1660,
10% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q3_K working on CUDA.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on CUDA, and with this CUDA is done.
* k_quants: WIP super-blocks with 64 weights
Q6_K working on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Q4_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q2_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q3_K working on ARM_NEON, but quite a bit slower than 256 weights.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on ARM_NEON, but quite a bit slower than 256 weights.
With that, we have full support for ARM_NEON, although
performance is not quite there.
* k_quants: WIP super-blocks with 64 weights
Slightly more efficient Q3_K and Q5_K
* k_quants: WIP super-blocks with 64 weights
Another small improvement for Q3_K and Q5_K on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Yet another speedup for Q5_K on ARM_NEON.
We are now within 10% of the QK_K = 256 version.
* k_quants: WIP super-blocks with 64 weights
* We are able to pass preprocessor macros to the Metal
compiler
* Q6_K works and is actually slightly more efficient than
the QK_K = 256 version (25.2 ms vs 25.8 ms)
* k_quants: WIP super-blocks with 64 weights
Q4_K works on Metal and is actually slightly faster
than QK_K = 256 (21.95 ms vs 24.0 ms).
* k_quants: WIP super-blocks with 64 weights
Q2_K works on Metal and is very slightly faster
than QK_K = 256 (23.8 ms vs 24.2 ms).
* k_quants: WIP super-blocks with 64 weights
Q3_K works on Metal and is slightly faster
than QK_K = 256 (26.6 ms vs 28.3 ms).
* k_quants: WIP super-blocks with 64 weights
Q5_K works on Metal and is slightly faster
than QK_K = 256 (23.7 ms vs 26.3 ms).
* k_quants: call them _K, not _k, also on Metal
* k_quants: correctly define QK_K in llama.cpp
* Fixed bug in q4_K quantization added with the 64-block addition
* Simplify via lambda
* k_quants: swicth Q3_K to 4-bit scales when QK_K = 64
Otherwise there isn't much benefit from this
quantization type. There is some very slight loss
in accuracy, but we reduce size by ~7%.
E.g., for OpenLLaMA-3B, Q3_K_S perplexity is
8.6131 with 8-bit scales and 8.6352 with 4-bit,
while file size decreases from 1.53G to 1.44G.
* k_quants: switch Q4_K to 4-bit scales when QK_K = 64
Here the loss in accuracy is greater than for Q3_K,
but the Q4_K points still move further to the left on
the perplexity vs size curve.
* k_quants: forgot to add the Metal changes in last commit
* k_quants: change Q5_K to be type 0 when QK_K = 64
Still needs AVX2 implementation
* k_quants: AVX2 implementation for new 64-weight Q5_K
* k_quants: 10% faster ARM_NEON Q5_K dot product
* k_quants: fixed issue caused by merging with master
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-26 18:43:07 +02:00
|
|
|
} block_q2_K;
|
Metal implementation for all k_quants (#1807)
* metal : improve q4_K
28.3 -> 26.0 ms/token by avoiding a branch in the
calculation of the scales.
* metal : small improvement for Q4_K
* metal : still optimizing Q4_K
This commit pushes it down to 25.3 ms / token.
The crazy idea of using 6 bits for the scales is really costly on
Metal: if I remove the bit fiddling necessary to make the block
scales, time goes almost to the Q4_0 23 ms/token.
Before pushing the k-quants upstream I had a Q4_K variant that
had used 8-bit scales. It wasn't more accurate, used 0.125 bits more per weight,
was running slightly slower on the CPU (due to the larger model size
and being memory bound there), and the difference was entirely
negligible under CUDA. So, I decided to publish the version with 6-bit
scales. Perhaps I should re-consider and change to 8-bit scales?
* metal : some more optimizations
Q2_K: 25.4 ms/token
Q6_K: 27.3 ms/token
Q4_0: 22.8 ms/token
Q4_1: 23.1 ms/token
* metal : Q3_K support
Something is not quite right yet.
* metal : Q5_K support
Initial version achieves 31.2 ms/token, 210 GB/s
* metal : still not able to figure out why q3_K does not work
* Minor
* metal : yet another failed attempt to make q3_K work
* metal : optimize Q5_K
31.2 ms -> 27.8 ms.
250 GB/s.
* metal : q3_K still not working
Adding a heavily commented q3_K metal kernel to explain
my obviously faulty logic. Perhaps someone could spot the issue?
* metal : q3_K finally working
Not optimized at all.
What was the issue? The scales are not 4-bytes aligned,
and I was accessing them with a uint32_t pointer.
When I tried that on CUDA, I got an error (illegal memory access)
and added a memcpy to a local array of 3 uint32_t's.
But on Metal it told me there is no memcpy, so I tried
accessing directly. There is no error, just garbage results.
At some point I did try accessing the scales with an uint16_t
pointer (the scales are for sure 2-byte aligned), but was
still getting garbage. I guess, there must have been another bug.
No access to scales is via a uint16_t pointer and, after starting
from scratch from the C dequantize function, it finally works.
* metal : Q3_K 1st optimization pass
* metal : Q3_K second optimization pass - 29.6 ms/token
* metal : Q3_K cleanup
* metal : fixed accidentally broken Q2_K
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-12 21:39:21 +02:00
|
|
|
// 84 bytes / block
|
|
|
|
|
|
|
|
typedef struct {
|
|
|
|
uint8_t hmask[QK_K/8]; // quants - high bit
|
|
|
|
uint8_t qs[QK_K/4]; // quants - low 2 bits
|
k-quants : support for super-block size of 64 (#2001)
* k_quants: WIP super-blocks with 64 weights
* k_quants: WIP super-blocks with 64 weights
Q6_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q4_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q2_K scalar and AVX2 works. Q2_K is way too slow (it is actually slower
than the scalar implementation)
* k_quants: WIP super-blocks with 64 weights
Q3_K scalar and AVX2 works.
* k_quants: WIP super-blocks with 64 weights
Q5_K scalar and AVX2 works, and with that all
k_quants are done on AVX2 and scalar
* k_quants: WIP super-blocks with 64 weights
Q6_K working on CUDA. Cannot make it run quite as gast as
with super-blocks with 256 weigths: 8% slower on 4080,
20% slower on the 1660 (but there we fit 1 less layer on the
GPU because pf the larger model size), so some fraction of
these 20% is due to that,
* k_quants: WIP super-blocks with 64 weights
Q4_K working on CUDA. ~10% slower on GTX-1660,
16% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q2_K working on CUDA. ~3% slower on GTX-1660,
10% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q3_K working on CUDA.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on CUDA, and with this CUDA is done.
* k_quants: WIP super-blocks with 64 weights
Q6_K working on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Q4_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q2_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q3_K working on ARM_NEON, but quite a bit slower than 256 weights.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on ARM_NEON, but quite a bit slower than 256 weights.
With that, we have full support for ARM_NEON, although
performance is not quite there.
* k_quants: WIP super-blocks with 64 weights
Slightly more efficient Q3_K and Q5_K
* k_quants: WIP super-blocks with 64 weights
Another small improvement for Q3_K and Q5_K on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Yet another speedup for Q5_K on ARM_NEON.
We are now within 10% of the QK_K = 256 version.
* k_quants: WIP super-blocks with 64 weights
* We are able to pass preprocessor macros to the Metal
compiler
* Q6_K works and is actually slightly more efficient than
the QK_K = 256 version (25.2 ms vs 25.8 ms)
* k_quants: WIP super-blocks with 64 weights
Q4_K works on Metal and is actually slightly faster
than QK_K = 256 (21.95 ms vs 24.0 ms).
* k_quants: WIP super-blocks with 64 weights
Q2_K works on Metal and is very slightly faster
than QK_K = 256 (23.8 ms vs 24.2 ms).
* k_quants: WIP super-blocks with 64 weights
Q3_K works on Metal and is slightly faster
than QK_K = 256 (26.6 ms vs 28.3 ms).
* k_quants: WIP super-blocks with 64 weights
Q5_K works on Metal and is slightly faster
than QK_K = 256 (23.7 ms vs 26.3 ms).
* k_quants: call them _K, not _k, also on Metal
* k_quants: correctly define QK_K in llama.cpp
* Fixed bug in q4_K quantization added with the 64-block addition
* Simplify via lambda
* k_quants: swicth Q3_K to 4-bit scales when QK_K = 64
Otherwise there isn't much benefit from this
quantization type. There is some very slight loss
in accuracy, but we reduce size by ~7%.
E.g., for OpenLLaMA-3B, Q3_K_S perplexity is
8.6131 with 8-bit scales and 8.6352 with 4-bit,
while file size decreases from 1.53G to 1.44G.
* k_quants: switch Q4_K to 4-bit scales when QK_K = 64
Here the loss in accuracy is greater than for Q3_K,
but the Q4_K points still move further to the left on
the perplexity vs size curve.
* k_quants: forgot to add the Metal changes in last commit
* k_quants: change Q5_K to be type 0 when QK_K = 64
Still needs AVX2 implementation
* k_quants: AVX2 implementation for new 64-weight Q5_K
* k_quants: 10% faster ARM_NEON Q5_K dot product
* k_quants: fixed issue caused by merging with master
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-26 18:43:07 +02:00
|
|
|
#if QK_K == 64
|
|
|
|
uint8_t scales[2];
|
|
|
|
#else
|
|
|
|
uint8_t scales[K_SCALE_SIZE]; // scales, quantized with 6 bits
|
|
|
|
#endif
|
|
|
|
half d; // super-block scale
|
|
|
|
} block_q3_K;
|
|
|
|
|
|
|
|
#if QK_K == 64
|
|
|
|
typedef struct {
|
|
|
|
half d[2]; // super-block scales/mins
|
|
|
|
uint8_t scales[2];
|
|
|
|
uint8_t qs[QK_K/2]; // 4-bit quants
|
|
|
|
} block_q4_K;
|
|
|
|
#else
|
2023-06-08 09:08:23 +02:00
|
|
|
typedef struct {
|
|
|
|
half d; // super-block scale for quantized scales
|
|
|
|
half dmin; // super-block scale for quantized mins
|
k-quants : support for super-block size of 64 (#2001)
* k_quants: WIP super-blocks with 64 weights
* k_quants: WIP super-blocks with 64 weights
Q6_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q4_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q2_K scalar and AVX2 works. Q2_K is way too slow (it is actually slower
than the scalar implementation)
* k_quants: WIP super-blocks with 64 weights
Q3_K scalar and AVX2 works.
* k_quants: WIP super-blocks with 64 weights
Q5_K scalar and AVX2 works, and with that all
k_quants are done on AVX2 and scalar
* k_quants: WIP super-blocks with 64 weights
Q6_K working on CUDA. Cannot make it run quite as gast as
with super-blocks with 256 weigths: 8% slower on 4080,
20% slower on the 1660 (but there we fit 1 less layer on the
GPU because pf the larger model size), so some fraction of
these 20% is due to that,
* k_quants: WIP super-blocks with 64 weights
Q4_K working on CUDA. ~10% slower on GTX-1660,
16% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q2_K working on CUDA. ~3% slower on GTX-1660,
10% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q3_K working on CUDA.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on CUDA, and with this CUDA is done.
* k_quants: WIP super-blocks with 64 weights
Q6_K working on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Q4_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q2_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q3_K working on ARM_NEON, but quite a bit slower than 256 weights.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on ARM_NEON, but quite a bit slower than 256 weights.
With that, we have full support for ARM_NEON, although
performance is not quite there.
* k_quants: WIP super-blocks with 64 weights
Slightly more efficient Q3_K and Q5_K
* k_quants: WIP super-blocks with 64 weights
Another small improvement for Q3_K and Q5_K on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Yet another speedup for Q5_K on ARM_NEON.
We are now within 10% of the QK_K = 256 version.
* k_quants: WIP super-blocks with 64 weights
* We are able to pass preprocessor macros to the Metal
compiler
* Q6_K works and is actually slightly more efficient than
the QK_K = 256 version (25.2 ms vs 25.8 ms)
* k_quants: WIP super-blocks with 64 weights
Q4_K works on Metal and is actually slightly faster
than QK_K = 256 (21.95 ms vs 24.0 ms).
* k_quants: WIP super-blocks with 64 weights
Q2_K works on Metal and is very slightly faster
than QK_K = 256 (23.8 ms vs 24.2 ms).
* k_quants: WIP super-blocks with 64 weights
Q3_K works on Metal and is slightly faster
than QK_K = 256 (26.6 ms vs 28.3 ms).
* k_quants: WIP super-blocks with 64 weights
Q5_K works on Metal and is slightly faster
than QK_K = 256 (23.7 ms vs 26.3 ms).
* k_quants: call them _K, not _k, also on Metal
* k_quants: correctly define QK_K in llama.cpp
* Fixed bug in q4_K quantization added with the 64-block addition
* Simplify via lambda
* k_quants: swicth Q3_K to 4-bit scales when QK_K = 64
Otherwise there isn't much benefit from this
quantization type. There is some very slight loss
in accuracy, but we reduce size by ~7%.
E.g., for OpenLLaMA-3B, Q3_K_S perplexity is
8.6131 with 8-bit scales and 8.6352 with 4-bit,
while file size decreases from 1.53G to 1.44G.
* k_quants: switch Q4_K to 4-bit scales when QK_K = 64
Here the loss in accuracy is greater than for Q3_K,
but the Q4_K points still move further to the left on
the perplexity vs size curve.
* k_quants: forgot to add the Metal changes in last commit
* k_quants: change Q5_K to be type 0 when QK_K = 64
Still needs AVX2 implementation
* k_quants: AVX2 implementation for new 64-weight Q5_K
* k_quants: 10% faster ARM_NEON Q5_K dot product
* k_quants: fixed issue caused by merging with master
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-26 18:43:07 +02:00
|
|
|
uint8_t scales[K_SCALE_SIZE]; // scales and mins, quantized with 6 bits
|
2023-06-08 09:08:23 +02:00
|
|
|
uint8_t qs[QK_K/2]; // 4--bit quants
|
k-quants : support for super-block size of 64 (#2001)
* k_quants: WIP super-blocks with 64 weights
* k_quants: WIP super-blocks with 64 weights
Q6_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q4_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q2_K scalar and AVX2 works. Q2_K is way too slow (it is actually slower
than the scalar implementation)
* k_quants: WIP super-blocks with 64 weights
Q3_K scalar and AVX2 works.
* k_quants: WIP super-blocks with 64 weights
Q5_K scalar and AVX2 works, and with that all
k_quants are done on AVX2 and scalar
* k_quants: WIP super-blocks with 64 weights
Q6_K working on CUDA. Cannot make it run quite as gast as
with super-blocks with 256 weigths: 8% slower on 4080,
20% slower on the 1660 (but there we fit 1 less layer on the
GPU because pf the larger model size), so some fraction of
these 20% is due to that,
* k_quants: WIP super-blocks with 64 weights
Q4_K working on CUDA. ~10% slower on GTX-1660,
16% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q2_K working on CUDA. ~3% slower on GTX-1660,
10% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q3_K working on CUDA.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on CUDA, and with this CUDA is done.
* k_quants: WIP super-blocks with 64 weights
Q6_K working on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Q4_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q2_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q3_K working on ARM_NEON, but quite a bit slower than 256 weights.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on ARM_NEON, but quite a bit slower than 256 weights.
With that, we have full support for ARM_NEON, although
performance is not quite there.
* k_quants: WIP super-blocks with 64 weights
Slightly more efficient Q3_K and Q5_K
* k_quants: WIP super-blocks with 64 weights
Another small improvement for Q3_K and Q5_K on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Yet another speedup for Q5_K on ARM_NEON.
We are now within 10% of the QK_K = 256 version.
* k_quants: WIP super-blocks with 64 weights
* We are able to pass preprocessor macros to the Metal
compiler
* Q6_K works and is actually slightly more efficient than
the QK_K = 256 version (25.2 ms vs 25.8 ms)
* k_quants: WIP super-blocks with 64 weights
Q4_K works on Metal and is actually slightly faster
than QK_K = 256 (21.95 ms vs 24.0 ms).
* k_quants: WIP super-blocks with 64 weights
Q2_K works on Metal and is very slightly faster
than QK_K = 256 (23.8 ms vs 24.2 ms).
* k_quants: WIP super-blocks with 64 weights
Q3_K works on Metal and is slightly faster
than QK_K = 256 (26.6 ms vs 28.3 ms).
* k_quants: WIP super-blocks with 64 weights
Q5_K works on Metal and is slightly faster
than QK_K = 256 (23.7 ms vs 26.3 ms).
* k_quants: call them _K, not _k, also on Metal
* k_quants: correctly define QK_K in llama.cpp
* Fixed bug in q4_K quantization added with the 64-block addition
* Simplify via lambda
* k_quants: swicth Q3_K to 4-bit scales when QK_K = 64
Otherwise there isn't much benefit from this
quantization type. There is some very slight loss
in accuracy, but we reduce size by ~7%.
E.g., for OpenLLaMA-3B, Q3_K_S perplexity is
8.6131 with 8-bit scales and 8.6352 with 4-bit,
while file size decreases from 1.53G to 1.44G.
* k_quants: switch Q4_K to 4-bit scales when QK_K = 64
Here the loss in accuracy is greater than for Q3_K,
but the Q4_K points still move further to the left on
the perplexity vs size curve.
* k_quants: forgot to add the Metal changes in last commit
* k_quants: change Q5_K to be type 0 when QK_K = 64
Still needs AVX2 implementation
* k_quants: AVX2 implementation for new 64-weight Q5_K
* k_quants: 10% faster ARM_NEON Q5_K dot product
* k_quants: fixed issue caused by merging with master
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-26 18:43:07 +02:00
|
|
|
} block_q4_K;
|
|
|
|
#endif
|
Metal implementation for all k_quants (#1807)
* metal : improve q4_K
28.3 -> 26.0 ms/token by avoiding a branch in the
calculation of the scales.
* metal : small improvement for Q4_K
* metal : still optimizing Q4_K
This commit pushes it down to 25.3 ms / token.
The crazy idea of using 6 bits for the scales is really costly on
Metal: if I remove the bit fiddling necessary to make the block
scales, time goes almost to the Q4_0 23 ms/token.
Before pushing the k-quants upstream I had a Q4_K variant that
had used 8-bit scales. It wasn't more accurate, used 0.125 bits more per weight,
was running slightly slower on the CPU (due to the larger model size
and being memory bound there), and the difference was entirely
negligible under CUDA. So, I decided to publish the version with 6-bit
scales. Perhaps I should re-consider and change to 8-bit scales?
* metal : some more optimizations
Q2_K: 25.4 ms/token
Q6_K: 27.3 ms/token
Q4_0: 22.8 ms/token
Q4_1: 23.1 ms/token
* metal : Q3_K support
Something is not quite right yet.
* metal : Q5_K support
Initial version achieves 31.2 ms/token, 210 GB/s
* metal : still not able to figure out why q3_K does not work
* Minor
* metal : yet another failed attempt to make q3_K work
* metal : optimize Q5_K
31.2 ms -> 27.8 ms.
250 GB/s.
* metal : q3_K still not working
Adding a heavily commented q3_K metal kernel to explain
my obviously faulty logic. Perhaps someone could spot the issue?
* metal : q3_K finally working
Not optimized at all.
What was the issue? The scales are not 4-bytes aligned,
and I was accessing them with a uint32_t pointer.
When I tried that on CUDA, I got an error (illegal memory access)
and added a memcpy to a local array of 3 uint32_t's.
But on Metal it told me there is no memcpy, so I tried
accessing directly. There is no error, just garbage results.
At some point I did try accessing the scales with an uint16_t
pointer (the scales are for sure 2-byte aligned), but was
still getting garbage. I guess, there must have been another bug.
No access to scales is via a uint16_t pointer and, after starting
from scratch from the C dequantize function, it finally works.
* metal : Q3_K 1st optimization pass
* metal : Q3_K second optimization pass - 29.6 ms/token
* metal : Q3_K cleanup
* metal : fixed accidentally broken Q2_K
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-12 21:39:21 +02:00
|
|
|
|
k-quants : support for super-block size of 64 (#2001)
* k_quants: WIP super-blocks with 64 weights
* k_quants: WIP super-blocks with 64 weights
Q6_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q4_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q2_K scalar and AVX2 works. Q2_K is way too slow (it is actually slower
than the scalar implementation)
* k_quants: WIP super-blocks with 64 weights
Q3_K scalar and AVX2 works.
* k_quants: WIP super-blocks with 64 weights
Q5_K scalar and AVX2 works, and with that all
k_quants are done on AVX2 and scalar
* k_quants: WIP super-blocks with 64 weights
Q6_K working on CUDA. Cannot make it run quite as gast as
with super-blocks with 256 weigths: 8% slower on 4080,
20% slower on the 1660 (but there we fit 1 less layer on the
GPU because pf the larger model size), so some fraction of
these 20% is due to that,
* k_quants: WIP super-blocks with 64 weights
Q4_K working on CUDA. ~10% slower on GTX-1660,
16% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q2_K working on CUDA. ~3% slower on GTX-1660,
10% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q3_K working on CUDA.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on CUDA, and with this CUDA is done.
* k_quants: WIP super-blocks with 64 weights
Q6_K working on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Q4_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q2_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q3_K working on ARM_NEON, but quite a bit slower than 256 weights.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on ARM_NEON, but quite a bit slower than 256 weights.
With that, we have full support for ARM_NEON, although
performance is not quite there.
* k_quants: WIP super-blocks with 64 weights
Slightly more efficient Q3_K and Q5_K
* k_quants: WIP super-blocks with 64 weights
Another small improvement for Q3_K and Q5_K on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Yet another speedup for Q5_K on ARM_NEON.
We are now within 10% of the QK_K = 256 version.
* k_quants: WIP super-blocks with 64 weights
* We are able to pass preprocessor macros to the Metal
compiler
* Q6_K works and is actually slightly more efficient than
the QK_K = 256 version (25.2 ms vs 25.8 ms)
* k_quants: WIP super-blocks with 64 weights
Q4_K works on Metal and is actually slightly faster
than QK_K = 256 (21.95 ms vs 24.0 ms).
* k_quants: WIP super-blocks with 64 weights
Q2_K works on Metal and is very slightly faster
than QK_K = 256 (23.8 ms vs 24.2 ms).
* k_quants: WIP super-blocks with 64 weights
Q3_K works on Metal and is slightly faster
than QK_K = 256 (26.6 ms vs 28.3 ms).
* k_quants: WIP super-blocks with 64 weights
Q5_K works on Metal and is slightly faster
than QK_K = 256 (23.7 ms vs 26.3 ms).
* k_quants: call them _K, not _k, also on Metal
* k_quants: correctly define QK_K in llama.cpp
* Fixed bug in q4_K quantization added with the 64-block addition
* Simplify via lambda
* k_quants: swicth Q3_K to 4-bit scales when QK_K = 64
Otherwise there isn't much benefit from this
quantization type. There is some very slight loss
in accuracy, but we reduce size by ~7%.
E.g., for OpenLLaMA-3B, Q3_K_S perplexity is
8.6131 with 8-bit scales and 8.6352 with 4-bit,
while file size decreases from 1.53G to 1.44G.
* k_quants: switch Q4_K to 4-bit scales when QK_K = 64
Here the loss in accuracy is greater than for Q3_K,
but the Q4_K points still move further to the left on
the perplexity vs size curve.
* k_quants: forgot to add the Metal changes in last commit
* k_quants: change Q5_K to be type 0 when QK_K = 64
Still needs AVX2 implementation
* k_quants: AVX2 implementation for new 64-weight Q5_K
* k_quants: 10% faster ARM_NEON Q5_K dot product
* k_quants: fixed issue caused by merging with master
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-26 18:43:07 +02:00
|
|
|
#if QK_K == 64
|
|
|
|
typedef struct {
|
|
|
|
half d; // super-block scales/mins
|
|
|
|
int8_t scales[QK_K/16]; // 8-bit block scales
|
|
|
|
uint8_t qh[QK_K/8]; // quants, high bit
|
|
|
|
uint8_t qs[QK_K/2]; // quants, low 4 bits
|
|
|
|
} block_q5_K;
|
|
|
|
#else
|
Metal implementation for all k_quants (#1807)
* metal : improve q4_K
28.3 -> 26.0 ms/token by avoiding a branch in the
calculation of the scales.
* metal : small improvement for Q4_K
* metal : still optimizing Q4_K
This commit pushes it down to 25.3 ms / token.
The crazy idea of using 6 bits for the scales is really costly on
Metal: if I remove the bit fiddling necessary to make the block
scales, time goes almost to the Q4_0 23 ms/token.
Before pushing the k-quants upstream I had a Q4_K variant that
had used 8-bit scales. It wasn't more accurate, used 0.125 bits more per weight,
was running slightly slower on the CPU (due to the larger model size
and being memory bound there), and the difference was entirely
negligible under CUDA. So, I decided to publish the version with 6-bit
scales. Perhaps I should re-consider and change to 8-bit scales?
* metal : some more optimizations
Q2_K: 25.4 ms/token
Q6_K: 27.3 ms/token
Q4_0: 22.8 ms/token
Q4_1: 23.1 ms/token
* metal : Q3_K support
Something is not quite right yet.
* metal : Q5_K support
Initial version achieves 31.2 ms/token, 210 GB/s
* metal : still not able to figure out why q3_K does not work
* Minor
* metal : yet another failed attempt to make q3_K work
* metal : optimize Q5_K
31.2 ms -> 27.8 ms.
250 GB/s.
* metal : q3_K still not working
Adding a heavily commented q3_K metal kernel to explain
my obviously faulty logic. Perhaps someone could spot the issue?
* metal : q3_K finally working
Not optimized at all.
What was the issue? The scales are not 4-bytes aligned,
and I was accessing them with a uint32_t pointer.
When I tried that on CUDA, I got an error (illegal memory access)
and added a memcpy to a local array of 3 uint32_t's.
But on Metal it told me there is no memcpy, so I tried
accessing directly. There is no error, just garbage results.
At some point I did try accessing the scales with an uint16_t
pointer (the scales are for sure 2-byte aligned), but was
still getting garbage. I guess, there must have been another bug.
No access to scales is via a uint16_t pointer and, after starting
from scratch from the C dequantize function, it finally works.
* metal : Q3_K 1st optimization pass
* metal : Q3_K second optimization pass - 29.6 ms/token
* metal : Q3_K cleanup
* metal : fixed accidentally broken Q2_K
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-12 21:39:21 +02:00
|
|
|
typedef struct {
|
|
|
|
half d; // super-block scale for quantized scales
|
|
|
|
half dmin; // super-block scale for quantized mins
|
|
|
|
uint8_t scales[3*QK_K/64]; // scales and mins, quantized with 6 bits
|
|
|
|
uint8_t qh[QK_K/8]; // quants, high bit
|
|
|
|
uint8_t qs[QK_K/2]; // quants, low 4 bits
|
k-quants : support for super-block size of 64 (#2001)
* k_quants: WIP super-blocks with 64 weights
* k_quants: WIP super-blocks with 64 weights
Q6_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q4_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q2_K scalar and AVX2 works. Q2_K is way too slow (it is actually slower
than the scalar implementation)
* k_quants: WIP super-blocks with 64 weights
Q3_K scalar and AVX2 works.
* k_quants: WIP super-blocks with 64 weights
Q5_K scalar and AVX2 works, and with that all
k_quants are done on AVX2 and scalar
* k_quants: WIP super-blocks with 64 weights
Q6_K working on CUDA. Cannot make it run quite as gast as
with super-blocks with 256 weigths: 8% slower on 4080,
20% slower on the 1660 (but there we fit 1 less layer on the
GPU because pf the larger model size), so some fraction of
these 20% is due to that,
* k_quants: WIP super-blocks with 64 weights
Q4_K working on CUDA. ~10% slower on GTX-1660,
16% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q2_K working on CUDA. ~3% slower on GTX-1660,
10% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q3_K working on CUDA.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on CUDA, and with this CUDA is done.
* k_quants: WIP super-blocks with 64 weights
Q6_K working on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Q4_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q2_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q3_K working on ARM_NEON, but quite a bit slower than 256 weights.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on ARM_NEON, but quite a bit slower than 256 weights.
With that, we have full support for ARM_NEON, although
performance is not quite there.
* k_quants: WIP super-blocks with 64 weights
Slightly more efficient Q3_K and Q5_K
* k_quants: WIP super-blocks with 64 weights
Another small improvement for Q3_K and Q5_K on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Yet another speedup for Q5_K on ARM_NEON.
We are now within 10% of the QK_K = 256 version.
* k_quants: WIP super-blocks with 64 weights
* We are able to pass preprocessor macros to the Metal
compiler
* Q6_K works and is actually slightly more efficient than
the QK_K = 256 version (25.2 ms vs 25.8 ms)
* k_quants: WIP super-blocks with 64 weights
Q4_K works on Metal and is actually slightly faster
than QK_K = 256 (21.95 ms vs 24.0 ms).
* k_quants: WIP super-blocks with 64 weights
Q2_K works on Metal and is very slightly faster
than QK_K = 256 (23.8 ms vs 24.2 ms).
* k_quants: WIP super-blocks with 64 weights
Q3_K works on Metal and is slightly faster
than QK_K = 256 (26.6 ms vs 28.3 ms).
* k_quants: WIP super-blocks with 64 weights
Q5_K works on Metal and is slightly faster
than QK_K = 256 (23.7 ms vs 26.3 ms).
* k_quants: call them _K, not _k, also on Metal
* k_quants: correctly define QK_K in llama.cpp
* Fixed bug in q4_K quantization added with the 64-block addition
* Simplify via lambda
* k_quants: swicth Q3_K to 4-bit scales when QK_K = 64
Otherwise there isn't much benefit from this
quantization type. There is some very slight loss
in accuracy, but we reduce size by ~7%.
E.g., for OpenLLaMA-3B, Q3_K_S perplexity is
8.6131 with 8-bit scales and 8.6352 with 4-bit,
while file size decreases from 1.53G to 1.44G.
* k_quants: switch Q4_K to 4-bit scales when QK_K = 64
Here the loss in accuracy is greater than for Q3_K,
but the Q4_K points still move further to the left on
the perplexity vs size curve.
* k_quants: forgot to add the Metal changes in last commit
* k_quants: change Q5_K to be type 0 when QK_K = 64
Still needs AVX2 implementation
* k_quants: AVX2 implementation for new 64-weight Q5_K
* k_quants: 10% faster ARM_NEON Q5_K dot product
* k_quants: fixed issue caused by merging with master
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-26 18:43:07 +02:00
|
|
|
} block_q5_K;
|
Metal implementation for all k_quants (#1807)
* metal : improve q4_K
28.3 -> 26.0 ms/token by avoiding a branch in the
calculation of the scales.
* metal : small improvement for Q4_K
* metal : still optimizing Q4_K
This commit pushes it down to 25.3 ms / token.
The crazy idea of using 6 bits for the scales is really costly on
Metal: if I remove the bit fiddling necessary to make the block
scales, time goes almost to the Q4_0 23 ms/token.
Before pushing the k-quants upstream I had a Q4_K variant that
had used 8-bit scales. It wasn't more accurate, used 0.125 bits more per weight,
was running slightly slower on the CPU (due to the larger model size
and being memory bound there), and the difference was entirely
negligible under CUDA. So, I decided to publish the version with 6-bit
scales. Perhaps I should re-consider and change to 8-bit scales?
* metal : some more optimizations
Q2_K: 25.4 ms/token
Q6_K: 27.3 ms/token
Q4_0: 22.8 ms/token
Q4_1: 23.1 ms/token
* metal : Q3_K support
Something is not quite right yet.
* metal : Q5_K support
Initial version achieves 31.2 ms/token, 210 GB/s
* metal : still not able to figure out why q3_K does not work
* Minor
* metal : yet another failed attempt to make q3_K work
* metal : optimize Q5_K
31.2 ms -> 27.8 ms.
250 GB/s.
* metal : q3_K still not working
Adding a heavily commented q3_K metal kernel to explain
my obviously faulty logic. Perhaps someone could spot the issue?
* metal : q3_K finally working
Not optimized at all.
What was the issue? The scales are not 4-bytes aligned,
and I was accessing them with a uint32_t pointer.
When I tried that on CUDA, I got an error (illegal memory access)
and added a memcpy to a local array of 3 uint32_t's.
But on Metal it told me there is no memcpy, so I tried
accessing directly. There is no error, just garbage results.
At some point I did try accessing the scales with an uint16_t
pointer (the scales are for sure 2-byte aligned), but was
still getting garbage. I guess, there must have been another bug.
No access to scales is via a uint16_t pointer and, after starting
from scratch from the C dequantize function, it finally works.
* metal : Q3_K 1st optimization pass
* metal : Q3_K second optimization pass - 29.6 ms/token
* metal : Q3_K cleanup
* metal : fixed accidentally broken Q2_K
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-12 21:39:21 +02:00
|
|
|
// 176 bytes / block
|
k-quants : support for super-block size of 64 (#2001)
* k_quants: WIP super-blocks with 64 weights
* k_quants: WIP super-blocks with 64 weights
Q6_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q4_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q2_K scalar and AVX2 works. Q2_K is way too slow (it is actually slower
than the scalar implementation)
* k_quants: WIP super-blocks with 64 weights
Q3_K scalar and AVX2 works.
* k_quants: WIP super-blocks with 64 weights
Q5_K scalar and AVX2 works, and with that all
k_quants are done on AVX2 and scalar
* k_quants: WIP super-blocks with 64 weights
Q6_K working on CUDA. Cannot make it run quite as gast as
with super-blocks with 256 weigths: 8% slower on 4080,
20% slower on the 1660 (but there we fit 1 less layer on the
GPU because pf the larger model size), so some fraction of
these 20% is due to that,
* k_quants: WIP super-blocks with 64 weights
Q4_K working on CUDA. ~10% slower on GTX-1660,
16% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q2_K working on CUDA. ~3% slower on GTX-1660,
10% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q3_K working on CUDA.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on CUDA, and with this CUDA is done.
* k_quants: WIP super-blocks with 64 weights
Q6_K working on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Q4_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q2_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q3_K working on ARM_NEON, but quite a bit slower than 256 weights.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on ARM_NEON, but quite a bit slower than 256 weights.
With that, we have full support for ARM_NEON, although
performance is not quite there.
* k_quants: WIP super-blocks with 64 weights
Slightly more efficient Q3_K and Q5_K
* k_quants: WIP super-blocks with 64 weights
Another small improvement for Q3_K and Q5_K on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Yet another speedup for Q5_K on ARM_NEON.
We are now within 10% of the QK_K = 256 version.
* k_quants: WIP super-blocks with 64 weights
* We are able to pass preprocessor macros to the Metal
compiler
* Q6_K works and is actually slightly more efficient than
the QK_K = 256 version (25.2 ms vs 25.8 ms)
* k_quants: WIP super-blocks with 64 weights
Q4_K works on Metal and is actually slightly faster
than QK_K = 256 (21.95 ms vs 24.0 ms).
* k_quants: WIP super-blocks with 64 weights
Q2_K works on Metal and is very slightly faster
than QK_K = 256 (23.8 ms vs 24.2 ms).
* k_quants: WIP super-blocks with 64 weights
Q3_K works on Metal and is slightly faster
than QK_K = 256 (26.6 ms vs 28.3 ms).
* k_quants: WIP super-blocks with 64 weights
Q5_K works on Metal and is slightly faster
than QK_K = 256 (23.7 ms vs 26.3 ms).
* k_quants: call them _K, not _k, also on Metal
* k_quants: correctly define QK_K in llama.cpp
* Fixed bug in q4_K quantization added with the 64-block addition
* Simplify via lambda
* k_quants: swicth Q3_K to 4-bit scales when QK_K = 64
Otherwise there isn't much benefit from this
quantization type. There is some very slight loss
in accuracy, but we reduce size by ~7%.
E.g., for OpenLLaMA-3B, Q3_K_S perplexity is
8.6131 with 8-bit scales and 8.6352 with 4-bit,
while file size decreases from 1.53G to 1.44G.
* k_quants: switch Q4_K to 4-bit scales when QK_K = 64
Here the loss in accuracy is greater than for Q3_K,
but the Q4_K points still move further to the left on
the perplexity vs size curve.
* k_quants: forgot to add the Metal changes in last commit
* k_quants: change Q5_K to be type 0 when QK_K = 64
Still needs AVX2 implementation
* k_quants: AVX2 implementation for new 64-weight Q5_K
* k_quants: 10% faster ARM_NEON Q5_K dot product
* k_quants: fixed issue caused by merging with master
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-26 18:43:07 +02:00
|
|
|
#endif
|
2023-06-08 09:08:23 +02:00
|
|
|
|
2023-06-08 18:46:22 +02:00
|
|
|
typedef struct {
|
|
|
|
uint8_t ql[QK_K/2]; // quants, lower 4 bits
|
|
|
|
uint8_t qh[QK_K/4]; // quants, upper 2 bits
|
|
|
|
int8_t scales[QK_K/16]; // scales, quantized with 8 bits
|
|
|
|
half d; // super-block scale
|
k-quants : support for super-block size of 64 (#2001)
* k_quants: WIP super-blocks with 64 weights
* k_quants: WIP super-blocks with 64 weights
Q6_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q4_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q2_K scalar and AVX2 works. Q2_K is way too slow (it is actually slower
than the scalar implementation)
* k_quants: WIP super-blocks with 64 weights
Q3_K scalar and AVX2 works.
* k_quants: WIP super-blocks with 64 weights
Q5_K scalar and AVX2 works, and with that all
k_quants are done on AVX2 and scalar
* k_quants: WIP super-blocks with 64 weights
Q6_K working on CUDA. Cannot make it run quite as gast as
with super-blocks with 256 weigths: 8% slower on 4080,
20% slower on the 1660 (but there we fit 1 less layer on the
GPU because pf the larger model size), so some fraction of
these 20% is due to that,
* k_quants: WIP super-blocks with 64 weights
Q4_K working on CUDA. ~10% slower on GTX-1660,
16% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q2_K working on CUDA. ~3% slower on GTX-1660,
10% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q3_K working on CUDA.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on CUDA, and with this CUDA is done.
* k_quants: WIP super-blocks with 64 weights
Q6_K working on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Q4_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q2_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q3_K working on ARM_NEON, but quite a bit slower than 256 weights.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on ARM_NEON, but quite a bit slower than 256 weights.
With that, we have full support for ARM_NEON, although
performance is not quite there.
* k_quants: WIP super-blocks with 64 weights
Slightly more efficient Q3_K and Q5_K
* k_quants: WIP super-blocks with 64 weights
Another small improvement for Q3_K and Q5_K on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Yet another speedup for Q5_K on ARM_NEON.
We are now within 10% of the QK_K = 256 version.
* k_quants: WIP super-blocks with 64 weights
* We are able to pass preprocessor macros to the Metal
compiler
* Q6_K works and is actually slightly more efficient than
the QK_K = 256 version (25.2 ms vs 25.8 ms)
* k_quants: WIP super-blocks with 64 weights
Q4_K works on Metal and is actually slightly faster
than QK_K = 256 (21.95 ms vs 24.0 ms).
* k_quants: WIP super-blocks with 64 weights
Q2_K works on Metal and is very slightly faster
than QK_K = 256 (23.8 ms vs 24.2 ms).
* k_quants: WIP super-blocks with 64 weights
Q3_K works on Metal and is slightly faster
than QK_K = 256 (26.6 ms vs 28.3 ms).
* k_quants: WIP super-blocks with 64 weights
Q5_K works on Metal and is slightly faster
than QK_K = 256 (23.7 ms vs 26.3 ms).
* k_quants: call them _K, not _k, also on Metal
* k_quants: correctly define QK_K in llama.cpp
* Fixed bug in q4_K quantization added with the 64-block addition
* Simplify via lambda
* k_quants: swicth Q3_K to 4-bit scales when QK_K = 64
Otherwise there isn't much benefit from this
quantization type. There is some very slight loss
in accuracy, but we reduce size by ~7%.
E.g., for OpenLLaMA-3B, Q3_K_S perplexity is
8.6131 with 8-bit scales and 8.6352 with 4-bit,
while file size decreases from 1.53G to 1.44G.
* k_quants: switch Q4_K to 4-bit scales when QK_K = 64
Here the loss in accuracy is greater than for Q3_K,
but the Q4_K points still move further to the left on
the perplexity vs size curve.
* k_quants: forgot to add the Metal changes in last commit
* k_quants: change Q5_K to be type 0 when QK_K = 64
Still needs AVX2 implementation
* k_quants: AVX2 implementation for new 64-weight Q5_K
* k_quants: 10% faster ARM_NEON Q5_K dot product
* k_quants: fixed issue caused by merging with master
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-26 18:43:07 +02:00
|
|
|
} block_q6_K;
|
Metal implementation for all k_quants (#1807)
* metal : improve q4_K
28.3 -> 26.0 ms/token by avoiding a branch in the
calculation of the scales.
* metal : small improvement for Q4_K
* metal : still optimizing Q4_K
This commit pushes it down to 25.3 ms / token.
The crazy idea of using 6 bits for the scales is really costly on
Metal: if I remove the bit fiddling necessary to make the block
scales, time goes almost to the Q4_0 23 ms/token.
Before pushing the k-quants upstream I had a Q4_K variant that
had used 8-bit scales. It wasn't more accurate, used 0.125 bits more per weight,
was running slightly slower on the CPU (due to the larger model size
and being memory bound there), and the difference was entirely
negligible under CUDA. So, I decided to publish the version with 6-bit
scales. Perhaps I should re-consider and change to 8-bit scales?
* metal : some more optimizations
Q2_K: 25.4 ms/token
Q6_K: 27.3 ms/token
Q4_0: 22.8 ms/token
Q4_1: 23.1 ms/token
* metal : Q3_K support
Something is not quite right yet.
* metal : Q5_K support
Initial version achieves 31.2 ms/token, 210 GB/s
* metal : still not able to figure out why q3_K does not work
* Minor
* metal : yet another failed attempt to make q3_K work
* metal : optimize Q5_K
31.2 ms -> 27.8 ms.
250 GB/s.
* metal : q3_K still not working
Adding a heavily commented q3_K metal kernel to explain
my obviously faulty logic. Perhaps someone could spot the issue?
* metal : q3_K finally working
Not optimized at all.
What was the issue? The scales are not 4-bytes aligned,
and I was accessing them with a uint32_t pointer.
When I tried that on CUDA, I got an error (illegal memory access)
and added a memcpy to a local array of 3 uint32_t's.
But on Metal it told me there is no memcpy, so I tried
accessing directly. There is no error, just garbage results.
At some point I did try accessing the scales with an uint16_t
pointer (the scales are for sure 2-byte aligned), but was
still getting garbage. I guess, there must have been another bug.
No access to scales is via a uint16_t pointer and, after starting
from scratch from the C dequantize function, it finally works.
* metal : Q3_K 1st optimization pass
* metal : Q3_K second optimization pass - 29.6 ms/token
* metal : Q3_K cleanup
* metal : fixed accidentally broken Q2_K
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-12 21:39:21 +02:00
|
|
|
// 210 bytes / block
|
2023-06-08 18:46:22 +02:00
|
|
|
|
2024-01-08 16:02:32 +01:00
|
|
|
typedef struct {
|
|
|
|
half d;
|
|
|
|
uint16_t qs[QK_K/8];
|
|
|
|
} block_iq2_xxs;
|
|
|
|
// 66 bytes / block for QK_K = 256, so 2.0625 bpw
|
|
|
|
|
2024-01-11 20:39:39 +01:00
|
|
|
typedef struct {
|
|
|
|
half d;
|
|
|
|
uint16_t qs[QK_K/8];
|
|
|
|
uint8_t scales[QK_K/32];
|
|
|
|
} block_iq2_xs;
|
|
|
|
// 74 bytes / block for QK_K = 256, so 2.3125 bpw
|
|
|
|
|
2024-01-30 14:14:12 +01:00
|
|
|
typedef struct {
|
|
|
|
half d;
|
|
|
|
uint8_t qs[3*QK_K/8];
|
|
|
|
} block_iq3_xxs;
|
|
|
|
// 98 bytes / block for QK_K = 256, so 3.0625 bpw
|
|
|
|
|
2023-06-08 21:28:21 +02:00
|
|
|
//====================================== dot products =========================
|
|
|
|
|
2023-12-13 13:04:25 +01:00
|
|
|
void kernel_mul_mv_q2_K_f32_impl(
|
2023-06-08 21:28:21 +02:00
|
|
|
device const void * src0,
|
|
|
|
device const float * src1,
|
|
|
|
device float * dst,
|
|
|
|
constant int64_t & ne00,
|
2023-12-13 13:04:25 +01:00
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant int64_t & ne10,
|
|
|
|
constant int64_t & ne12,
|
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
|
|
|
constant uint & r2,
|
|
|
|
constant uint & r3,
|
2023-08-16 22:07:04 +02:00
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
2023-12-07 21:26:54 +01:00
|
|
|
uint tiisg[[thread_index_in_simdgroup]],
|
|
|
|
uint sgitg[[simdgroup_index_in_threadgroup]]) {
|
2023-06-08 21:28:21 +02:00
|
|
|
|
|
|
|
const int nb = ne00/QK_K;
|
2023-07-21 09:44:40 +02:00
|
|
|
const int r0 = tgpig.x;
|
|
|
|
const int r1 = tgpig.y;
|
2023-12-07 21:26:54 +01:00
|
|
|
const int im = tgpig.z;
|
2023-06-08 21:28:21 +02:00
|
|
|
|
2023-07-21 09:44:40 +02:00
|
|
|
const int first_row = (r0 * N_SIMDGROUP + sgitg) * N_DST;
|
|
|
|
const int ib_row = first_row * nb;
|
2023-12-07 21:26:54 +01:00
|
|
|
|
|
|
|
const uint i12 = im%ne12;
|
|
|
|
const uint i13 = im/ne12;
|
|
|
|
|
|
|
|
const uint offset0 = (i12/r2)*(nb*ne01) + (i13/r3)*(nb*ne01*ne02);
|
|
|
|
|
2023-08-16 22:07:04 +02:00
|
|
|
device const block_q2_K * x = (device const block_q2_K *) src0 + ib_row + offset0;
|
2023-12-07 21:26:54 +01:00
|
|
|
device const float * y = (device const float *) src1 + r1*ne10 + im*ne00*ne1;
|
|
|
|
|
2023-07-21 09:44:40 +02:00
|
|
|
float yl[32];
|
|
|
|
float sumf[N_DST]={0.f}, all_sum;
|
2023-06-08 21:28:21 +02:00
|
|
|
|
2023-07-21 09:44:40 +02:00
|
|
|
const int step = sizeof(block_q2_K) * nb;
|
k-quants : support for super-block size of 64 (#2001)
* k_quants: WIP super-blocks with 64 weights
* k_quants: WIP super-blocks with 64 weights
Q6_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q4_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q2_K scalar and AVX2 works. Q2_K is way too slow (it is actually slower
than the scalar implementation)
* k_quants: WIP super-blocks with 64 weights
Q3_K scalar and AVX2 works.
* k_quants: WIP super-blocks with 64 weights
Q5_K scalar and AVX2 works, and with that all
k_quants are done on AVX2 and scalar
* k_quants: WIP super-blocks with 64 weights
Q6_K working on CUDA. Cannot make it run quite as gast as
with super-blocks with 256 weigths: 8% slower on 4080,
20% slower on the 1660 (but there we fit 1 less layer on the
GPU because pf the larger model size), so some fraction of
these 20% is due to that,
* k_quants: WIP super-blocks with 64 weights
Q4_K working on CUDA. ~10% slower on GTX-1660,
16% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q2_K working on CUDA. ~3% slower on GTX-1660,
10% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q3_K working on CUDA.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on CUDA, and with this CUDA is done.
* k_quants: WIP super-blocks with 64 weights
Q6_K working on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Q4_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q2_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q3_K working on ARM_NEON, but quite a bit slower than 256 weights.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on ARM_NEON, but quite a bit slower than 256 weights.
With that, we have full support for ARM_NEON, although
performance is not quite there.
* k_quants: WIP super-blocks with 64 weights
Slightly more efficient Q3_K and Q5_K
* k_quants: WIP super-blocks with 64 weights
Another small improvement for Q3_K and Q5_K on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Yet another speedup for Q5_K on ARM_NEON.
We are now within 10% of the QK_K = 256 version.
* k_quants: WIP super-blocks with 64 weights
* We are able to pass preprocessor macros to the Metal
compiler
* Q6_K works and is actually slightly more efficient than
the QK_K = 256 version (25.2 ms vs 25.8 ms)
* k_quants: WIP super-blocks with 64 weights
Q4_K works on Metal and is actually slightly faster
than QK_K = 256 (21.95 ms vs 24.0 ms).
* k_quants: WIP super-blocks with 64 weights
Q2_K works on Metal and is very slightly faster
than QK_K = 256 (23.8 ms vs 24.2 ms).
* k_quants: WIP super-blocks with 64 weights
Q3_K works on Metal and is slightly faster
than QK_K = 256 (26.6 ms vs 28.3 ms).
* k_quants: WIP super-blocks with 64 weights
Q5_K works on Metal and is slightly faster
than QK_K = 256 (23.7 ms vs 26.3 ms).
* k_quants: call them _K, not _k, also on Metal
* k_quants: correctly define QK_K in llama.cpp
* Fixed bug in q4_K quantization added with the 64-block addition
* Simplify via lambda
* k_quants: swicth Q3_K to 4-bit scales when QK_K = 64
Otherwise there isn't much benefit from this
quantization type. There is some very slight loss
in accuracy, but we reduce size by ~7%.
E.g., for OpenLLaMA-3B, Q3_K_S perplexity is
8.6131 with 8-bit scales and 8.6352 with 4-bit,
while file size decreases from 1.53G to 1.44G.
* k_quants: switch Q4_K to 4-bit scales when QK_K = 64
Here the loss in accuracy is greater than for Q3_K,
but the Q4_K points still move further to the left on
the perplexity vs size curve.
* k_quants: forgot to add the Metal changes in last commit
* k_quants: change Q5_K to be type 0 when QK_K = 64
Still needs AVX2 implementation
* k_quants: AVX2 implementation for new 64-weight Q5_K
* k_quants: 10% faster ARM_NEON Q5_K dot product
* k_quants: fixed issue caused by merging with master
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-26 18:43:07 +02:00
|
|
|
|
|
|
|
#if QK_K == 256
|
2023-07-21 09:44:40 +02:00
|
|
|
const int ix = tiisg/8; // 0...3
|
|
|
|
const int it = tiisg%8; // 0...7
|
2023-12-07 21:26:54 +01:00
|
|
|
const int iq = it/4; // 0 or 1
|
2023-07-21 09:44:40 +02:00
|
|
|
const int ir = it%4; // 0...3
|
|
|
|
const int is = (8*ir)/16;// 0 or 1
|
|
|
|
|
2023-12-07 21:26:54 +01:00
|
|
|
device const float * y4 = y + ix * QK_K + 128 * iq + 8 * ir;
|
2023-07-21 09:44:40 +02:00
|
|
|
|
|
|
|
for (int ib = ix; ib < nb; ib += 4) {
|
|
|
|
|
|
|
|
float4 sumy = {0.f, 0.f, 0.f, 0.f};
|
|
|
|
for (int i = 0; i < 8; ++i) {
|
|
|
|
yl[i+ 0] = y4[i+ 0]; sumy[0] += yl[i+ 0];
|
|
|
|
yl[i+ 8] = y4[i+32]; sumy[1] += yl[i+ 8];
|
|
|
|
yl[i+16] = y4[i+64]; sumy[2] += yl[i+16];
|
|
|
|
yl[i+24] = y4[i+96]; sumy[3] += yl[i+24];
|
2023-06-08 21:28:21 +02:00
|
|
|
}
|
|
|
|
|
2023-12-07 21:26:54 +01:00
|
|
|
device const uint8_t * sc = (device const uint8_t *)x[ib].scales + 8*iq + is;
|
|
|
|
device const uint16_t * qs = (device const uint16_t *)x[ib].qs + 16 * iq + 4 * ir;
|
2023-07-21 09:44:40 +02:00
|
|
|
device const half * dh = &x[ib].d;
|
Metal implementation for all k_quants (#1807)
* metal : improve q4_K
28.3 -> 26.0 ms/token by avoiding a branch in the
calculation of the scales.
* metal : small improvement for Q4_K
* metal : still optimizing Q4_K
This commit pushes it down to 25.3 ms / token.
The crazy idea of using 6 bits for the scales is really costly on
Metal: if I remove the bit fiddling necessary to make the block
scales, time goes almost to the Q4_0 23 ms/token.
Before pushing the k-quants upstream I had a Q4_K variant that
had used 8-bit scales. It wasn't more accurate, used 0.125 bits more per weight,
was running slightly slower on the CPU (due to the larger model size
and being memory bound there), and the difference was entirely
negligible under CUDA. So, I decided to publish the version with 6-bit
scales. Perhaps I should re-consider and change to 8-bit scales?
* metal : some more optimizations
Q2_K: 25.4 ms/token
Q6_K: 27.3 ms/token
Q4_0: 22.8 ms/token
Q4_1: 23.1 ms/token
* metal : Q3_K support
Something is not quite right yet.
* metal : Q5_K support
Initial version achieves 31.2 ms/token, 210 GB/s
* metal : still not able to figure out why q3_K does not work
* Minor
* metal : yet another failed attempt to make q3_K work
* metal : optimize Q5_K
31.2 ms -> 27.8 ms.
250 GB/s.
* metal : q3_K still not working
Adding a heavily commented q3_K metal kernel to explain
my obviously faulty logic. Perhaps someone could spot the issue?
* metal : q3_K finally working
Not optimized at all.
What was the issue? The scales are not 4-bytes aligned,
and I was accessing them with a uint32_t pointer.
When I tried that on CUDA, I got an error (illegal memory access)
and added a memcpy to a local array of 3 uint32_t's.
But on Metal it told me there is no memcpy, so I tried
accessing directly. There is no error, just garbage results.
At some point I did try accessing the scales with an uint16_t
pointer (the scales are for sure 2-byte aligned), but was
still getting garbage. I guess, there must have been another bug.
No access to scales is via a uint16_t pointer and, after starting
from scratch from the C dequantize function, it finally works.
* metal : Q3_K 1st optimization pass
* metal : Q3_K second optimization pass - 29.6 ms/token
* metal : Q3_K cleanup
* metal : fixed accidentally broken Q2_K
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-12 21:39:21 +02:00
|
|
|
|
2023-07-21 09:44:40 +02:00
|
|
|
for (int row = 0; row < N_DST; row++) {
|
2023-06-08 21:28:21 +02:00
|
|
|
|
2023-07-21 09:44:40 +02:00
|
|
|
float4 acc1 = {0.f, 0.f, 0.f, 0.f};
|
|
|
|
float4 acc2 = {0.f, 0.f, 0.f, 0.f};
|
|
|
|
for (int i = 0; i < 8; i += 2) {
|
|
|
|
acc1[0] += yl[i+ 0] * (qs[i/2] & 0x0003);
|
|
|
|
acc2[0] += yl[i+ 1] * (qs[i/2] & 0x0300);
|
|
|
|
acc1[1] += yl[i+ 8] * (qs[i/2] & 0x000c);
|
|
|
|
acc2[1] += yl[i+ 9] * (qs[i/2] & 0x0c00);
|
|
|
|
acc1[2] += yl[i+16] * (qs[i/2] & 0x0030);
|
|
|
|
acc2[2] += yl[i+17] * (qs[i/2] & 0x3000);
|
|
|
|
acc1[3] += yl[i+24] * (qs[i/2] & 0x00c0);
|
|
|
|
acc2[3] += yl[i+25] * (qs[i/2] & 0xc000);
|
|
|
|
}
|
|
|
|
float dall = dh[0];
|
|
|
|
float dmin = dh[1] * 1.f/16.f;
|
|
|
|
sumf[row] += dall * ((acc1[0] + 1.f/256.f * acc2[0]) * (sc[0] & 0xF) * 1.f/ 1.f +
|
|
|
|
(acc1[1] + 1.f/256.f * acc2[1]) * (sc[2] & 0xF) * 1.f/ 4.f +
|
|
|
|
(acc1[2] + 1.f/256.f * acc2[2]) * (sc[4] & 0xF) * 1.f/16.f +
|
|
|
|
(acc1[3] + 1.f/256.f * acc2[3]) * (sc[6] & 0xF) * 1.f/64.f) -
|
|
|
|
dmin * (sumy[0] * (sc[0] & 0xF0) + sumy[1] * (sc[2] & 0xF0) + sumy[2] * (sc[4] & 0xF0) + sumy[3] * (sc[6] & 0xF0));
|
|
|
|
|
|
|
|
qs += step/2;
|
|
|
|
sc += step;
|
|
|
|
dh += step/2;
|
|
|
|
}
|
|
|
|
|
|
|
|
y4 += 4 * QK_K;
|
2023-06-08 21:28:21 +02:00
|
|
|
}
|
k-quants : support for super-block size of 64 (#2001)
* k_quants: WIP super-blocks with 64 weights
* k_quants: WIP super-blocks with 64 weights
Q6_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q4_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q2_K scalar and AVX2 works. Q2_K is way too slow (it is actually slower
than the scalar implementation)
* k_quants: WIP super-blocks with 64 weights
Q3_K scalar and AVX2 works.
* k_quants: WIP super-blocks with 64 weights
Q5_K scalar and AVX2 works, and with that all
k_quants are done on AVX2 and scalar
* k_quants: WIP super-blocks with 64 weights
Q6_K working on CUDA. Cannot make it run quite as gast as
with super-blocks with 256 weigths: 8% slower on 4080,
20% slower on the 1660 (but there we fit 1 less layer on the
GPU because pf the larger model size), so some fraction of
these 20% is due to that,
* k_quants: WIP super-blocks with 64 weights
Q4_K working on CUDA. ~10% slower on GTX-1660,
16% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q2_K working on CUDA. ~3% slower on GTX-1660,
10% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q3_K working on CUDA.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on CUDA, and with this CUDA is done.
* k_quants: WIP super-blocks with 64 weights
Q6_K working on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Q4_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q2_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q3_K working on ARM_NEON, but quite a bit slower than 256 weights.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on ARM_NEON, but quite a bit slower than 256 weights.
With that, we have full support for ARM_NEON, although
performance is not quite there.
* k_quants: WIP super-blocks with 64 weights
Slightly more efficient Q3_K and Q5_K
* k_quants: WIP super-blocks with 64 weights
Another small improvement for Q3_K and Q5_K on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Yet another speedup for Q5_K on ARM_NEON.
We are now within 10% of the QK_K = 256 version.
* k_quants: WIP super-blocks with 64 weights
* We are able to pass preprocessor macros to the Metal
compiler
* Q6_K works and is actually slightly more efficient than
the QK_K = 256 version (25.2 ms vs 25.8 ms)
* k_quants: WIP super-blocks with 64 weights
Q4_K works on Metal and is actually slightly faster
than QK_K = 256 (21.95 ms vs 24.0 ms).
* k_quants: WIP super-blocks with 64 weights
Q2_K works on Metal and is very slightly faster
than QK_K = 256 (23.8 ms vs 24.2 ms).
* k_quants: WIP super-blocks with 64 weights
Q3_K works on Metal and is slightly faster
than QK_K = 256 (26.6 ms vs 28.3 ms).
* k_quants: WIP super-blocks with 64 weights
Q5_K works on Metal and is slightly faster
than QK_K = 256 (23.7 ms vs 26.3 ms).
* k_quants: call them _K, not _k, also on Metal
* k_quants: correctly define QK_K in llama.cpp
* Fixed bug in q4_K quantization added with the 64-block addition
* Simplify via lambda
* k_quants: swicth Q3_K to 4-bit scales when QK_K = 64
Otherwise there isn't much benefit from this
quantization type. There is some very slight loss
in accuracy, but we reduce size by ~7%.
E.g., for OpenLLaMA-3B, Q3_K_S perplexity is
8.6131 with 8-bit scales and 8.6352 with 4-bit,
while file size decreases from 1.53G to 1.44G.
* k_quants: switch Q4_K to 4-bit scales when QK_K = 64
Here the loss in accuracy is greater than for Q3_K,
but the Q4_K points still move further to the left on
the perplexity vs size curve.
* k_quants: forgot to add the Metal changes in last commit
* k_quants: change Q5_K to be type 0 when QK_K = 64
Still needs AVX2 implementation
* k_quants: AVX2 implementation for new 64-weight Q5_K
* k_quants: 10% faster ARM_NEON Q5_K dot product
* k_quants: fixed issue caused by merging with master
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-26 18:43:07 +02:00
|
|
|
#else
|
2023-07-21 09:44:40 +02:00
|
|
|
const int ix = tiisg/2; // 0...15
|
|
|
|
const int it = tiisg%2; // 0...1
|
|
|
|
|
|
|
|
device const float * y4 = y + ix * QK_K + 8 * it;
|
2023-06-08 21:28:21 +02:00
|
|
|
|
2023-07-21 09:44:40 +02:00
|
|
|
for (int ib = ix; ib < nb; ib += 16) {
|
Metal implementation for all k_quants (#1807)
* metal : improve q4_K
28.3 -> 26.0 ms/token by avoiding a branch in the
calculation of the scales.
* metal : small improvement for Q4_K
* metal : still optimizing Q4_K
This commit pushes it down to 25.3 ms / token.
The crazy idea of using 6 bits for the scales is really costly on
Metal: if I remove the bit fiddling necessary to make the block
scales, time goes almost to the Q4_0 23 ms/token.
Before pushing the k-quants upstream I had a Q4_K variant that
had used 8-bit scales. It wasn't more accurate, used 0.125 bits more per weight,
was running slightly slower on the CPU (due to the larger model size
and being memory bound there), and the difference was entirely
negligible under CUDA. So, I decided to publish the version with 6-bit
scales. Perhaps I should re-consider and change to 8-bit scales?
* metal : some more optimizations
Q2_K: 25.4 ms/token
Q6_K: 27.3 ms/token
Q4_0: 22.8 ms/token
Q4_1: 23.1 ms/token
* metal : Q3_K support
Something is not quite right yet.
* metal : Q5_K support
Initial version achieves 31.2 ms/token, 210 GB/s
* metal : still not able to figure out why q3_K does not work
* Minor
* metal : yet another failed attempt to make q3_K work
* metal : optimize Q5_K
31.2 ms -> 27.8 ms.
250 GB/s.
* metal : q3_K still not working
Adding a heavily commented q3_K metal kernel to explain
my obviously faulty logic. Perhaps someone could spot the issue?
* metal : q3_K finally working
Not optimized at all.
What was the issue? The scales are not 4-bytes aligned,
and I was accessing them with a uint32_t pointer.
When I tried that on CUDA, I got an error (illegal memory access)
and added a memcpy to a local array of 3 uint32_t's.
But on Metal it told me there is no memcpy, so I tried
accessing directly. There is no error, just garbage results.
At some point I did try accessing the scales with an uint16_t
pointer (the scales are for sure 2-byte aligned), but was
still getting garbage. I guess, there must have been another bug.
No access to scales is via a uint16_t pointer and, after starting
from scratch from the C dequantize function, it finally works.
* metal : Q3_K 1st optimization pass
* metal : Q3_K second optimization pass - 29.6 ms/token
* metal : Q3_K cleanup
* metal : fixed accidentally broken Q2_K
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-12 21:39:21 +02:00
|
|
|
|
2023-07-21 09:44:40 +02:00
|
|
|
float4 sumy = {0.f, 0.f, 0.f, 0.f};
|
|
|
|
for (int i = 0; i < 8; ++i) {
|
|
|
|
yl[i+ 0] = y4[i+ 0]; sumy[0] += yl[i+ 0];
|
|
|
|
yl[i+ 8] = y4[i+16]; sumy[1] += yl[i+ 8];
|
|
|
|
yl[i+16] = y4[i+32]; sumy[2] += yl[i+16];
|
|
|
|
yl[i+24] = y4[i+48]; sumy[3] += yl[i+24];
|
|
|
|
}
|
k-quants : support for super-block size of 64 (#2001)
* k_quants: WIP super-blocks with 64 weights
* k_quants: WIP super-blocks with 64 weights
Q6_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q4_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q2_K scalar and AVX2 works. Q2_K is way too slow (it is actually slower
than the scalar implementation)
* k_quants: WIP super-blocks with 64 weights
Q3_K scalar and AVX2 works.
* k_quants: WIP super-blocks with 64 weights
Q5_K scalar and AVX2 works, and with that all
k_quants are done on AVX2 and scalar
* k_quants: WIP super-blocks with 64 weights
Q6_K working on CUDA. Cannot make it run quite as gast as
with super-blocks with 256 weigths: 8% slower on 4080,
20% slower on the 1660 (but there we fit 1 less layer on the
GPU because pf the larger model size), so some fraction of
these 20% is due to that,
* k_quants: WIP super-blocks with 64 weights
Q4_K working on CUDA. ~10% slower on GTX-1660,
16% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q2_K working on CUDA. ~3% slower on GTX-1660,
10% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q3_K working on CUDA.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on CUDA, and with this CUDA is done.
* k_quants: WIP super-blocks with 64 weights
Q6_K working on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Q4_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q2_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q3_K working on ARM_NEON, but quite a bit slower than 256 weights.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on ARM_NEON, but quite a bit slower than 256 weights.
With that, we have full support for ARM_NEON, although
performance is not quite there.
* k_quants: WIP super-blocks with 64 weights
Slightly more efficient Q3_K and Q5_K
* k_quants: WIP super-blocks with 64 weights
Another small improvement for Q3_K and Q5_K on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Yet another speedup for Q5_K on ARM_NEON.
We are now within 10% of the QK_K = 256 version.
* k_quants: WIP super-blocks with 64 weights
* We are able to pass preprocessor macros to the Metal
compiler
* Q6_K works and is actually slightly more efficient than
the QK_K = 256 version (25.2 ms vs 25.8 ms)
* k_quants: WIP super-blocks with 64 weights
Q4_K works on Metal and is actually slightly faster
than QK_K = 256 (21.95 ms vs 24.0 ms).
* k_quants: WIP super-blocks with 64 weights
Q2_K works on Metal and is very slightly faster
than QK_K = 256 (23.8 ms vs 24.2 ms).
* k_quants: WIP super-blocks with 64 weights
Q3_K works on Metal and is slightly faster
than QK_K = 256 (26.6 ms vs 28.3 ms).
* k_quants: WIP super-blocks with 64 weights
Q5_K works on Metal and is slightly faster
than QK_K = 256 (23.7 ms vs 26.3 ms).
* k_quants: call them _K, not _k, also on Metal
* k_quants: correctly define QK_K in llama.cpp
* Fixed bug in q4_K quantization added with the 64-block addition
* Simplify via lambda
* k_quants: swicth Q3_K to 4-bit scales when QK_K = 64
Otherwise there isn't much benefit from this
quantization type. There is some very slight loss
in accuracy, but we reduce size by ~7%.
E.g., for OpenLLaMA-3B, Q3_K_S perplexity is
8.6131 with 8-bit scales and 8.6352 with 4-bit,
while file size decreases from 1.53G to 1.44G.
* k_quants: switch Q4_K to 4-bit scales when QK_K = 64
Here the loss in accuracy is greater than for Q3_K,
but the Q4_K points still move further to the left on
the perplexity vs size curve.
* k_quants: forgot to add the Metal changes in last commit
* k_quants: change Q5_K to be type 0 when QK_K = 64
Still needs AVX2 implementation
* k_quants: AVX2 implementation for new 64-weight Q5_K
* k_quants: 10% faster ARM_NEON Q5_K dot product
* k_quants: fixed issue caused by merging with master
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-26 18:43:07 +02:00
|
|
|
|
2023-07-21 09:44:40 +02:00
|
|
|
device const uint8_t * sc = (device const uint8_t *)x[ib].scales;
|
|
|
|
device const uint16_t * qs = (device const uint16_t *)x[ib].qs + 4 * it;
|
|
|
|
device const half * dh = &x[ib].d;
|
k-quants : support for super-block size of 64 (#2001)
* k_quants: WIP super-blocks with 64 weights
* k_quants: WIP super-blocks with 64 weights
Q6_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q4_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q2_K scalar and AVX2 works. Q2_K is way too slow (it is actually slower
than the scalar implementation)
* k_quants: WIP super-blocks with 64 weights
Q3_K scalar and AVX2 works.
* k_quants: WIP super-blocks with 64 weights
Q5_K scalar and AVX2 works, and with that all
k_quants are done on AVX2 and scalar
* k_quants: WIP super-blocks with 64 weights
Q6_K working on CUDA. Cannot make it run quite as gast as
with super-blocks with 256 weigths: 8% slower on 4080,
20% slower on the 1660 (but there we fit 1 less layer on the
GPU because pf the larger model size), so some fraction of
these 20% is due to that,
* k_quants: WIP super-blocks with 64 weights
Q4_K working on CUDA. ~10% slower on GTX-1660,
16% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q2_K working on CUDA. ~3% slower on GTX-1660,
10% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q3_K working on CUDA.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on CUDA, and with this CUDA is done.
* k_quants: WIP super-blocks with 64 weights
Q6_K working on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Q4_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q2_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q3_K working on ARM_NEON, but quite a bit slower than 256 weights.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on ARM_NEON, but quite a bit slower than 256 weights.
With that, we have full support for ARM_NEON, although
performance is not quite there.
* k_quants: WIP super-blocks with 64 weights
Slightly more efficient Q3_K and Q5_K
* k_quants: WIP super-blocks with 64 weights
Another small improvement for Q3_K and Q5_K on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Yet another speedup for Q5_K on ARM_NEON.
We are now within 10% of the QK_K = 256 version.
* k_quants: WIP super-blocks with 64 weights
* We are able to pass preprocessor macros to the Metal
compiler
* Q6_K works and is actually slightly more efficient than
the QK_K = 256 version (25.2 ms vs 25.8 ms)
* k_quants: WIP super-blocks with 64 weights
Q4_K works on Metal and is actually slightly faster
than QK_K = 256 (21.95 ms vs 24.0 ms).
* k_quants: WIP super-blocks with 64 weights
Q2_K works on Metal and is very slightly faster
than QK_K = 256 (23.8 ms vs 24.2 ms).
* k_quants: WIP super-blocks with 64 weights
Q3_K works on Metal and is slightly faster
than QK_K = 256 (26.6 ms vs 28.3 ms).
* k_quants: WIP super-blocks with 64 weights
Q5_K works on Metal and is slightly faster
than QK_K = 256 (23.7 ms vs 26.3 ms).
* k_quants: call them _K, not _k, also on Metal
* k_quants: correctly define QK_K in llama.cpp
* Fixed bug in q4_K quantization added with the 64-block addition
* Simplify via lambda
* k_quants: swicth Q3_K to 4-bit scales when QK_K = 64
Otherwise there isn't much benefit from this
quantization type. There is some very slight loss
in accuracy, but we reduce size by ~7%.
E.g., for OpenLLaMA-3B, Q3_K_S perplexity is
8.6131 with 8-bit scales and 8.6352 with 4-bit,
while file size decreases from 1.53G to 1.44G.
* k_quants: switch Q4_K to 4-bit scales when QK_K = 64
Here the loss in accuracy is greater than for Q3_K,
but the Q4_K points still move further to the left on
the perplexity vs size curve.
* k_quants: forgot to add the Metal changes in last commit
* k_quants: change Q5_K to be type 0 when QK_K = 64
Still needs AVX2 implementation
* k_quants: AVX2 implementation for new 64-weight Q5_K
* k_quants: 10% faster ARM_NEON Q5_K dot product
* k_quants: fixed issue caused by merging with master
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-26 18:43:07 +02:00
|
|
|
|
2023-07-21 09:44:40 +02:00
|
|
|
for (int row = 0; row < N_DST; row++) {
|
k-quants : support for super-block size of 64 (#2001)
* k_quants: WIP super-blocks with 64 weights
* k_quants: WIP super-blocks with 64 weights
Q6_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q4_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q2_K scalar and AVX2 works. Q2_K is way too slow (it is actually slower
than the scalar implementation)
* k_quants: WIP super-blocks with 64 weights
Q3_K scalar and AVX2 works.
* k_quants: WIP super-blocks with 64 weights
Q5_K scalar and AVX2 works, and with that all
k_quants are done on AVX2 and scalar
* k_quants: WIP super-blocks with 64 weights
Q6_K working on CUDA. Cannot make it run quite as gast as
with super-blocks with 256 weigths: 8% slower on 4080,
20% slower on the 1660 (but there we fit 1 less layer on the
GPU because pf the larger model size), so some fraction of
these 20% is due to that,
* k_quants: WIP super-blocks with 64 weights
Q4_K working on CUDA. ~10% slower on GTX-1660,
16% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q2_K working on CUDA. ~3% slower on GTX-1660,
10% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q3_K working on CUDA.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on CUDA, and with this CUDA is done.
* k_quants: WIP super-blocks with 64 weights
Q6_K working on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Q4_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q2_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q3_K working on ARM_NEON, but quite a bit slower than 256 weights.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on ARM_NEON, but quite a bit slower than 256 weights.
With that, we have full support for ARM_NEON, although
performance is not quite there.
* k_quants: WIP super-blocks with 64 weights
Slightly more efficient Q3_K and Q5_K
* k_quants: WIP super-blocks with 64 weights
Another small improvement for Q3_K and Q5_K on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Yet another speedup for Q5_K on ARM_NEON.
We are now within 10% of the QK_K = 256 version.
* k_quants: WIP super-blocks with 64 weights
* We are able to pass preprocessor macros to the Metal
compiler
* Q6_K works and is actually slightly more efficient than
the QK_K = 256 version (25.2 ms vs 25.8 ms)
* k_quants: WIP super-blocks with 64 weights
Q4_K works on Metal and is actually slightly faster
than QK_K = 256 (21.95 ms vs 24.0 ms).
* k_quants: WIP super-blocks with 64 weights
Q2_K works on Metal and is very slightly faster
than QK_K = 256 (23.8 ms vs 24.2 ms).
* k_quants: WIP super-blocks with 64 weights
Q3_K works on Metal and is slightly faster
than QK_K = 256 (26.6 ms vs 28.3 ms).
* k_quants: WIP super-blocks with 64 weights
Q5_K works on Metal and is slightly faster
than QK_K = 256 (23.7 ms vs 26.3 ms).
* k_quants: call them _K, not _k, also on Metal
* k_quants: correctly define QK_K in llama.cpp
* Fixed bug in q4_K quantization added with the 64-block addition
* Simplify via lambda
* k_quants: swicth Q3_K to 4-bit scales when QK_K = 64
Otherwise there isn't much benefit from this
quantization type. There is some very slight loss
in accuracy, but we reduce size by ~7%.
E.g., for OpenLLaMA-3B, Q3_K_S perplexity is
8.6131 with 8-bit scales and 8.6352 with 4-bit,
while file size decreases from 1.53G to 1.44G.
* k_quants: switch Q4_K to 4-bit scales when QK_K = 64
Here the loss in accuracy is greater than for Q3_K,
but the Q4_K points still move further to the left on
the perplexity vs size curve.
* k_quants: forgot to add the Metal changes in last commit
* k_quants: change Q5_K to be type 0 when QK_K = 64
Still needs AVX2 implementation
* k_quants: AVX2 implementation for new 64-weight Q5_K
* k_quants: 10% faster ARM_NEON Q5_K dot product
* k_quants: fixed issue caused by merging with master
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-26 18:43:07 +02:00
|
|
|
|
2023-07-21 09:44:40 +02:00
|
|
|
float4 acc1 = {0.f, 0.f, 0.f, 0.f};
|
|
|
|
float4 acc2 = {0.f, 0.f, 0.f, 0.f};
|
|
|
|
for (int i = 0; i < 8; i += 2) {
|
|
|
|
acc1[0] += yl[i+ 0] * (qs[i/2] & 0x0003);
|
|
|
|
acc2[0] += yl[i+ 1] * (qs[i/2] & 0x0300);
|
|
|
|
acc1[1] += yl[i+ 8] * (qs[i/2] & 0x000c);
|
|
|
|
acc2[1] += yl[i+ 9] * (qs[i/2] & 0x0c00);
|
|
|
|
acc1[2] += yl[i+16] * (qs[i/2] & 0x0030);
|
|
|
|
acc2[2] += yl[i+17] * (qs[i/2] & 0x3000);
|
|
|
|
acc1[3] += yl[i+24] * (qs[i/2] & 0x00c0);
|
|
|
|
acc2[3] += yl[i+25] * (qs[i/2] & 0xc000);
|
|
|
|
}
|
k-quants : support for super-block size of 64 (#2001)
* k_quants: WIP super-blocks with 64 weights
* k_quants: WIP super-blocks with 64 weights
Q6_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q4_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q2_K scalar and AVX2 works. Q2_K is way too slow (it is actually slower
than the scalar implementation)
* k_quants: WIP super-blocks with 64 weights
Q3_K scalar and AVX2 works.
* k_quants: WIP super-blocks with 64 weights
Q5_K scalar and AVX2 works, and with that all
k_quants are done on AVX2 and scalar
* k_quants: WIP super-blocks with 64 weights
Q6_K working on CUDA. Cannot make it run quite as gast as
with super-blocks with 256 weigths: 8% slower on 4080,
20% slower on the 1660 (but there we fit 1 less layer on the
GPU because pf the larger model size), so some fraction of
these 20% is due to that,
* k_quants: WIP super-blocks with 64 weights
Q4_K working on CUDA. ~10% slower on GTX-1660,
16% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q2_K working on CUDA. ~3% slower on GTX-1660,
10% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q3_K working on CUDA.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on CUDA, and with this CUDA is done.
* k_quants: WIP super-blocks with 64 weights
Q6_K working on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Q4_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q2_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q3_K working on ARM_NEON, but quite a bit slower than 256 weights.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on ARM_NEON, but quite a bit slower than 256 weights.
With that, we have full support for ARM_NEON, although
performance is not quite there.
* k_quants: WIP super-blocks with 64 weights
Slightly more efficient Q3_K and Q5_K
* k_quants: WIP super-blocks with 64 weights
Another small improvement for Q3_K and Q5_K on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Yet another speedup for Q5_K on ARM_NEON.
We are now within 10% of the QK_K = 256 version.
* k_quants: WIP super-blocks with 64 weights
* We are able to pass preprocessor macros to the Metal
compiler
* Q6_K works and is actually slightly more efficient than
the QK_K = 256 version (25.2 ms vs 25.8 ms)
* k_quants: WIP super-blocks with 64 weights
Q4_K works on Metal and is actually slightly faster
than QK_K = 256 (21.95 ms vs 24.0 ms).
* k_quants: WIP super-blocks with 64 weights
Q2_K works on Metal and is very slightly faster
than QK_K = 256 (23.8 ms vs 24.2 ms).
* k_quants: WIP super-blocks with 64 weights
Q3_K works on Metal and is slightly faster
than QK_K = 256 (26.6 ms vs 28.3 ms).
* k_quants: WIP super-blocks with 64 weights
Q5_K works on Metal and is slightly faster
than QK_K = 256 (23.7 ms vs 26.3 ms).
* k_quants: call them _K, not _k, also on Metal
* k_quants: correctly define QK_K in llama.cpp
* Fixed bug in q4_K quantization added with the 64-block addition
* Simplify via lambda
* k_quants: swicth Q3_K to 4-bit scales when QK_K = 64
Otherwise there isn't much benefit from this
quantization type. There is some very slight loss
in accuracy, but we reduce size by ~7%.
E.g., for OpenLLaMA-3B, Q3_K_S perplexity is
8.6131 with 8-bit scales and 8.6352 with 4-bit,
while file size decreases from 1.53G to 1.44G.
* k_quants: switch Q4_K to 4-bit scales when QK_K = 64
Here the loss in accuracy is greater than for Q3_K,
but the Q4_K points still move further to the left on
the perplexity vs size curve.
* k_quants: forgot to add the Metal changes in last commit
* k_quants: change Q5_K to be type 0 when QK_K = 64
Still needs AVX2 implementation
* k_quants: AVX2 implementation for new 64-weight Q5_K
* k_quants: 10% faster ARM_NEON Q5_K dot product
* k_quants: fixed issue caused by merging with master
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-26 18:43:07 +02:00
|
|
|
|
2023-07-21 09:44:40 +02:00
|
|
|
float dall = dh[0];
|
|
|
|
float dmin = dh[1];
|
|
|
|
sumf[row] += dall * ((acc1[0] + 1.f/256.f * acc2[0]) * (sc[0] & 0xF) * 1.f/ 1.f +
|
|
|
|
(acc1[1] + 1.f/256.f * acc2[1]) * (sc[1] & 0xF) * 1.f/ 4.f +
|
|
|
|
(acc1[2] + 1.f/256.f * acc2[2]) * (sc[2] & 0xF) * 1.f/16.f +
|
|
|
|
(acc1[3] + 1.f/256.f * acc2[3]) * (sc[3] & 0xF) * 1.f/64.f) -
|
|
|
|
dmin * (sumy[0] * (sc[0] >> 4) + sumy[1] * (sc[1] >> 4) + sumy[2] * (sc[2] >> 4) + sumy[3] * (sc[3] >> 4));
|
|
|
|
|
|
|
|
qs += step/2;
|
|
|
|
sc += step;
|
|
|
|
dh += step/2;
|
k-quants : support for super-block size of 64 (#2001)
* k_quants: WIP super-blocks with 64 weights
* k_quants: WIP super-blocks with 64 weights
Q6_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q4_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q2_K scalar and AVX2 works. Q2_K is way too slow (it is actually slower
than the scalar implementation)
* k_quants: WIP super-blocks with 64 weights
Q3_K scalar and AVX2 works.
* k_quants: WIP super-blocks with 64 weights
Q5_K scalar and AVX2 works, and with that all
k_quants are done on AVX2 and scalar
* k_quants: WIP super-blocks with 64 weights
Q6_K working on CUDA. Cannot make it run quite as gast as
with super-blocks with 256 weigths: 8% slower on 4080,
20% slower on the 1660 (but there we fit 1 less layer on the
GPU because pf the larger model size), so some fraction of
these 20% is due to that,
* k_quants: WIP super-blocks with 64 weights
Q4_K working on CUDA. ~10% slower on GTX-1660,
16% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q2_K working on CUDA. ~3% slower on GTX-1660,
10% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q3_K working on CUDA.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on CUDA, and with this CUDA is done.
* k_quants: WIP super-blocks with 64 weights
Q6_K working on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Q4_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q2_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q3_K working on ARM_NEON, but quite a bit slower than 256 weights.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on ARM_NEON, but quite a bit slower than 256 weights.
With that, we have full support for ARM_NEON, although
performance is not quite there.
* k_quants: WIP super-blocks with 64 weights
Slightly more efficient Q3_K and Q5_K
* k_quants: WIP super-blocks with 64 weights
Another small improvement for Q3_K and Q5_K on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Yet another speedup for Q5_K on ARM_NEON.
We are now within 10% of the QK_K = 256 version.
* k_quants: WIP super-blocks with 64 weights
* We are able to pass preprocessor macros to the Metal
compiler
* Q6_K works and is actually slightly more efficient than
the QK_K = 256 version (25.2 ms vs 25.8 ms)
* k_quants: WIP super-blocks with 64 weights
Q4_K works on Metal and is actually slightly faster
than QK_K = 256 (21.95 ms vs 24.0 ms).
* k_quants: WIP super-blocks with 64 weights
Q2_K works on Metal and is very slightly faster
than QK_K = 256 (23.8 ms vs 24.2 ms).
* k_quants: WIP super-blocks with 64 weights
Q3_K works on Metal and is slightly faster
than QK_K = 256 (26.6 ms vs 28.3 ms).
* k_quants: WIP super-blocks with 64 weights
Q5_K works on Metal and is slightly faster
than QK_K = 256 (23.7 ms vs 26.3 ms).
* k_quants: call them _K, not _k, also on Metal
* k_quants: correctly define QK_K in llama.cpp
* Fixed bug in q4_K quantization added with the 64-block addition
* Simplify via lambda
* k_quants: swicth Q3_K to 4-bit scales when QK_K = 64
Otherwise there isn't much benefit from this
quantization type. There is some very slight loss
in accuracy, but we reduce size by ~7%.
E.g., for OpenLLaMA-3B, Q3_K_S perplexity is
8.6131 with 8-bit scales and 8.6352 with 4-bit,
while file size decreases from 1.53G to 1.44G.
* k_quants: switch Q4_K to 4-bit scales when QK_K = 64
Here the loss in accuracy is greater than for Q3_K,
but the Q4_K points still move further to the left on
the perplexity vs size curve.
* k_quants: forgot to add the Metal changes in last commit
* k_quants: change Q5_K to be type 0 when QK_K = 64
Still needs AVX2 implementation
* k_quants: AVX2 implementation for new 64-weight Q5_K
* k_quants: 10% faster ARM_NEON Q5_K dot product
* k_quants: fixed issue caused by merging with master
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-26 18:43:07 +02:00
|
|
|
}
|
2023-07-21 09:44:40 +02:00
|
|
|
|
|
|
|
y4 += 16 * QK_K;
|
k-quants : support for super-block size of 64 (#2001)
* k_quants: WIP super-blocks with 64 weights
* k_quants: WIP super-blocks with 64 weights
Q6_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q4_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q2_K scalar and AVX2 works. Q2_K is way too slow (it is actually slower
than the scalar implementation)
* k_quants: WIP super-blocks with 64 weights
Q3_K scalar and AVX2 works.
* k_quants: WIP super-blocks with 64 weights
Q5_K scalar and AVX2 works, and with that all
k_quants are done on AVX2 and scalar
* k_quants: WIP super-blocks with 64 weights
Q6_K working on CUDA. Cannot make it run quite as gast as
with super-blocks with 256 weigths: 8% slower on 4080,
20% slower on the 1660 (but there we fit 1 less layer on the
GPU because pf the larger model size), so some fraction of
these 20% is due to that,
* k_quants: WIP super-blocks with 64 weights
Q4_K working on CUDA. ~10% slower on GTX-1660,
16% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q2_K working on CUDA. ~3% slower on GTX-1660,
10% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q3_K working on CUDA.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on CUDA, and with this CUDA is done.
* k_quants: WIP super-blocks with 64 weights
Q6_K working on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Q4_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q2_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q3_K working on ARM_NEON, but quite a bit slower than 256 weights.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on ARM_NEON, but quite a bit slower than 256 weights.
With that, we have full support for ARM_NEON, although
performance is not quite there.
* k_quants: WIP super-blocks with 64 weights
Slightly more efficient Q3_K and Q5_K
* k_quants: WIP super-blocks with 64 weights
Another small improvement for Q3_K and Q5_K on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Yet another speedup for Q5_K on ARM_NEON.
We are now within 10% of the QK_K = 256 version.
* k_quants: WIP super-blocks with 64 weights
* We are able to pass preprocessor macros to the Metal
compiler
* Q6_K works and is actually slightly more efficient than
the QK_K = 256 version (25.2 ms vs 25.8 ms)
* k_quants: WIP super-blocks with 64 weights
Q4_K works on Metal and is actually slightly faster
than QK_K = 256 (21.95 ms vs 24.0 ms).
* k_quants: WIP super-blocks with 64 weights
Q2_K works on Metal and is very slightly faster
than QK_K = 256 (23.8 ms vs 24.2 ms).
* k_quants: WIP super-blocks with 64 weights
Q3_K works on Metal and is slightly faster
than QK_K = 256 (26.6 ms vs 28.3 ms).
* k_quants: WIP super-blocks with 64 weights
Q5_K works on Metal and is slightly faster
than QK_K = 256 (23.7 ms vs 26.3 ms).
* k_quants: call them _K, not _k, also on Metal
* k_quants: correctly define QK_K in llama.cpp
* Fixed bug in q4_K quantization added with the 64-block addition
* Simplify via lambda
* k_quants: swicth Q3_K to 4-bit scales when QK_K = 64
Otherwise there isn't much benefit from this
quantization type. There is some very slight loss
in accuracy, but we reduce size by ~7%.
E.g., for OpenLLaMA-3B, Q3_K_S perplexity is
8.6131 with 8-bit scales and 8.6352 with 4-bit,
while file size decreases from 1.53G to 1.44G.
* k_quants: switch Q4_K to 4-bit scales when QK_K = 64
Here the loss in accuracy is greater than for Q3_K,
but the Q4_K points still move further to the left on
the perplexity vs size curve.
* k_quants: forgot to add the Metal changes in last commit
* k_quants: change Q5_K to be type 0 when QK_K = 64
Still needs AVX2 implementation
* k_quants: AVX2 implementation for new 64-weight Q5_K
* k_quants: 10% faster ARM_NEON Q5_K dot product
* k_quants: fixed issue caused by merging with master
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-26 18:43:07 +02:00
|
|
|
}
|
|
|
|
#endif
|
|
|
|
|
2023-07-21 09:44:40 +02:00
|
|
|
for (int row = 0; row < N_DST; ++row) {
|
|
|
|
all_sum = simd_sum(sumf[row]);
|
|
|
|
if (tiisg == 0) {
|
2023-12-07 21:26:54 +01:00
|
|
|
dst[r1*ne0 + im*ne0*ne1 + first_row + row] = all_sum;
|
2023-07-21 09:44:40 +02:00
|
|
|
}
|
2023-06-08 21:28:21 +02:00
|
|
|
}
|
Metal implementation for all k_quants (#1807)
* metal : improve q4_K
28.3 -> 26.0 ms/token by avoiding a branch in the
calculation of the scales.
* metal : small improvement for Q4_K
* metal : still optimizing Q4_K
This commit pushes it down to 25.3 ms / token.
The crazy idea of using 6 bits for the scales is really costly on
Metal: if I remove the bit fiddling necessary to make the block
scales, time goes almost to the Q4_0 23 ms/token.
Before pushing the k-quants upstream I had a Q4_K variant that
had used 8-bit scales. It wasn't more accurate, used 0.125 bits more per weight,
was running slightly slower on the CPU (due to the larger model size
and being memory bound there), and the difference was entirely
negligible under CUDA. So, I decided to publish the version with 6-bit
scales. Perhaps I should re-consider and change to 8-bit scales?
* metal : some more optimizations
Q2_K: 25.4 ms/token
Q6_K: 27.3 ms/token
Q4_0: 22.8 ms/token
Q4_1: 23.1 ms/token
* metal : Q3_K support
Something is not quite right yet.
* metal : Q5_K support
Initial version achieves 31.2 ms/token, 210 GB/s
* metal : still not able to figure out why q3_K does not work
* Minor
* metal : yet another failed attempt to make q3_K work
* metal : optimize Q5_K
31.2 ms -> 27.8 ms.
250 GB/s.
* metal : q3_K still not working
Adding a heavily commented q3_K metal kernel to explain
my obviously faulty logic. Perhaps someone could spot the issue?
* metal : q3_K finally working
Not optimized at all.
What was the issue? The scales are not 4-bytes aligned,
and I was accessing them with a uint32_t pointer.
When I tried that on CUDA, I got an error (illegal memory access)
and added a memcpy to a local array of 3 uint32_t's.
But on Metal it told me there is no memcpy, so I tried
accessing directly. There is no error, just garbage results.
At some point I did try accessing the scales with an uint16_t
pointer (the scales are for sure 2-byte aligned), but was
still getting garbage. I guess, there must have been another bug.
No access to scales is via a uint16_t pointer and, after starting
from scratch from the C dequantize function, it finally works.
* metal : Q3_K 1st optimization pass
* metal : Q3_K second optimization pass - 29.6 ms/token
* metal : Q3_K cleanup
* metal : fixed accidentally broken Q2_K
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-12 21:39:21 +02:00
|
|
|
}
|
2023-06-08 21:28:21 +02:00
|
|
|
|
2023-12-13 13:04:25 +01:00
|
|
|
[[host_name("kernel_mul_mv_q2_K_f32")]]
|
|
|
|
kernel void kernel_mul_mv_q2_K_f32(
|
Metal implementation for all k_quants (#1807)
* metal : improve q4_K
28.3 -> 26.0 ms/token by avoiding a branch in the
calculation of the scales.
* metal : small improvement for Q4_K
* metal : still optimizing Q4_K
This commit pushes it down to 25.3 ms / token.
The crazy idea of using 6 bits for the scales is really costly on
Metal: if I remove the bit fiddling necessary to make the block
scales, time goes almost to the Q4_0 23 ms/token.
Before pushing the k-quants upstream I had a Q4_K variant that
had used 8-bit scales. It wasn't more accurate, used 0.125 bits more per weight,
was running slightly slower on the CPU (due to the larger model size
and being memory bound there), and the difference was entirely
negligible under CUDA. So, I decided to publish the version with 6-bit
scales. Perhaps I should re-consider and change to 8-bit scales?
* metal : some more optimizations
Q2_K: 25.4 ms/token
Q6_K: 27.3 ms/token
Q4_0: 22.8 ms/token
Q4_1: 23.1 ms/token
* metal : Q3_K support
Something is not quite right yet.
* metal : Q5_K support
Initial version achieves 31.2 ms/token, 210 GB/s
* metal : still not able to figure out why q3_K does not work
* Minor
* metal : yet another failed attempt to make q3_K work
* metal : optimize Q5_K
31.2 ms -> 27.8 ms.
250 GB/s.
* metal : q3_K still not working
Adding a heavily commented q3_K metal kernel to explain
my obviously faulty logic. Perhaps someone could spot the issue?
* metal : q3_K finally working
Not optimized at all.
What was the issue? The scales are not 4-bytes aligned,
and I was accessing them with a uint32_t pointer.
When I tried that on CUDA, I got an error (illegal memory access)
and added a memcpy to a local array of 3 uint32_t's.
But on Metal it told me there is no memcpy, so I tried
accessing directly. There is no error, just garbage results.
At some point I did try accessing the scales with an uint16_t
pointer (the scales are for sure 2-byte aligned), but was
still getting garbage. I guess, there must have been another bug.
No access to scales is via a uint16_t pointer and, after starting
from scratch from the C dequantize function, it finally works.
* metal : Q3_K 1st optimization pass
* metal : Q3_K second optimization pass - 29.6 ms/token
* metal : Q3_K cleanup
* metal : fixed accidentally broken Q2_K
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-12 21:39:21 +02:00
|
|
|
device const void * src0,
|
|
|
|
device const float * src1,
|
|
|
|
device float * dst,
|
|
|
|
constant int64_t & ne00,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant uint64_t & nb00,
|
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant uint64_t & nb02,
|
|
|
|
constant int64_t & ne10,
|
|
|
|
constant int64_t & ne11,
|
|
|
|
constant int64_t & ne12,
|
|
|
|
constant uint64_t & nb10,
|
|
|
|
constant uint64_t & nb11,
|
|
|
|
constant uint64_t & nb12,
|
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
|
|
|
constant uint & r2,
|
|
|
|
constant uint & r3,
|
2023-08-16 22:07:04 +02:00
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
2023-12-13 13:04:25 +01:00
|
|
|
uint tiisg[[thread_index_in_simdgroup]],
|
|
|
|
uint sgitg[[simdgroup_index_in_threadgroup]]) {
|
|
|
|
|
|
|
|
kernel_mul_mv_q2_K_f32_impl(src0, src1, dst, ne00, ne01, ne02, ne10, ne12, ne0, ne1, r2, r3, tgpig, tiisg, sgitg);
|
|
|
|
}
|
|
|
|
|
|
|
|
#if QK_K == 256
|
|
|
|
void kernel_mul_mv_q3_K_f32_impl(
|
|
|
|
device const void * src0,
|
|
|
|
device const float * src1,
|
|
|
|
device float * dst,
|
|
|
|
constant int64_t & ne00,
|
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant int64_t & ne10,
|
|
|
|
constant int64_t & ne12,
|
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
|
|
|
constant uint & r2,
|
|
|
|
constant uint & r3,
|
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint tiisg[[thread_index_in_simdgroup]],
|
|
|
|
uint sgitg[[simdgroup_index_in_threadgroup]]) {
|
Metal implementation for all k_quants (#1807)
* metal : improve q4_K
28.3 -> 26.0 ms/token by avoiding a branch in the
calculation of the scales.
* metal : small improvement for Q4_K
* metal : still optimizing Q4_K
This commit pushes it down to 25.3 ms / token.
The crazy idea of using 6 bits for the scales is really costly on
Metal: if I remove the bit fiddling necessary to make the block
scales, time goes almost to the Q4_0 23 ms/token.
Before pushing the k-quants upstream I had a Q4_K variant that
had used 8-bit scales. It wasn't more accurate, used 0.125 bits more per weight,
was running slightly slower on the CPU (due to the larger model size
and being memory bound there), and the difference was entirely
negligible under CUDA. So, I decided to publish the version with 6-bit
scales. Perhaps I should re-consider and change to 8-bit scales?
* metal : some more optimizations
Q2_K: 25.4 ms/token
Q6_K: 27.3 ms/token
Q4_0: 22.8 ms/token
Q4_1: 23.1 ms/token
* metal : Q3_K support
Something is not quite right yet.
* metal : Q5_K support
Initial version achieves 31.2 ms/token, 210 GB/s
* metal : still not able to figure out why q3_K does not work
* Minor
* metal : yet another failed attempt to make q3_K work
* metal : optimize Q5_K
31.2 ms -> 27.8 ms.
250 GB/s.
* metal : q3_K still not working
Adding a heavily commented q3_K metal kernel to explain
my obviously faulty logic. Perhaps someone could spot the issue?
* metal : q3_K finally working
Not optimized at all.
What was the issue? The scales are not 4-bytes aligned,
and I was accessing them with a uint32_t pointer.
When I tried that on CUDA, I got an error (illegal memory access)
and added a memcpy to a local array of 3 uint32_t's.
But on Metal it told me there is no memcpy, so I tried
accessing directly. There is no error, just garbage results.
At some point I did try accessing the scales with an uint16_t
pointer (the scales are for sure 2-byte aligned), but was
still getting garbage. I guess, there must have been another bug.
No access to scales is via a uint16_t pointer and, after starting
from scratch from the C dequantize function, it finally works.
* metal : Q3_K 1st optimization pass
* metal : Q3_K second optimization pass - 29.6 ms/token
* metal : Q3_K cleanup
* metal : fixed accidentally broken Q2_K
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-12 21:39:21 +02:00
|
|
|
|
|
|
|
const int nb = ne00/QK_K;
|
|
|
|
|
|
|
|
const int64_t r0 = tgpig.x;
|
|
|
|
const int64_t r1 = tgpig.y;
|
2023-12-07 21:26:54 +01:00
|
|
|
const int64_t im = tgpig.z;
|
Metal implementation for all k_quants (#1807)
* metal : improve q4_K
28.3 -> 26.0 ms/token by avoiding a branch in the
calculation of the scales.
* metal : small improvement for Q4_K
* metal : still optimizing Q4_K
This commit pushes it down to 25.3 ms / token.
The crazy idea of using 6 bits for the scales is really costly on
Metal: if I remove the bit fiddling necessary to make the block
scales, time goes almost to the Q4_0 23 ms/token.
Before pushing the k-quants upstream I had a Q4_K variant that
had used 8-bit scales. It wasn't more accurate, used 0.125 bits more per weight,
was running slightly slower on the CPU (due to the larger model size
and being memory bound there), and the difference was entirely
negligible under CUDA. So, I decided to publish the version with 6-bit
scales. Perhaps I should re-consider and change to 8-bit scales?
* metal : some more optimizations
Q2_K: 25.4 ms/token
Q6_K: 27.3 ms/token
Q4_0: 22.8 ms/token
Q4_1: 23.1 ms/token
* metal : Q3_K support
Something is not quite right yet.
* metal : Q5_K support
Initial version achieves 31.2 ms/token, 210 GB/s
* metal : still not able to figure out why q3_K does not work
* Minor
* metal : yet another failed attempt to make q3_K work
* metal : optimize Q5_K
31.2 ms -> 27.8 ms.
250 GB/s.
* metal : q3_K still not working
Adding a heavily commented q3_K metal kernel to explain
my obviously faulty logic. Perhaps someone could spot the issue?
* metal : q3_K finally working
Not optimized at all.
What was the issue? The scales are not 4-bytes aligned,
and I was accessing them with a uint32_t pointer.
When I tried that on CUDA, I got an error (illegal memory access)
and added a memcpy to a local array of 3 uint32_t's.
But on Metal it told me there is no memcpy, so I tried
accessing directly. There is no error, just garbage results.
At some point I did try accessing the scales with an uint16_t
pointer (the scales are for sure 2-byte aligned), but was
still getting garbage. I guess, there must have been another bug.
No access to scales is via a uint16_t pointer and, after starting
from scratch from the C dequantize function, it finally works.
* metal : Q3_K 1st optimization pass
* metal : Q3_K second optimization pass - 29.6 ms/token
* metal : Q3_K cleanup
* metal : fixed accidentally broken Q2_K
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-12 21:39:21 +02:00
|
|
|
|
2023-07-21 16:05:30 +02:00
|
|
|
const int first_row = (r0 * N_SIMDGROUP + sgitg) * 2;
|
2023-12-07 21:26:54 +01:00
|
|
|
|
|
|
|
const uint i12 = im%ne12;
|
|
|
|
const uint i13 = im/ne12;
|
|
|
|
|
|
|
|
const uint offset0 = (i12/r2)*(nb*ne01) + (i13/r3)*(nb*ne01*ne02);
|
|
|
|
|
2023-08-16 22:07:04 +02:00
|
|
|
device const block_q3_K * x = (device const block_q3_K *) src0 + first_row*nb + offset0;
|
2023-12-07 21:26:54 +01:00
|
|
|
device const float * yy = (device const float *) src1 + r1*ne10 + im*ne00*ne1;
|
k-quants : support for super-block size of 64 (#2001)
* k_quants: WIP super-blocks with 64 weights
* k_quants: WIP super-blocks with 64 weights
Q6_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q4_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q2_K scalar and AVX2 works. Q2_K is way too slow (it is actually slower
than the scalar implementation)
* k_quants: WIP super-blocks with 64 weights
Q3_K scalar and AVX2 works.
* k_quants: WIP super-blocks with 64 weights
Q5_K scalar and AVX2 works, and with that all
k_quants are done on AVX2 and scalar
* k_quants: WIP super-blocks with 64 weights
Q6_K working on CUDA. Cannot make it run quite as gast as
with super-blocks with 256 weigths: 8% slower on 4080,
20% slower on the 1660 (but there we fit 1 less layer on the
GPU because pf the larger model size), so some fraction of
these 20% is due to that,
* k_quants: WIP super-blocks with 64 weights
Q4_K working on CUDA. ~10% slower on GTX-1660,
16% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q2_K working on CUDA. ~3% slower on GTX-1660,
10% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q3_K working on CUDA.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on CUDA, and with this CUDA is done.
* k_quants: WIP super-blocks with 64 weights
Q6_K working on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Q4_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q2_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q3_K working on ARM_NEON, but quite a bit slower than 256 weights.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on ARM_NEON, but quite a bit slower than 256 weights.
With that, we have full support for ARM_NEON, although
performance is not quite there.
* k_quants: WIP super-blocks with 64 weights
Slightly more efficient Q3_K and Q5_K
* k_quants: WIP super-blocks with 64 weights
Another small improvement for Q3_K and Q5_K on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Yet another speedup for Q5_K on ARM_NEON.
We are now within 10% of the QK_K = 256 version.
* k_quants: WIP super-blocks with 64 weights
* We are able to pass preprocessor macros to the Metal
compiler
* Q6_K works and is actually slightly more efficient than
the QK_K = 256 version (25.2 ms vs 25.8 ms)
* k_quants: WIP super-blocks with 64 weights
Q4_K works on Metal and is actually slightly faster
than QK_K = 256 (21.95 ms vs 24.0 ms).
* k_quants: WIP super-blocks with 64 weights
Q2_K works on Metal and is very slightly faster
than QK_K = 256 (23.8 ms vs 24.2 ms).
* k_quants: WIP super-blocks with 64 weights
Q3_K works on Metal and is slightly faster
than QK_K = 256 (26.6 ms vs 28.3 ms).
* k_quants: WIP super-blocks with 64 weights
Q5_K works on Metal and is slightly faster
than QK_K = 256 (23.7 ms vs 26.3 ms).
* k_quants: call them _K, not _k, also on Metal
* k_quants: correctly define QK_K in llama.cpp
* Fixed bug in q4_K quantization added with the 64-block addition
* Simplify via lambda
* k_quants: swicth Q3_K to 4-bit scales when QK_K = 64
Otherwise there isn't much benefit from this
quantization type. There is some very slight loss
in accuracy, but we reduce size by ~7%.
E.g., for OpenLLaMA-3B, Q3_K_S perplexity is
8.6131 with 8-bit scales and 8.6352 with 4-bit,
while file size decreases from 1.53G to 1.44G.
* k_quants: switch Q4_K to 4-bit scales when QK_K = 64
Here the loss in accuracy is greater than for Q3_K,
but the Q4_K points still move further to the left on
the perplexity vs size curve.
* k_quants: forgot to add the Metal changes in last commit
* k_quants: change Q5_K to be type 0 when QK_K = 64
Still needs AVX2 implementation
* k_quants: AVX2 implementation for new 64-weight Q5_K
* k_quants: 10% faster ARM_NEON Q5_K dot product
* k_quants: fixed issue caused by merging with master
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-26 18:43:07 +02:00
|
|
|
|
2023-09-08 18:01:04 +02:00
|
|
|
float yl[32];
|
k-quants : support for super-block size of 64 (#2001)
* k_quants: WIP super-blocks with 64 weights
* k_quants: WIP super-blocks with 64 weights
Q6_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q4_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q2_K scalar and AVX2 works. Q2_K is way too slow (it is actually slower
than the scalar implementation)
* k_quants: WIP super-blocks with 64 weights
Q3_K scalar and AVX2 works.
* k_quants: WIP super-blocks with 64 weights
Q5_K scalar and AVX2 works, and with that all
k_quants are done on AVX2 and scalar
* k_quants: WIP super-blocks with 64 weights
Q6_K working on CUDA. Cannot make it run quite as gast as
with super-blocks with 256 weigths: 8% slower on 4080,
20% slower on the 1660 (but there we fit 1 less layer on the
GPU because pf the larger model size), so some fraction of
these 20% is due to that,
* k_quants: WIP super-blocks with 64 weights
Q4_K working on CUDA. ~10% slower on GTX-1660,
16% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q2_K working on CUDA. ~3% slower on GTX-1660,
10% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q3_K working on CUDA.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on CUDA, and with this CUDA is done.
* k_quants: WIP super-blocks with 64 weights
Q6_K working on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Q4_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q2_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q3_K working on ARM_NEON, but quite a bit slower than 256 weights.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on ARM_NEON, but quite a bit slower than 256 weights.
With that, we have full support for ARM_NEON, although
performance is not quite there.
* k_quants: WIP super-blocks with 64 weights
Slightly more efficient Q3_K and Q5_K
* k_quants: WIP super-blocks with 64 weights
Another small improvement for Q3_K and Q5_K on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Yet another speedup for Q5_K on ARM_NEON.
We are now within 10% of the QK_K = 256 version.
* k_quants: WIP super-blocks with 64 weights
* We are able to pass preprocessor macros to the Metal
compiler
* Q6_K works and is actually slightly more efficient than
the QK_K = 256 version (25.2 ms vs 25.8 ms)
* k_quants: WIP super-blocks with 64 weights
Q4_K works on Metal and is actually slightly faster
than QK_K = 256 (21.95 ms vs 24.0 ms).
* k_quants: WIP super-blocks with 64 weights
Q2_K works on Metal and is very slightly faster
than QK_K = 256 (23.8 ms vs 24.2 ms).
* k_quants: WIP super-blocks with 64 weights
Q3_K works on Metal and is slightly faster
than QK_K = 256 (26.6 ms vs 28.3 ms).
* k_quants: WIP super-blocks with 64 weights
Q5_K works on Metal and is slightly faster
than QK_K = 256 (23.7 ms vs 26.3 ms).
* k_quants: call them _K, not _k, also on Metal
* k_quants: correctly define QK_K in llama.cpp
* Fixed bug in q4_K quantization added with the 64-block addition
* Simplify via lambda
* k_quants: swicth Q3_K to 4-bit scales when QK_K = 64
Otherwise there isn't much benefit from this
quantization type. There is some very slight loss
in accuracy, but we reduce size by ~7%.
E.g., for OpenLLaMA-3B, Q3_K_S perplexity is
8.6131 with 8-bit scales and 8.6352 with 4-bit,
while file size decreases from 1.53G to 1.44G.
* k_quants: switch Q4_K to 4-bit scales when QK_K = 64
Here the loss in accuracy is greater than for Q3_K,
but the Q4_K points still move further to the left on
the perplexity vs size curve.
* k_quants: forgot to add the Metal changes in last commit
* k_quants: change Q5_K to be type 0 when QK_K = 64
Still needs AVX2 implementation
* k_quants: AVX2 implementation for new 64-weight Q5_K
* k_quants: 10% faster ARM_NEON Q5_K dot product
* k_quants: fixed issue caused by merging with master
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-26 18:43:07 +02:00
|
|
|
|
2023-09-28 18:04:36 +02:00
|
|
|
//const uint16_t kmask1 = 0x3030;
|
|
|
|
//const uint16_t kmask2 = 0x0f0f;
|
k-quants : support for super-block size of 64 (#2001)
* k_quants: WIP super-blocks with 64 weights
* k_quants: WIP super-blocks with 64 weights
Q6_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q4_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q2_K scalar and AVX2 works. Q2_K is way too slow (it is actually slower
than the scalar implementation)
* k_quants: WIP super-blocks with 64 weights
Q3_K scalar and AVX2 works.
* k_quants: WIP super-blocks with 64 weights
Q5_K scalar and AVX2 works, and with that all
k_quants are done on AVX2 and scalar
* k_quants: WIP super-blocks with 64 weights
Q6_K working on CUDA. Cannot make it run quite as gast as
with super-blocks with 256 weigths: 8% slower on 4080,
20% slower on the 1660 (but there we fit 1 less layer on the
GPU because pf the larger model size), so some fraction of
these 20% is due to that,
* k_quants: WIP super-blocks with 64 weights
Q4_K working on CUDA. ~10% slower on GTX-1660,
16% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q2_K working on CUDA. ~3% slower on GTX-1660,
10% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q3_K working on CUDA.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on CUDA, and with this CUDA is done.
* k_quants: WIP super-blocks with 64 weights
Q6_K working on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Q4_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q2_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q3_K working on ARM_NEON, but quite a bit slower than 256 weights.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on ARM_NEON, but quite a bit slower than 256 weights.
With that, we have full support for ARM_NEON, although
performance is not quite there.
* k_quants: WIP super-blocks with 64 weights
Slightly more efficient Q3_K and Q5_K
* k_quants: WIP super-blocks with 64 weights
Another small improvement for Q3_K and Q5_K on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Yet another speedup for Q5_K on ARM_NEON.
We are now within 10% of the QK_K = 256 version.
* k_quants: WIP super-blocks with 64 weights
* We are able to pass preprocessor macros to the Metal
compiler
* Q6_K works and is actually slightly more efficient than
the QK_K = 256 version (25.2 ms vs 25.8 ms)
* k_quants: WIP super-blocks with 64 weights
Q4_K works on Metal and is actually slightly faster
than QK_K = 256 (21.95 ms vs 24.0 ms).
* k_quants: WIP super-blocks with 64 weights
Q2_K works on Metal and is very slightly faster
than QK_K = 256 (23.8 ms vs 24.2 ms).
* k_quants: WIP super-blocks with 64 weights
Q3_K works on Metal and is slightly faster
than QK_K = 256 (26.6 ms vs 28.3 ms).
* k_quants: WIP super-blocks with 64 weights
Q5_K works on Metal and is slightly faster
than QK_K = 256 (23.7 ms vs 26.3 ms).
* k_quants: call them _K, not _k, also on Metal
* k_quants: correctly define QK_K in llama.cpp
* Fixed bug in q4_K quantization added with the 64-block addition
* Simplify via lambda
* k_quants: swicth Q3_K to 4-bit scales when QK_K = 64
Otherwise there isn't much benefit from this
quantization type. There is some very slight loss
in accuracy, but we reduce size by ~7%.
E.g., for OpenLLaMA-3B, Q3_K_S perplexity is
8.6131 with 8-bit scales and 8.6352 with 4-bit,
while file size decreases from 1.53G to 1.44G.
* k_quants: switch Q4_K to 4-bit scales when QK_K = 64
Here the loss in accuracy is greater than for Q3_K,
but the Q4_K points still move further to the left on
the perplexity vs size curve.
* k_quants: forgot to add the Metal changes in last commit
* k_quants: change Q5_K to be type 0 when QK_K = 64
Still needs AVX2 implementation
* k_quants: AVX2 implementation for new 64-weight Q5_K
* k_quants: 10% faster ARM_NEON Q5_K dot product
* k_quants: fixed issue caused by merging with master
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-26 18:43:07 +02:00
|
|
|
|
2023-09-08 18:01:04 +02:00
|
|
|
const int tid = tiisg/4;
|
|
|
|
const int ix = tiisg%4;
|
|
|
|
const int ip = tid/4; // 0 or 1
|
|
|
|
const int il = 2*((tid%4)/2); // 0 or 2
|
Metal implementation for all k_quants (#1807)
* metal : improve q4_K
28.3 -> 26.0 ms/token by avoiding a branch in the
calculation of the scales.
* metal : small improvement for Q4_K
* metal : still optimizing Q4_K
This commit pushes it down to 25.3 ms / token.
The crazy idea of using 6 bits for the scales is really costly on
Metal: if I remove the bit fiddling necessary to make the block
scales, time goes almost to the Q4_0 23 ms/token.
Before pushing the k-quants upstream I had a Q4_K variant that
had used 8-bit scales. It wasn't more accurate, used 0.125 bits more per weight,
was running slightly slower on the CPU (due to the larger model size
and being memory bound there), and the difference was entirely
negligible under CUDA. So, I decided to publish the version with 6-bit
scales. Perhaps I should re-consider and change to 8-bit scales?
* metal : some more optimizations
Q2_K: 25.4 ms/token
Q6_K: 27.3 ms/token
Q4_0: 22.8 ms/token
Q4_1: 23.1 ms/token
* metal : Q3_K support
Something is not quite right yet.
* metal : Q5_K support
Initial version achieves 31.2 ms/token, 210 GB/s
* metal : still not able to figure out why q3_K does not work
* Minor
* metal : yet another failed attempt to make q3_K work
* metal : optimize Q5_K
31.2 ms -> 27.8 ms.
250 GB/s.
* metal : q3_K still not working
Adding a heavily commented q3_K metal kernel to explain
my obviously faulty logic. Perhaps someone could spot the issue?
* metal : q3_K finally working
Not optimized at all.
What was the issue? The scales are not 4-bytes aligned,
and I was accessing them with a uint32_t pointer.
When I tried that on CUDA, I got an error (illegal memory access)
and added a memcpy to a local array of 3 uint32_t's.
But on Metal it told me there is no memcpy, so I tried
accessing directly. There is no error, just garbage results.
At some point I did try accessing the scales with an uint16_t
pointer (the scales are for sure 2-byte aligned), but was
still getting garbage. I guess, there must have been another bug.
No access to scales is via a uint16_t pointer and, after starting
from scratch from the C dequantize function, it finally works.
* metal : Q3_K 1st optimization pass
* metal : Q3_K second optimization pass - 29.6 ms/token
* metal : Q3_K cleanup
* metal : fixed accidentally broken Q2_K
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-12 21:39:21 +02:00
|
|
|
const int ir = tid%2;
|
|
|
|
const int n = 8;
|
|
|
|
const int l0 = n*ir;
|
|
|
|
|
2023-09-08 18:01:04 +02:00
|
|
|
// One would think that the Metal compiler would figure out that ip and il can only have
|
|
|
|
// 4 possible states, and optimize accordingly. Well, no. It needs help, and we do it
|
|
|
|
// with these two tales.
|
|
|
|
//
|
|
|
|
// Possible masks for the high bit
|
|
|
|
const ushort4 mm[4] = {{0x0001, 0x0100, 0x0002, 0x0200}, // ip = 0, il = 0
|
|
|
|
{0x0004, 0x0400, 0x0008, 0x0800}, // ip = 0, il = 2
|
|
|
|
{0x0010, 0x1000, 0x0020, 0x2000}, // ip = 1, il = 0
|
|
|
|
{0x0040, 0x4000, 0x0080, 0x8000}}; // ip = 1, il = 2
|
|
|
|
|
|
|
|
// Possible masks for the low 2 bits
|
|
|
|
const int4 qm[2] = {{0x0003, 0x0300, 0x000c, 0x0c00}, {0x0030, 0x3000, 0x00c0, 0xc000}};
|
|
|
|
|
|
|
|
const ushort4 hm = mm[2*ip + il/2];
|
Metal implementation for all k_quants (#1807)
* metal : improve q4_K
28.3 -> 26.0 ms/token by avoiding a branch in the
calculation of the scales.
* metal : small improvement for Q4_K
* metal : still optimizing Q4_K
This commit pushes it down to 25.3 ms / token.
The crazy idea of using 6 bits for the scales is really costly on
Metal: if I remove the bit fiddling necessary to make the block
scales, time goes almost to the Q4_0 23 ms/token.
Before pushing the k-quants upstream I had a Q4_K variant that
had used 8-bit scales. It wasn't more accurate, used 0.125 bits more per weight,
was running slightly slower on the CPU (due to the larger model size
and being memory bound there), and the difference was entirely
negligible under CUDA. So, I decided to publish the version with 6-bit
scales. Perhaps I should re-consider and change to 8-bit scales?
* metal : some more optimizations
Q2_K: 25.4 ms/token
Q6_K: 27.3 ms/token
Q4_0: 22.8 ms/token
Q4_1: 23.1 ms/token
* metal : Q3_K support
Something is not quite right yet.
* metal : Q5_K support
Initial version achieves 31.2 ms/token, 210 GB/s
* metal : still not able to figure out why q3_K does not work
* Minor
* metal : yet another failed attempt to make q3_K work
* metal : optimize Q5_K
31.2 ms -> 27.8 ms.
250 GB/s.
* metal : q3_K still not working
Adding a heavily commented q3_K metal kernel to explain
my obviously faulty logic. Perhaps someone could spot the issue?
* metal : q3_K finally working
Not optimized at all.
What was the issue? The scales are not 4-bytes aligned,
and I was accessing them with a uint32_t pointer.
When I tried that on CUDA, I got an error (illegal memory access)
and added a memcpy to a local array of 3 uint32_t's.
But on Metal it told me there is no memcpy, so I tried
accessing directly. There is no error, just garbage results.
At some point I did try accessing the scales with an uint16_t
pointer (the scales are for sure 2-byte aligned), but was
still getting garbage. I guess, there must have been another bug.
No access to scales is via a uint16_t pointer and, after starting
from scratch from the C dequantize function, it finally works.
* metal : Q3_K 1st optimization pass
* metal : Q3_K second optimization pass - 29.6 ms/token
* metal : Q3_K cleanup
* metal : fixed accidentally broken Q2_K
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-12 21:39:21 +02:00
|
|
|
|
|
|
|
const int shift = 2*il;
|
2023-09-08 18:01:04 +02:00
|
|
|
const float v1 = il == 0 ? 4.f : 64.f;
|
|
|
|
const float v2 = 4.f * v1;
|
Metal implementation for all k_quants (#1807)
* metal : improve q4_K
28.3 -> 26.0 ms/token by avoiding a branch in the
calculation of the scales.
* metal : small improvement for Q4_K
* metal : still optimizing Q4_K
This commit pushes it down to 25.3 ms / token.
The crazy idea of using 6 bits for the scales is really costly on
Metal: if I remove the bit fiddling necessary to make the block
scales, time goes almost to the Q4_0 23 ms/token.
Before pushing the k-quants upstream I had a Q4_K variant that
had used 8-bit scales. It wasn't more accurate, used 0.125 bits more per weight,
was running slightly slower on the CPU (due to the larger model size
and being memory bound there), and the difference was entirely
negligible under CUDA. So, I decided to publish the version with 6-bit
scales. Perhaps I should re-consider and change to 8-bit scales?
* metal : some more optimizations
Q2_K: 25.4 ms/token
Q6_K: 27.3 ms/token
Q4_0: 22.8 ms/token
Q4_1: 23.1 ms/token
* metal : Q3_K support
Something is not quite right yet.
* metal : Q5_K support
Initial version achieves 31.2 ms/token, 210 GB/s
* metal : still not able to figure out why q3_K does not work
* Minor
* metal : yet another failed attempt to make q3_K work
* metal : optimize Q5_K
31.2 ms -> 27.8 ms.
250 GB/s.
* metal : q3_K still not working
Adding a heavily commented q3_K metal kernel to explain
my obviously faulty logic. Perhaps someone could spot the issue?
* metal : q3_K finally working
Not optimized at all.
What was the issue? The scales are not 4-bytes aligned,
and I was accessing them with a uint32_t pointer.
When I tried that on CUDA, I got an error (illegal memory access)
and added a memcpy to a local array of 3 uint32_t's.
But on Metal it told me there is no memcpy, so I tried
accessing directly. There is no error, just garbage results.
At some point I did try accessing the scales with an uint16_t
pointer (the scales are for sure 2-byte aligned), but was
still getting garbage. I guess, there must have been another bug.
No access to scales is via a uint16_t pointer and, after starting
from scratch from the C dequantize function, it finally works.
* metal : Q3_K 1st optimization pass
* metal : Q3_K second optimization pass - 29.6 ms/token
* metal : Q3_K cleanup
* metal : fixed accidentally broken Q2_K
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-12 21:39:21 +02:00
|
|
|
|
|
|
|
const uint16_t s_shift1 = 4*ip;
|
2023-09-08 18:01:04 +02:00
|
|
|
const uint16_t s_shift2 = s_shift1 + il;
|
Metal implementation for all k_quants (#1807)
* metal : improve q4_K
28.3 -> 26.0 ms/token by avoiding a branch in the
calculation of the scales.
* metal : small improvement for Q4_K
* metal : still optimizing Q4_K
This commit pushes it down to 25.3 ms / token.
The crazy idea of using 6 bits for the scales is really costly on
Metal: if I remove the bit fiddling necessary to make the block
scales, time goes almost to the Q4_0 23 ms/token.
Before pushing the k-quants upstream I had a Q4_K variant that
had used 8-bit scales. It wasn't more accurate, used 0.125 bits more per weight,
was running slightly slower on the CPU (due to the larger model size
and being memory bound there), and the difference was entirely
negligible under CUDA. So, I decided to publish the version with 6-bit
scales. Perhaps I should re-consider and change to 8-bit scales?
* metal : some more optimizations
Q2_K: 25.4 ms/token
Q6_K: 27.3 ms/token
Q4_0: 22.8 ms/token
Q4_1: 23.1 ms/token
* metal : Q3_K support
Something is not quite right yet.
* metal : Q5_K support
Initial version achieves 31.2 ms/token, 210 GB/s
* metal : still not able to figure out why q3_K does not work
* Minor
* metal : yet another failed attempt to make q3_K work
* metal : optimize Q5_K
31.2 ms -> 27.8 ms.
250 GB/s.
* metal : q3_K still not working
Adding a heavily commented q3_K metal kernel to explain
my obviously faulty logic. Perhaps someone could spot the issue?
* metal : q3_K finally working
Not optimized at all.
What was the issue? The scales are not 4-bytes aligned,
and I was accessing them with a uint32_t pointer.
When I tried that on CUDA, I got an error (illegal memory access)
and added a memcpy to a local array of 3 uint32_t's.
But on Metal it told me there is no memcpy, so I tried
accessing directly. There is no error, just garbage results.
At some point I did try accessing the scales with an uint16_t
pointer (the scales are for sure 2-byte aligned), but was
still getting garbage. I guess, there must have been another bug.
No access to scales is via a uint16_t pointer and, after starting
from scratch from the C dequantize function, it finally works.
* metal : Q3_K 1st optimization pass
* metal : Q3_K second optimization pass - 29.6 ms/token
* metal : Q3_K cleanup
* metal : fixed accidentally broken Q2_K
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-12 21:39:21 +02:00
|
|
|
|
|
|
|
const int q_offset = 32*ip + l0;
|
|
|
|
const int y_offset = 128*ip + 32*il + l0;
|
|
|
|
|
2023-07-21 16:05:30 +02:00
|
|
|
const int step = sizeof(block_q3_K) * nb / 2;
|
Metal implementation for all k_quants (#1807)
* metal : improve q4_K
28.3 -> 26.0 ms/token by avoiding a branch in the
calculation of the scales.
* metal : small improvement for Q4_K
* metal : still optimizing Q4_K
This commit pushes it down to 25.3 ms / token.
The crazy idea of using 6 bits for the scales is really costly on
Metal: if I remove the bit fiddling necessary to make the block
scales, time goes almost to the Q4_0 23 ms/token.
Before pushing the k-quants upstream I had a Q4_K variant that
had used 8-bit scales. It wasn't more accurate, used 0.125 bits more per weight,
was running slightly slower on the CPU (due to the larger model size
and being memory bound there), and the difference was entirely
negligible under CUDA. So, I decided to publish the version with 6-bit
scales. Perhaps I should re-consider and change to 8-bit scales?
* metal : some more optimizations
Q2_K: 25.4 ms/token
Q6_K: 27.3 ms/token
Q4_0: 22.8 ms/token
Q4_1: 23.1 ms/token
* metal : Q3_K support
Something is not quite right yet.
* metal : Q5_K support
Initial version achieves 31.2 ms/token, 210 GB/s
* metal : still not able to figure out why q3_K does not work
* Minor
* metal : yet another failed attempt to make q3_K work
* metal : optimize Q5_K
31.2 ms -> 27.8 ms.
250 GB/s.
* metal : q3_K still not working
Adding a heavily commented q3_K metal kernel to explain
my obviously faulty logic. Perhaps someone could spot the issue?
* metal : q3_K finally working
Not optimized at all.
What was the issue? The scales are not 4-bytes aligned,
and I was accessing them with a uint32_t pointer.
When I tried that on CUDA, I got an error (illegal memory access)
and added a memcpy to a local array of 3 uint32_t's.
But on Metal it told me there is no memcpy, so I tried
accessing directly. There is no error, just garbage results.
At some point I did try accessing the scales with an uint16_t
pointer (the scales are for sure 2-byte aligned), but was
still getting garbage. I guess, there must have been another bug.
No access to scales is via a uint16_t pointer and, after starting
from scratch from the C dequantize function, it finally works.
* metal : Q3_K 1st optimization pass
* metal : Q3_K second optimization pass - 29.6 ms/token
* metal : Q3_K cleanup
* metal : fixed accidentally broken Q2_K
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-12 21:39:21 +02:00
|
|
|
|
2023-07-21 16:05:30 +02:00
|
|
|
device const float * y1 = yy + ix*QK_K + y_offset;
|
Metal implementation for all k_quants (#1807)
* metal : improve q4_K
28.3 -> 26.0 ms/token by avoiding a branch in the
calculation of the scales.
* metal : small improvement for Q4_K
* metal : still optimizing Q4_K
This commit pushes it down to 25.3 ms / token.
The crazy idea of using 6 bits for the scales is really costly on
Metal: if I remove the bit fiddling necessary to make the block
scales, time goes almost to the Q4_0 23 ms/token.
Before pushing the k-quants upstream I had a Q4_K variant that
had used 8-bit scales. It wasn't more accurate, used 0.125 bits more per weight,
was running slightly slower on the CPU (due to the larger model size
and being memory bound there), and the difference was entirely
negligible under CUDA. So, I decided to publish the version with 6-bit
scales. Perhaps I should re-consider and change to 8-bit scales?
* metal : some more optimizations
Q2_K: 25.4 ms/token
Q6_K: 27.3 ms/token
Q4_0: 22.8 ms/token
Q4_1: 23.1 ms/token
* metal : Q3_K support
Something is not quite right yet.
* metal : Q5_K support
Initial version achieves 31.2 ms/token, 210 GB/s
* metal : still not able to figure out why q3_K does not work
* Minor
* metal : yet another failed attempt to make q3_K work
* metal : optimize Q5_K
31.2 ms -> 27.8 ms.
250 GB/s.
* metal : q3_K still not working
Adding a heavily commented q3_K metal kernel to explain
my obviously faulty logic. Perhaps someone could spot the issue?
* metal : q3_K finally working
Not optimized at all.
What was the issue? The scales are not 4-bytes aligned,
and I was accessing them with a uint32_t pointer.
When I tried that on CUDA, I got an error (illegal memory access)
and added a memcpy to a local array of 3 uint32_t's.
But on Metal it told me there is no memcpy, so I tried
accessing directly. There is no error, just garbage results.
At some point I did try accessing the scales with an uint16_t
pointer (the scales are for sure 2-byte aligned), but was
still getting garbage. I guess, there must have been another bug.
No access to scales is via a uint16_t pointer and, after starting
from scratch from the C dequantize function, it finally works.
* metal : Q3_K 1st optimization pass
* metal : Q3_K second optimization pass - 29.6 ms/token
* metal : Q3_K cleanup
* metal : fixed accidentally broken Q2_K
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-12 21:39:21 +02:00
|
|
|
|
2023-09-08 18:01:04 +02:00
|
|
|
uint32_t scales32, aux32;
|
|
|
|
thread uint16_t * scales16 = (thread uint16_t *)&scales32;
|
|
|
|
thread const int8_t * scales = (thread const int8_t *)&scales32;
|
|
|
|
|
|
|
|
float sumf1[2] = {0.f};
|
|
|
|
float sumf2[2] = {0.f};
|
|
|
|
for (int i = ix; i < nb; i += 4) {
|
Metal implementation for all k_quants (#1807)
* metal : improve q4_K
28.3 -> 26.0 ms/token by avoiding a branch in the
calculation of the scales.
* metal : small improvement for Q4_K
* metal : still optimizing Q4_K
This commit pushes it down to 25.3 ms / token.
The crazy idea of using 6 bits for the scales is really costly on
Metal: if I remove the bit fiddling necessary to make the block
scales, time goes almost to the Q4_0 23 ms/token.
Before pushing the k-quants upstream I had a Q4_K variant that
had used 8-bit scales. It wasn't more accurate, used 0.125 bits more per weight,
was running slightly slower on the CPU (due to the larger model size
and being memory bound there), and the difference was entirely
negligible under CUDA. So, I decided to publish the version with 6-bit
scales. Perhaps I should re-consider and change to 8-bit scales?
* metal : some more optimizations
Q2_K: 25.4 ms/token
Q6_K: 27.3 ms/token
Q4_0: 22.8 ms/token
Q4_1: 23.1 ms/token
* metal : Q3_K support
Something is not quite right yet.
* metal : Q5_K support
Initial version achieves 31.2 ms/token, 210 GB/s
* metal : still not able to figure out why q3_K does not work
* Minor
* metal : yet another failed attempt to make q3_K work
* metal : optimize Q5_K
31.2 ms -> 27.8 ms.
250 GB/s.
* metal : q3_K still not working
Adding a heavily commented q3_K metal kernel to explain
my obviously faulty logic. Perhaps someone could spot the issue?
* metal : q3_K finally working
Not optimized at all.
What was the issue? The scales are not 4-bytes aligned,
and I was accessing them with a uint32_t pointer.
When I tried that on CUDA, I got an error (illegal memory access)
and added a memcpy to a local array of 3 uint32_t's.
But on Metal it told me there is no memcpy, so I tried
accessing directly. There is no error, just garbage results.
At some point I did try accessing the scales with an uint16_t
pointer (the scales are for sure 2-byte aligned), but was
still getting garbage. I guess, there must have been another bug.
No access to scales is via a uint16_t pointer and, after starting
from scratch from the C dequantize function, it finally works.
* metal : Q3_K 1st optimization pass
* metal : Q3_K second optimization pass - 29.6 ms/token
* metal : Q3_K cleanup
* metal : fixed accidentally broken Q2_K
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-12 21:39:21 +02:00
|
|
|
|
2023-07-21 16:05:30 +02:00
|
|
|
for (int l = 0; l < 8; ++l) {
|
2023-09-08 18:01:04 +02:00
|
|
|
yl[l+ 0] = y1[l+ 0];
|
|
|
|
yl[l+ 8] = y1[l+16];
|
|
|
|
yl[l+16] = y1[l+32];
|
|
|
|
yl[l+24] = y1[l+48];
|
Metal implementation for all k_quants (#1807)
* metal : improve q4_K
28.3 -> 26.0 ms/token by avoiding a branch in the
calculation of the scales.
* metal : small improvement for Q4_K
* metal : still optimizing Q4_K
This commit pushes it down to 25.3 ms / token.
The crazy idea of using 6 bits for the scales is really costly on
Metal: if I remove the bit fiddling necessary to make the block
scales, time goes almost to the Q4_0 23 ms/token.
Before pushing the k-quants upstream I had a Q4_K variant that
had used 8-bit scales. It wasn't more accurate, used 0.125 bits more per weight,
was running slightly slower on the CPU (due to the larger model size
and being memory bound there), and the difference was entirely
negligible under CUDA. So, I decided to publish the version with 6-bit
scales. Perhaps I should re-consider and change to 8-bit scales?
* metal : some more optimizations
Q2_K: 25.4 ms/token
Q6_K: 27.3 ms/token
Q4_0: 22.8 ms/token
Q4_1: 23.1 ms/token
* metal : Q3_K support
Something is not quite right yet.
* metal : Q5_K support
Initial version achieves 31.2 ms/token, 210 GB/s
* metal : still not able to figure out why q3_K does not work
* Minor
* metal : yet another failed attempt to make q3_K work
* metal : optimize Q5_K
31.2 ms -> 27.8 ms.
250 GB/s.
* metal : q3_K still not working
Adding a heavily commented q3_K metal kernel to explain
my obviously faulty logic. Perhaps someone could spot the issue?
* metal : q3_K finally working
Not optimized at all.
What was the issue? The scales are not 4-bytes aligned,
and I was accessing them with a uint32_t pointer.
When I tried that on CUDA, I got an error (illegal memory access)
and added a memcpy to a local array of 3 uint32_t's.
But on Metal it told me there is no memcpy, so I tried
accessing directly. There is no error, just garbage results.
At some point I did try accessing the scales with an uint16_t
pointer (the scales are for sure 2-byte aligned), but was
still getting garbage. I guess, there must have been another bug.
No access to scales is via a uint16_t pointer and, after starting
from scratch from the C dequantize function, it finally works.
* metal : Q3_K 1st optimization pass
* metal : Q3_K second optimization pass - 29.6 ms/token
* metal : Q3_K cleanup
* metal : fixed accidentally broken Q2_K
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-12 21:39:21 +02:00
|
|
|
}
|
2023-07-21 16:05:30 +02:00
|
|
|
|
|
|
|
device const uint16_t * q = (device const uint16_t *)(x[i].qs + q_offset);
|
|
|
|
device const uint16_t * h = (device const uint16_t *)(x[i].hmask + l0);
|
|
|
|
device const uint16_t * a = (device const uint16_t *)(x[i].scales);
|
|
|
|
device const half * dh = &x[i].d;
|
|
|
|
|
|
|
|
for (int row = 0; row < 2; ++row) {
|
|
|
|
|
|
|
|
const float d_all = (float)dh[0];
|
|
|
|
|
2023-09-08 18:01:04 +02:00
|
|
|
scales16[0] = a[4];
|
|
|
|
scales16[1] = a[5];
|
|
|
|
aux32 = ((scales32 >> s_shift2) << 4) & 0x30303030;
|
|
|
|
scales16[0] = a[il+0];
|
|
|
|
scales16[1] = a[il+1];
|
|
|
|
scales32 = ((scales32 >> s_shift1) & 0x0f0f0f0f) | aux32;
|
|
|
|
|
|
|
|
float s1 = 0, s2 = 0, s3 = 0, s4 = 0, s5 = 0, s6 = 0;
|
2023-07-21 16:05:30 +02:00
|
|
|
for (int l = 0; l < n; l += 2) {
|
2023-09-08 18:01:04 +02:00
|
|
|
const int32_t qs = q[l/2];
|
|
|
|
s1 += yl[l+0] * (qs & qm[il/2][0]);
|
|
|
|
s2 += yl[l+1] * (qs & qm[il/2][1]);
|
|
|
|
s3 += ((h[l/2] & hm[0]) ? 0.f : yl[l+0]) + ((h[l/2] & hm[1]) ? 0.f : yl[l+1]);
|
|
|
|
s4 += yl[l+16] * (qs & qm[il/2][2]);
|
|
|
|
s5 += yl[l+17] * (qs & qm[il/2][3]);
|
|
|
|
s6 += ((h[l/2] & hm[2]) ? 0.f : yl[l+16]) + ((h[l/2] & hm[3]) ? 0.f : yl[l+17]);
|
2023-07-21 16:05:30 +02:00
|
|
|
}
|
2023-09-08 18:01:04 +02:00
|
|
|
float d1 = d_all * (s1 + 1.f/256.f * s2 - s3*v1);
|
|
|
|
float d2 = d_all * (s4 + 1.f/256.f * s5 - s6*v2);
|
|
|
|
sumf1[row] += d1 * (scales[0] - 32);
|
|
|
|
sumf2[row] += d2 * (scales[2] - 32);
|
2023-07-21 16:05:30 +02:00
|
|
|
|
2023-09-08 18:01:04 +02:00
|
|
|
s1 = s2 = s3 = s4 = s5 = s6 = 0;
|
2023-07-21 16:05:30 +02:00
|
|
|
for (int l = 0; l < n; l += 2) {
|
2023-09-08 18:01:04 +02:00
|
|
|
const int32_t qs = q[l/2+8];
|
|
|
|
s1 += yl[l+8] * (qs & qm[il/2][0]);
|
|
|
|
s2 += yl[l+9] * (qs & qm[il/2][1]);
|
|
|
|
s3 += ((h[l/2+8] & hm[0]) ? 0.f : yl[l+8]) + ((h[l/2+8] & hm[1]) ? 0.f : yl[l+9]);
|
|
|
|
s4 += yl[l+24] * (qs & qm[il/2][2]);
|
|
|
|
s5 += yl[l+25] * (qs & qm[il/2][3]);
|
|
|
|
s6 += ((h[l/2+8] & hm[2]) ? 0.f : yl[l+24]) + ((h[l/2+8] & hm[3]) ? 0.f : yl[l+25]);
|
2023-07-21 16:05:30 +02:00
|
|
|
}
|
2023-09-08 18:01:04 +02:00
|
|
|
d1 = d_all * (s1 + 1.f/256.f * s2 - s3*v1);
|
|
|
|
d2 = d_all * (s4 + 1.f/256.f * s5 - s6*v2);
|
|
|
|
sumf1[row] += d1 * (scales[1] - 32);
|
|
|
|
sumf2[row] += d2 * (scales[3] - 32);
|
2023-07-21 16:05:30 +02:00
|
|
|
|
|
|
|
q += step;
|
|
|
|
h += step;
|
|
|
|
a += step;
|
|
|
|
dh += step;
|
|
|
|
|
Metal implementation for all k_quants (#1807)
* metal : improve q4_K
28.3 -> 26.0 ms/token by avoiding a branch in the
calculation of the scales.
* metal : small improvement for Q4_K
* metal : still optimizing Q4_K
This commit pushes it down to 25.3 ms / token.
The crazy idea of using 6 bits for the scales is really costly on
Metal: if I remove the bit fiddling necessary to make the block
scales, time goes almost to the Q4_0 23 ms/token.
Before pushing the k-quants upstream I had a Q4_K variant that
had used 8-bit scales. It wasn't more accurate, used 0.125 bits more per weight,
was running slightly slower on the CPU (due to the larger model size
and being memory bound there), and the difference was entirely
negligible under CUDA. So, I decided to publish the version with 6-bit
scales. Perhaps I should re-consider and change to 8-bit scales?
* metal : some more optimizations
Q2_K: 25.4 ms/token
Q6_K: 27.3 ms/token
Q4_0: 22.8 ms/token
Q4_1: 23.1 ms/token
* metal : Q3_K support
Something is not quite right yet.
* metal : Q5_K support
Initial version achieves 31.2 ms/token, 210 GB/s
* metal : still not able to figure out why q3_K does not work
* Minor
* metal : yet another failed attempt to make q3_K work
* metal : optimize Q5_K
31.2 ms -> 27.8 ms.
250 GB/s.
* metal : q3_K still not working
Adding a heavily commented q3_K metal kernel to explain
my obviously faulty logic. Perhaps someone could spot the issue?
* metal : q3_K finally working
Not optimized at all.
What was the issue? The scales are not 4-bytes aligned,
and I was accessing them with a uint32_t pointer.
When I tried that on CUDA, I got an error (illegal memory access)
and added a memcpy to a local array of 3 uint32_t's.
But on Metal it told me there is no memcpy, so I tried
accessing directly. There is no error, just garbage results.
At some point I did try accessing the scales with an uint16_t
pointer (the scales are for sure 2-byte aligned), but was
still getting garbage. I guess, there must have been another bug.
No access to scales is via a uint16_t pointer and, after starting
from scratch from the C dequantize function, it finally works.
* metal : Q3_K 1st optimization pass
* metal : Q3_K second optimization pass - 29.6 ms/token
* metal : Q3_K cleanup
* metal : fixed accidentally broken Q2_K
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-12 21:39:21 +02:00
|
|
|
}
|
2023-07-21 16:05:30 +02:00
|
|
|
|
2023-09-08 18:01:04 +02:00
|
|
|
y1 += 4 * QK_K;
|
Metal implementation for all k_quants (#1807)
* metal : improve q4_K
28.3 -> 26.0 ms/token by avoiding a branch in the
calculation of the scales.
* metal : small improvement for Q4_K
* metal : still optimizing Q4_K
This commit pushes it down to 25.3 ms / token.
The crazy idea of using 6 bits for the scales is really costly on
Metal: if I remove the bit fiddling necessary to make the block
scales, time goes almost to the Q4_0 23 ms/token.
Before pushing the k-quants upstream I had a Q4_K variant that
had used 8-bit scales. It wasn't more accurate, used 0.125 bits more per weight,
was running slightly slower on the CPU (due to the larger model size
and being memory bound there), and the difference was entirely
negligible under CUDA. So, I decided to publish the version with 6-bit
scales. Perhaps I should re-consider and change to 8-bit scales?
* metal : some more optimizations
Q2_K: 25.4 ms/token
Q6_K: 27.3 ms/token
Q4_0: 22.8 ms/token
Q4_1: 23.1 ms/token
* metal : Q3_K support
Something is not quite right yet.
* metal : Q5_K support
Initial version achieves 31.2 ms/token, 210 GB/s
* metal : still not able to figure out why q3_K does not work
* Minor
* metal : yet another failed attempt to make q3_K work
* metal : optimize Q5_K
31.2 ms -> 27.8 ms.
250 GB/s.
* metal : q3_K still not working
Adding a heavily commented q3_K metal kernel to explain
my obviously faulty logic. Perhaps someone could spot the issue?
* metal : q3_K finally working
Not optimized at all.
What was the issue? The scales are not 4-bytes aligned,
and I was accessing them with a uint32_t pointer.
When I tried that on CUDA, I got an error (illegal memory access)
and added a memcpy to a local array of 3 uint32_t's.
But on Metal it told me there is no memcpy, so I tried
accessing directly. There is no error, just garbage results.
At some point I did try accessing the scales with an uint16_t
pointer (the scales are for sure 2-byte aligned), but was
still getting garbage. I guess, there must have been another bug.
No access to scales is via a uint16_t pointer and, after starting
from scratch from the C dequantize function, it finally works.
* metal : Q3_K 1st optimization pass
* metal : Q3_K second optimization pass - 29.6 ms/token
* metal : Q3_K cleanup
* metal : fixed accidentally broken Q2_K
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-12 21:39:21 +02:00
|
|
|
|
|
|
|
}
|
|
|
|
|
2023-07-21 16:05:30 +02:00
|
|
|
for (int row = 0; row < 2; ++row) {
|
2023-09-08 18:01:04 +02:00
|
|
|
const float sumf = (sumf1[row] + 0.25f * sumf2[row]) / (1 << shift);
|
|
|
|
sumf1[row] = simd_sum(sumf);
|
|
|
|
}
|
|
|
|
if (tiisg == 0) {
|
|
|
|
for (int row = 0; row < 2; ++row) {
|
2023-12-07 21:26:54 +01:00
|
|
|
dst[r1*ne0 + im*ne0*ne1 + first_row + row] = sumf1[row];
|
2023-07-21 16:05:30 +02:00
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
k-quants : support for super-block size of 64 (#2001)
* k_quants: WIP super-blocks with 64 weights
* k_quants: WIP super-blocks with 64 weights
Q6_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q4_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q2_K scalar and AVX2 works. Q2_K is way too slow (it is actually slower
than the scalar implementation)
* k_quants: WIP super-blocks with 64 weights
Q3_K scalar and AVX2 works.
* k_quants: WIP super-blocks with 64 weights
Q5_K scalar and AVX2 works, and with that all
k_quants are done on AVX2 and scalar
* k_quants: WIP super-blocks with 64 weights
Q6_K working on CUDA. Cannot make it run quite as gast as
with super-blocks with 256 weigths: 8% slower on 4080,
20% slower on the 1660 (but there we fit 1 less layer on the
GPU because pf the larger model size), so some fraction of
these 20% is due to that,
* k_quants: WIP super-blocks with 64 weights
Q4_K working on CUDA. ~10% slower on GTX-1660,
16% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q2_K working on CUDA. ~3% slower on GTX-1660,
10% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q3_K working on CUDA.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on CUDA, and with this CUDA is done.
* k_quants: WIP super-blocks with 64 weights
Q6_K working on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Q4_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q2_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q3_K working on ARM_NEON, but quite a bit slower than 256 weights.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on ARM_NEON, but quite a bit slower than 256 weights.
With that, we have full support for ARM_NEON, although
performance is not quite there.
* k_quants: WIP super-blocks with 64 weights
Slightly more efficient Q3_K and Q5_K
* k_quants: WIP super-blocks with 64 weights
Another small improvement for Q3_K and Q5_K on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Yet another speedup for Q5_K on ARM_NEON.
We are now within 10% of the QK_K = 256 version.
* k_quants: WIP super-blocks with 64 weights
* We are able to pass preprocessor macros to the Metal
compiler
* Q6_K works and is actually slightly more efficient than
the QK_K = 256 version (25.2 ms vs 25.8 ms)
* k_quants: WIP super-blocks with 64 weights
Q4_K works on Metal and is actually slightly faster
than QK_K = 256 (21.95 ms vs 24.0 ms).
* k_quants: WIP super-blocks with 64 weights
Q2_K works on Metal and is very slightly faster
than QK_K = 256 (23.8 ms vs 24.2 ms).
* k_quants: WIP super-blocks with 64 weights
Q3_K works on Metal and is slightly faster
than QK_K = 256 (26.6 ms vs 28.3 ms).
* k_quants: WIP super-blocks with 64 weights
Q5_K works on Metal and is slightly faster
than QK_K = 256 (23.7 ms vs 26.3 ms).
* k_quants: call them _K, not _k, also on Metal
* k_quants: correctly define QK_K in llama.cpp
* Fixed bug in q4_K quantization added with the 64-block addition
* Simplify via lambda
* k_quants: swicth Q3_K to 4-bit scales when QK_K = 64
Otherwise there isn't much benefit from this
quantization type. There is some very slight loss
in accuracy, but we reduce size by ~7%.
E.g., for OpenLLaMA-3B, Q3_K_S perplexity is
8.6131 with 8-bit scales and 8.6352 with 4-bit,
while file size decreases from 1.53G to 1.44G.
* k_quants: switch Q4_K to 4-bit scales when QK_K = 64
Here the loss in accuracy is greater than for Q3_K,
but the Q4_K points still move further to the left on
the perplexity vs size curve.
* k_quants: forgot to add the Metal changes in last commit
* k_quants: change Q5_K to be type 0 when QK_K = 64
Still needs AVX2 implementation
* k_quants: AVX2 implementation for new 64-weight Q5_K
* k_quants: 10% faster ARM_NEON Q5_K dot product
* k_quants: fixed issue caused by merging with master
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-26 18:43:07 +02:00
|
|
|
#else
|
2023-12-13 13:04:25 +01:00
|
|
|
void kernel_mul_mv_q3_K_f32_impl(
|
2023-07-21 16:05:30 +02:00
|
|
|
device const void * src0,
|
|
|
|
device const float * src1,
|
|
|
|
device float * dst,
|
|
|
|
constant int64_t & ne00,
|
2023-12-13 13:04:25 +01:00
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant int64_t & ne10,
|
|
|
|
constant int64_t & ne12,
|
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
|
|
|
constant uint & r2,
|
|
|
|
constant uint & r3,
|
2023-08-16 22:07:04 +02:00
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
2023-12-07 21:26:54 +01:00
|
|
|
uint tiisg[[thread_index_in_simdgroup]],
|
|
|
|
uint sgitg[[simdgroup_index_in_threadgroup]]) {
|
2023-07-21 16:05:30 +02:00
|
|
|
|
|
|
|
const int nb = ne00/QK_K;
|
|
|
|
|
|
|
|
const int64_t r0 = tgpig.x;
|
|
|
|
const int64_t r1 = tgpig.y;
|
2023-12-07 21:26:54 +01:00
|
|
|
const int64_t im = tgpig.z;
|
2023-07-21 16:05:30 +02:00
|
|
|
|
|
|
|
const int row = 2 * r0 + sgitg;
|
2023-12-07 21:26:54 +01:00
|
|
|
|
|
|
|
const uint i12 = im%ne12;
|
|
|
|
const uint i13 = im/ne12;
|
|
|
|
|
|
|
|
const uint offset0 = (i12/r2)*(nb*ne01) + (i13/r3)*(nb*ne01*ne02);
|
|
|
|
|
2023-08-16 22:07:04 +02:00
|
|
|
device const block_q3_K * x = (device const block_q3_K *) src0 + row*nb + offset0;
|
2023-12-07 21:26:54 +01:00
|
|
|
device const float * yy = (device const float *) src1 + r1*ne10 + im*ne00*ne1;
|
|
|
|
|
2023-07-21 16:05:30 +02:00
|
|
|
const int ix = tiisg/4;
|
|
|
|
const int il = 4 * (tiisg%4);// 0, 4, 8, 12
|
2023-12-07 21:26:54 +01:00
|
|
|
const int iq = il/8; // 0, 0, 1, 1
|
k-quants : support for super-block size of 64 (#2001)
* k_quants: WIP super-blocks with 64 weights
* k_quants: WIP super-blocks with 64 weights
Q6_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q4_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q2_K scalar and AVX2 works. Q2_K is way too slow (it is actually slower
than the scalar implementation)
* k_quants: WIP super-blocks with 64 weights
Q3_K scalar and AVX2 works.
* k_quants: WIP super-blocks with 64 weights
Q5_K scalar and AVX2 works, and with that all
k_quants are done on AVX2 and scalar
* k_quants: WIP super-blocks with 64 weights
Q6_K working on CUDA. Cannot make it run quite as gast as
with super-blocks with 256 weigths: 8% slower on 4080,
20% slower on the 1660 (but there we fit 1 less layer on the
GPU because pf the larger model size), so some fraction of
these 20% is due to that,
* k_quants: WIP super-blocks with 64 weights
Q4_K working on CUDA. ~10% slower on GTX-1660,
16% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q2_K working on CUDA. ~3% slower on GTX-1660,
10% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q3_K working on CUDA.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on CUDA, and with this CUDA is done.
* k_quants: WIP super-blocks with 64 weights
Q6_K working on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Q4_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q2_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q3_K working on ARM_NEON, but quite a bit slower than 256 weights.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on ARM_NEON, but quite a bit slower than 256 weights.
With that, we have full support for ARM_NEON, although
performance is not quite there.
* k_quants: WIP super-blocks with 64 weights
Slightly more efficient Q3_K and Q5_K
* k_quants: WIP super-blocks with 64 weights
Another small improvement for Q3_K and Q5_K on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Yet another speedup for Q5_K on ARM_NEON.
We are now within 10% of the QK_K = 256 version.
* k_quants: WIP super-blocks with 64 weights
* We are able to pass preprocessor macros to the Metal
compiler
* Q6_K works and is actually slightly more efficient than
the QK_K = 256 version (25.2 ms vs 25.8 ms)
* k_quants: WIP super-blocks with 64 weights
Q4_K works on Metal and is actually slightly faster
than QK_K = 256 (21.95 ms vs 24.0 ms).
* k_quants: WIP super-blocks with 64 weights
Q2_K works on Metal and is very slightly faster
than QK_K = 256 (23.8 ms vs 24.2 ms).
* k_quants: WIP super-blocks with 64 weights
Q3_K works on Metal and is slightly faster
than QK_K = 256 (26.6 ms vs 28.3 ms).
* k_quants: WIP super-blocks with 64 weights
Q5_K works on Metal and is slightly faster
than QK_K = 256 (23.7 ms vs 26.3 ms).
* k_quants: call them _K, not _k, also on Metal
* k_quants: correctly define QK_K in llama.cpp
* Fixed bug in q4_K quantization added with the 64-block addition
* Simplify via lambda
* k_quants: swicth Q3_K to 4-bit scales when QK_K = 64
Otherwise there isn't much benefit from this
quantization type. There is some very slight loss
in accuracy, but we reduce size by ~7%.
E.g., for OpenLLaMA-3B, Q3_K_S perplexity is
8.6131 with 8-bit scales and 8.6352 with 4-bit,
while file size decreases from 1.53G to 1.44G.
* k_quants: switch Q4_K to 4-bit scales when QK_K = 64
Here the loss in accuracy is greater than for Q3_K,
but the Q4_K points still move further to the left on
the perplexity vs size curve.
* k_quants: forgot to add the Metal changes in last commit
* k_quants: change Q5_K to be type 0 when QK_K = 64
Still needs AVX2 implementation
* k_quants: AVX2 implementation for new 64-weight Q5_K
* k_quants: 10% faster ARM_NEON Q5_K dot product
* k_quants: fixed issue caused by merging with master
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-26 18:43:07 +02:00
|
|
|
const int in = il%8; // 0, 4, 0, 4
|
|
|
|
|
2023-07-21 16:05:30 +02:00
|
|
|
float2 sum = {0.f, 0.f};
|
k-quants : support for super-block size of 64 (#2001)
* k_quants: WIP super-blocks with 64 weights
* k_quants: WIP super-blocks with 64 weights
Q6_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q4_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q2_K scalar and AVX2 works. Q2_K is way too slow (it is actually slower
than the scalar implementation)
* k_quants: WIP super-blocks with 64 weights
Q3_K scalar and AVX2 works.
* k_quants: WIP super-blocks with 64 weights
Q5_K scalar and AVX2 works, and with that all
k_quants are done on AVX2 and scalar
* k_quants: WIP super-blocks with 64 weights
Q6_K working on CUDA. Cannot make it run quite as gast as
with super-blocks with 256 weigths: 8% slower on 4080,
20% slower on the 1660 (but there we fit 1 less layer on the
GPU because pf the larger model size), so some fraction of
these 20% is due to that,
* k_quants: WIP super-blocks with 64 weights
Q4_K working on CUDA. ~10% slower on GTX-1660,
16% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q2_K working on CUDA. ~3% slower on GTX-1660,
10% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q3_K working on CUDA.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on CUDA, and with this CUDA is done.
* k_quants: WIP super-blocks with 64 weights
Q6_K working on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Q4_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q2_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q3_K working on ARM_NEON, but quite a bit slower than 256 weights.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on ARM_NEON, but quite a bit slower than 256 weights.
With that, we have full support for ARM_NEON, although
performance is not quite there.
* k_quants: WIP super-blocks with 64 weights
Slightly more efficient Q3_K and Q5_K
* k_quants: WIP super-blocks with 64 weights
Another small improvement for Q3_K and Q5_K on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Yet another speedup for Q5_K on ARM_NEON.
We are now within 10% of the QK_K = 256 version.
* k_quants: WIP super-blocks with 64 weights
* We are able to pass preprocessor macros to the Metal
compiler
* Q6_K works and is actually slightly more efficient than
the QK_K = 256 version (25.2 ms vs 25.8 ms)
* k_quants: WIP super-blocks with 64 weights
Q4_K works on Metal and is actually slightly faster
than QK_K = 256 (21.95 ms vs 24.0 ms).
* k_quants: WIP super-blocks with 64 weights
Q2_K works on Metal and is very slightly faster
than QK_K = 256 (23.8 ms vs 24.2 ms).
* k_quants: WIP super-blocks with 64 weights
Q3_K works on Metal and is slightly faster
than QK_K = 256 (26.6 ms vs 28.3 ms).
* k_quants: WIP super-blocks with 64 weights
Q5_K works on Metal and is slightly faster
than QK_K = 256 (23.7 ms vs 26.3 ms).
* k_quants: call them _K, not _k, also on Metal
* k_quants: correctly define QK_K in llama.cpp
* Fixed bug in q4_K quantization added with the 64-block addition
* Simplify via lambda
* k_quants: swicth Q3_K to 4-bit scales when QK_K = 64
Otherwise there isn't much benefit from this
quantization type. There is some very slight loss
in accuracy, but we reduce size by ~7%.
E.g., for OpenLLaMA-3B, Q3_K_S perplexity is
8.6131 with 8-bit scales and 8.6352 with 4-bit,
while file size decreases from 1.53G to 1.44G.
* k_quants: switch Q4_K to 4-bit scales when QK_K = 64
Here the loss in accuracy is greater than for Q3_K,
but the Q4_K points still move further to the left on
the perplexity vs size curve.
* k_quants: forgot to add the Metal changes in last commit
* k_quants: change Q5_K to be type 0 when QK_K = 64
Still needs AVX2 implementation
* k_quants: AVX2 implementation for new 64-weight Q5_K
* k_quants: 10% faster ARM_NEON Q5_K dot product
* k_quants: fixed issue caused by merging with master
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-26 18:43:07 +02:00
|
|
|
|
2023-07-21 16:05:30 +02:00
|
|
|
for (int i = ix; i < nb; i += 8) {
|
k-quants : support for super-block size of 64 (#2001)
* k_quants: WIP super-blocks with 64 weights
* k_quants: WIP super-blocks with 64 weights
Q6_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q4_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q2_K scalar and AVX2 works. Q2_K is way too slow (it is actually slower
than the scalar implementation)
* k_quants: WIP super-blocks with 64 weights
Q3_K scalar and AVX2 works.
* k_quants: WIP super-blocks with 64 weights
Q5_K scalar and AVX2 works, and with that all
k_quants are done on AVX2 and scalar
* k_quants: WIP super-blocks with 64 weights
Q6_K working on CUDA. Cannot make it run quite as gast as
with super-blocks with 256 weigths: 8% slower on 4080,
20% slower on the 1660 (but there we fit 1 less layer on the
GPU because pf the larger model size), so some fraction of
these 20% is due to that,
* k_quants: WIP super-blocks with 64 weights
Q4_K working on CUDA. ~10% slower on GTX-1660,
16% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q2_K working on CUDA. ~3% slower on GTX-1660,
10% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q3_K working on CUDA.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on CUDA, and with this CUDA is done.
* k_quants: WIP super-blocks with 64 weights
Q6_K working on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Q4_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q2_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q3_K working on ARM_NEON, but quite a bit slower than 256 weights.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on ARM_NEON, but quite a bit slower than 256 weights.
With that, we have full support for ARM_NEON, although
performance is not quite there.
* k_quants: WIP super-blocks with 64 weights
Slightly more efficient Q3_K and Q5_K
* k_quants: WIP super-blocks with 64 weights
Another small improvement for Q3_K and Q5_K on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Yet another speedup for Q5_K on ARM_NEON.
We are now within 10% of the QK_K = 256 version.
* k_quants: WIP super-blocks with 64 weights
* We are able to pass preprocessor macros to the Metal
compiler
* Q6_K works and is actually slightly more efficient than
the QK_K = 256 version (25.2 ms vs 25.8 ms)
* k_quants: WIP super-blocks with 64 weights
Q4_K works on Metal and is actually slightly faster
than QK_K = 256 (21.95 ms vs 24.0 ms).
* k_quants: WIP super-blocks with 64 weights
Q2_K works on Metal and is very slightly faster
than QK_K = 256 (23.8 ms vs 24.2 ms).
* k_quants: WIP super-blocks with 64 weights
Q3_K works on Metal and is slightly faster
than QK_K = 256 (26.6 ms vs 28.3 ms).
* k_quants: WIP super-blocks with 64 weights
Q5_K works on Metal and is slightly faster
than QK_K = 256 (23.7 ms vs 26.3 ms).
* k_quants: call them _K, not _k, also on Metal
* k_quants: correctly define QK_K in llama.cpp
* Fixed bug in q4_K quantization added with the 64-block addition
* Simplify via lambda
* k_quants: swicth Q3_K to 4-bit scales when QK_K = 64
Otherwise there isn't much benefit from this
quantization type. There is some very slight loss
in accuracy, but we reduce size by ~7%.
E.g., for OpenLLaMA-3B, Q3_K_S perplexity is
8.6131 with 8-bit scales and 8.6352 with 4-bit,
while file size decreases from 1.53G to 1.44G.
* k_quants: switch Q4_K to 4-bit scales when QK_K = 64
Here the loss in accuracy is greater than for Q3_K,
but the Q4_K points still move further to the left on
the perplexity vs size curve.
* k_quants: forgot to add the Metal changes in last commit
* k_quants: change Q5_K to be type 0 when QK_K = 64
Still needs AVX2 implementation
* k_quants: AVX2 implementation for new 64-weight Q5_K
* k_quants: 10% faster ARM_NEON Q5_K dot product
* k_quants: fixed issue caused by merging with master
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-26 18:43:07 +02:00
|
|
|
|
|
|
|
const float d_all = (float)(x[i].d);
|
|
|
|
|
2023-07-21 16:05:30 +02:00
|
|
|
device const uint16_t * q = (device const uint16_t *)(x[i].qs + il);
|
|
|
|
device const uint16_t * h = (device const uint16_t *)(x[i].hmask + in);
|
|
|
|
device const uint16_t * s = (device const uint16_t *)(x[i].scales);
|
|
|
|
device const float * y = yy + i * QK_K + il;
|
|
|
|
|
|
|
|
const float d1 = d_all * ((int32_t)(s[0] & 0x000F) - 8);
|
|
|
|
const float d2 = d_all * ((int32_t)(s[0] & 0x00F0) - 128) * 1.f/64.f;
|
|
|
|
const float d3 = d_all * ((int32_t)(s[0] & 0x0F00) - 2048) * 1.f/4096.f;
|
|
|
|
const float d4 = d_all * ((int32_t)(s[0] & 0xF000) - 32768) * 1.f/262144.f;
|
|
|
|
|
|
|
|
for (int l = 0; l < 4; l += 2) {
|
2023-12-07 21:26:54 +01:00
|
|
|
const uint16_t hm = h[l/2] >> iq;
|
2023-07-21 16:05:30 +02:00
|
|
|
sum[0] += y[l+ 0] * d1 * ((int32_t)(q[l/2] & 0x0003) - ((hm & 0x0001) ? 0 : 4))
|
|
|
|
+ y[l+16] * d2 * ((int32_t)(q[l/2] & 0x000c) - ((hm & 0x0004) ? 0 : 16))
|
|
|
|
+ y[l+32] * d3 * ((int32_t)(q[l/2] & 0x0030) - ((hm & 0x0010) ? 0 : 64))
|
|
|
|
+ y[l+48] * d4 * ((int32_t)(q[l/2] & 0x00c0) - ((hm & 0x0040) ? 0 : 256));
|
|
|
|
sum[1] += y[l+ 1] * d1 * ((int32_t)(q[l/2] & 0x0300) - ((hm & 0x0100) ? 0 : 1024))
|
|
|
|
+ y[l+17] * d2 * ((int32_t)(q[l/2] & 0x0c00) - ((hm & 0x0400) ? 0 : 4096))
|
|
|
|
+ y[l+33] * d3 * ((int32_t)(q[l/2] & 0x3000) - ((hm & 0x1000) ? 0 : 16384))
|
|
|
|
+ y[l+49] * d4 * ((int32_t)(q[l/2] & 0xc000) - ((hm & 0x4000) ? 0 : 65536));
|
k-quants : support for super-block size of 64 (#2001)
* k_quants: WIP super-blocks with 64 weights
* k_quants: WIP super-blocks with 64 weights
Q6_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q4_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q2_K scalar and AVX2 works. Q2_K is way too slow (it is actually slower
than the scalar implementation)
* k_quants: WIP super-blocks with 64 weights
Q3_K scalar and AVX2 works.
* k_quants: WIP super-blocks with 64 weights
Q5_K scalar and AVX2 works, and with that all
k_quants are done on AVX2 and scalar
* k_quants: WIP super-blocks with 64 weights
Q6_K working on CUDA. Cannot make it run quite as gast as
with super-blocks with 256 weigths: 8% slower on 4080,
20% slower on the 1660 (but there we fit 1 less layer on the
GPU because pf the larger model size), so some fraction of
these 20% is due to that,
* k_quants: WIP super-blocks with 64 weights
Q4_K working on CUDA. ~10% slower on GTX-1660,
16% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q2_K working on CUDA. ~3% slower on GTX-1660,
10% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q3_K working on CUDA.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on CUDA, and with this CUDA is done.
* k_quants: WIP super-blocks with 64 weights
Q6_K working on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Q4_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q2_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q3_K working on ARM_NEON, but quite a bit slower than 256 weights.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on ARM_NEON, but quite a bit slower than 256 weights.
With that, we have full support for ARM_NEON, although
performance is not quite there.
* k_quants: WIP super-blocks with 64 weights
Slightly more efficient Q3_K and Q5_K
* k_quants: WIP super-blocks with 64 weights
Another small improvement for Q3_K and Q5_K on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Yet another speedup for Q5_K on ARM_NEON.
We are now within 10% of the QK_K = 256 version.
* k_quants: WIP super-blocks with 64 weights
* We are able to pass preprocessor macros to the Metal
compiler
* Q6_K works and is actually slightly more efficient than
the QK_K = 256 version (25.2 ms vs 25.8 ms)
* k_quants: WIP super-blocks with 64 weights
Q4_K works on Metal and is actually slightly faster
than QK_K = 256 (21.95 ms vs 24.0 ms).
* k_quants: WIP super-blocks with 64 weights
Q2_K works on Metal and is very slightly faster
than QK_K = 256 (23.8 ms vs 24.2 ms).
* k_quants: WIP super-blocks with 64 weights
Q3_K works on Metal and is slightly faster
than QK_K = 256 (26.6 ms vs 28.3 ms).
* k_quants: WIP super-blocks with 64 weights
Q5_K works on Metal and is slightly faster
than QK_K = 256 (23.7 ms vs 26.3 ms).
* k_quants: call them _K, not _k, also on Metal
* k_quants: correctly define QK_K in llama.cpp
* Fixed bug in q4_K quantization added with the 64-block addition
* Simplify via lambda
* k_quants: swicth Q3_K to 4-bit scales when QK_K = 64
Otherwise there isn't much benefit from this
quantization type. There is some very slight loss
in accuracy, but we reduce size by ~7%.
E.g., for OpenLLaMA-3B, Q3_K_S perplexity is
8.6131 with 8-bit scales and 8.6352 with 4-bit,
while file size decreases from 1.53G to 1.44G.
* k_quants: switch Q4_K to 4-bit scales when QK_K = 64
Here the loss in accuracy is greater than for Q3_K,
but the Q4_K points still move further to the left on
the perplexity vs size curve.
* k_quants: forgot to add the Metal changes in last commit
* k_quants: change Q5_K to be type 0 when QK_K = 64
Still needs AVX2 implementation
* k_quants: AVX2 implementation for new 64-weight Q5_K
* k_quants: 10% faster ARM_NEON Q5_K dot product
* k_quants: fixed issue caused by merging with master
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-26 18:43:07 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
}
|
2023-07-21 16:05:30 +02:00
|
|
|
const float sumf = sum[0] + sum[1] * 1.f/256.f;
|
k-quants : support for super-block size of 64 (#2001)
* k_quants: WIP super-blocks with 64 weights
* k_quants: WIP super-blocks with 64 weights
Q6_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q4_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q2_K scalar and AVX2 works. Q2_K is way too slow (it is actually slower
than the scalar implementation)
* k_quants: WIP super-blocks with 64 weights
Q3_K scalar and AVX2 works.
* k_quants: WIP super-blocks with 64 weights
Q5_K scalar and AVX2 works, and with that all
k_quants are done on AVX2 and scalar
* k_quants: WIP super-blocks with 64 weights
Q6_K working on CUDA. Cannot make it run quite as gast as
with super-blocks with 256 weigths: 8% slower on 4080,
20% slower on the 1660 (but there we fit 1 less layer on the
GPU because pf the larger model size), so some fraction of
these 20% is due to that,
* k_quants: WIP super-blocks with 64 weights
Q4_K working on CUDA. ~10% slower on GTX-1660,
16% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q2_K working on CUDA. ~3% slower on GTX-1660,
10% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q3_K working on CUDA.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on CUDA, and with this CUDA is done.
* k_quants: WIP super-blocks with 64 weights
Q6_K working on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Q4_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q2_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q3_K working on ARM_NEON, but quite a bit slower than 256 weights.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on ARM_NEON, but quite a bit slower than 256 weights.
With that, we have full support for ARM_NEON, although
performance is not quite there.
* k_quants: WIP super-blocks with 64 weights
Slightly more efficient Q3_K and Q5_K
* k_quants: WIP super-blocks with 64 weights
Another small improvement for Q3_K and Q5_K on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Yet another speedup for Q5_K on ARM_NEON.
We are now within 10% of the QK_K = 256 version.
* k_quants: WIP super-blocks with 64 weights
* We are able to pass preprocessor macros to the Metal
compiler
* Q6_K works and is actually slightly more efficient than
the QK_K = 256 version (25.2 ms vs 25.8 ms)
* k_quants: WIP super-blocks with 64 weights
Q4_K works on Metal and is actually slightly faster
than QK_K = 256 (21.95 ms vs 24.0 ms).
* k_quants: WIP super-blocks with 64 weights
Q2_K works on Metal and is very slightly faster
than QK_K = 256 (23.8 ms vs 24.2 ms).
* k_quants: WIP super-blocks with 64 weights
Q3_K works on Metal and is slightly faster
than QK_K = 256 (26.6 ms vs 28.3 ms).
* k_quants: WIP super-blocks with 64 weights
Q5_K works on Metal and is slightly faster
than QK_K = 256 (23.7 ms vs 26.3 ms).
* k_quants: call them _K, not _k, also on Metal
* k_quants: correctly define QK_K in llama.cpp
* Fixed bug in q4_K quantization added with the 64-block addition
* Simplify via lambda
* k_quants: swicth Q3_K to 4-bit scales when QK_K = 64
Otherwise there isn't much benefit from this
quantization type. There is some very slight loss
in accuracy, but we reduce size by ~7%.
E.g., for OpenLLaMA-3B, Q3_K_S perplexity is
8.6131 with 8-bit scales and 8.6352 with 4-bit,
while file size decreases from 1.53G to 1.44G.
* k_quants: switch Q4_K to 4-bit scales when QK_K = 64
Here the loss in accuracy is greater than for Q3_K,
but the Q4_K points still move further to the left on
the perplexity vs size curve.
* k_quants: forgot to add the Metal changes in last commit
* k_quants: change Q5_K to be type 0 when QK_K = 64
Still needs AVX2 implementation
* k_quants: AVX2 implementation for new 64-weight Q5_K
* k_quants: 10% faster ARM_NEON Q5_K dot product
* k_quants: fixed issue caused by merging with master
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-26 18:43:07 +02:00
|
|
|
|
2023-07-21 16:05:30 +02:00
|
|
|
const float tot = simd_sum(sumf);
|
|
|
|
if (tiisg == 0) {
|
2023-12-07 21:26:54 +01:00
|
|
|
dst[r1*ne0 + im*ne0*ne1 + row] = tot;
|
Metal implementation for all k_quants (#1807)
* metal : improve q4_K
28.3 -> 26.0 ms/token by avoiding a branch in the
calculation of the scales.
* metal : small improvement for Q4_K
* metal : still optimizing Q4_K
This commit pushes it down to 25.3 ms / token.
The crazy idea of using 6 bits for the scales is really costly on
Metal: if I remove the bit fiddling necessary to make the block
scales, time goes almost to the Q4_0 23 ms/token.
Before pushing the k-quants upstream I had a Q4_K variant that
had used 8-bit scales. It wasn't more accurate, used 0.125 bits more per weight,
was running slightly slower on the CPU (due to the larger model size
and being memory bound there), and the difference was entirely
negligible under CUDA. So, I decided to publish the version with 6-bit
scales. Perhaps I should re-consider and change to 8-bit scales?
* metal : some more optimizations
Q2_K: 25.4 ms/token
Q6_K: 27.3 ms/token
Q4_0: 22.8 ms/token
Q4_1: 23.1 ms/token
* metal : Q3_K support
Something is not quite right yet.
* metal : Q5_K support
Initial version achieves 31.2 ms/token, 210 GB/s
* metal : still not able to figure out why q3_K does not work
* Minor
* metal : yet another failed attempt to make q3_K work
* metal : optimize Q5_K
31.2 ms -> 27.8 ms.
250 GB/s.
* metal : q3_K still not working
Adding a heavily commented q3_K metal kernel to explain
my obviously faulty logic. Perhaps someone could spot the issue?
* metal : q3_K finally working
Not optimized at all.
What was the issue? The scales are not 4-bytes aligned,
and I was accessing them with a uint32_t pointer.
When I tried that on CUDA, I got an error (illegal memory access)
and added a memcpy to a local array of 3 uint32_t's.
But on Metal it told me there is no memcpy, so I tried
accessing directly. There is no error, just garbage results.
At some point I did try accessing the scales with an uint16_t
pointer (the scales are for sure 2-byte aligned), but was
still getting garbage. I guess, there must have been another bug.
No access to scales is via a uint16_t pointer and, after starting
from scratch from the C dequantize function, it finally works.
* metal : Q3_K 1st optimization pass
* metal : Q3_K second optimization pass - 29.6 ms/token
* metal : Q3_K cleanup
* metal : fixed accidentally broken Q2_K
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-12 21:39:21 +02:00
|
|
|
}
|
2023-06-08 21:28:21 +02:00
|
|
|
|
|
|
|
}
|
2023-07-21 16:05:30 +02:00
|
|
|
#endif
|
2023-06-08 21:28:21 +02:00
|
|
|
|
2023-12-13 13:04:25 +01:00
|
|
|
[[host_name("kernel_mul_mv_q3_K_f32")]]
|
|
|
|
kernel void kernel_mul_mv_q3_K_f32(
|
|
|
|
device const void * src0,
|
|
|
|
device const float * src1,
|
|
|
|
device float * dst,
|
|
|
|
constant int64_t & ne00,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant uint64_t & nb00,
|
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant uint64_t & nb02,
|
|
|
|
constant int64_t & ne10,
|
|
|
|
constant int64_t & ne11,
|
|
|
|
constant int64_t & ne12,
|
|
|
|
constant uint64_t & nb10,
|
|
|
|
constant uint64_t & nb11,
|
|
|
|
constant uint64_t & nb12,
|
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
|
|
|
constant uint & r2,
|
|
|
|
constant uint & r3,
|
2023-12-13 13:04:25 +01:00
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint tiisg[[thread_index_in_simdgroup]],
|
|
|
|
uint sgitg[[simdgroup_index_in_threadgroup]]) {
|
|
|
|
|
|
|
|
kernel_mul_mv_q3_K_f32_impl(src0, src1, dst, ne00, ne01, ne02, ne10, ne12, ne0, ne1, r2, r3, tgpig, tiisg, sgitg);
|
|
|
|
}
|
|
|
|
|
2023-07-20 14:18:43 +02:00
|
|
|
#if QK_K == 256
|
2023-12-13 13:04:25 +01:00
|
|
|
void kernel_mul_mv_q4_K_f32_impl(
|
2023-06-08 09:08:23 +02:00
|
|
|
device const void * src0,
|
|
|
|
device const float * src1,
|
|
|
|
device float * dst,
|
|
|
|
constant int64_t & ne00,
|
2023-12-13 13:04:25 +01:00
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant int64_t & ne10,
|
|
|
|
constant int64_t & ne12,
|
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
|
|
|
constant uint & r2,
|
|
|
|
constant uint & r3,
|
2023-08-16 22:07:04 +02:00
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
2023-12-07 21:26:54 +01:00
|
|
|
uint tiisg[[thread_index_in_simdgroup]],
|
|
|
|
uint sgitg[[simdgroup_index_in_threadgroup]]) {
|
k-quants : support for super-block size of 64 (#2001)
* k_quants: WIP super-blocks with 64 weights
* k_quants: WIP super-blocks with 64 weights
Q6_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q4_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q2_K scalar and AVX2 works. Q2_K is way too slow (it is actually slower
than the scalar implementation)
* k_quants: WIP super-blocks with 64 weights
Q3_K scalar and AVX2 works.
* k_quants: WIP super-blocks with 64 weights
Q5_K scalar and AVX2 works, and with that all
k_quants are done on AVX2 and scalar
* k_quants: WIP super-blocks with 64 weights
Q6_K working on CUDA. Cannot make it run quite as gast as
with super-blocks with 256 weigths: 8% slower on 4080,
20% slower on the 1660 (but there we fit 1 less layer on the
GPU because pf the larger model size), so some fraction of
these 20% is due to that,
* k_quants: WIP super-blocks with 64 weights
Q4_K working on CUDA. ~10% slower on GTX-1660,
16% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q2_K working on CUDA. ~3% slower on GTX-1660,
10% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q3_K working on CUDA.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on CUDA, and with this CUDA is done.
* k_quants: WIP super-blocks with 64 weights
Q6_K working on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Q4_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q2_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q3_K working on ARM_NEON, but quite a bit slower than 256 weights.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on ARM_NEON, but quite a bit slower than 256 weights.
With that, we have full support for ARM_NEON, although
performance is not quite there.
* k_quants: WIP super-blocks with 64 weights
Slightly more efficient Q3_K and Q5_K
* k_quants: WIP super-blocks with 64 weights
Another small improvement for Q3_K and Q5_K on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Yet another speedup for Q5_K on ARM_NEON.
We are now within 10% of the QK_K = 256 version.
* k_quants: WIP super-blocks with 64 weights
* We are able to pass preprocessor macros to the Metal
compiler
* Q6_K works and is actually slightly more efficient than
the QK_K = 256 version (25.2 ms vs 25.8 ms)
* k_quants: WIP super-blocks with 64 weights
Q4_K works on Metal and is actually slightly faster
than QK_K = 256 (21.95 ms vs 24.0 ms).
* k_quants: WIP super-blocks with 64 weights
Q2_K works on Metal and is very slightly faster
than QK_K = 256 (23.8 ms vs 24.2 ms).
* k_quants: WIP super-blocks with 64 weights
Q3_K works on Metal and is slightly faster
than QK_K = 256 (26.6 ms vs 28.3 ms).
* k_quants: WIP super-blocks with 64 weights
Q5_K works on Metal and is slightly faster
than QK_K = 256 (23.7 ms vs 26.3 ms).
* k_quants: call them _K, not _k, also on Metal
* k_quants: correctly define QK_K in llama.cpp
* Fixed bug in q4_K quantization added with the 64-block addition
* Simplify via lambda
* k_quants: swicth Q3_K to 4-bit scales when QK_K = 64
Otherwise there isn't much benefit from this
quantization type. There is some very slight loss
in accuracy, but we reduce size by ~7%.
E.g., for OpenLLaMA-3B, Q3_K_S perplexity is
8.6131 with 8-bit scales and 8.6352 with 4-bit,
while file size decreases from 1.53G to 1.44G.
* k_quants: switch Q4_K to 4-bit scales when QK_K = 64
Here the loss in accuracy is greater than for Q3_K,
but the Q4_K points still move further to the left on
the perplexity vs size curve.
* k_quants: forgot to add the Metal changes in last commit
* k_quants: change Q5_K to be type 0 when QK_K = 64
Still needs AVX2 implementation
* k_quants: AVX2 implementation for new 64-weight Q5_K
* k_quants: 10% faster ARM_NEON Q5_K dot product
* k_quants: fixed issue caused by merging with master
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-26 18:43:07 +02:00
|
|
|
|
|
|
|
const uint16_t kmask1 = 0x3f3f;
|
|
|
|
const uint16_t kmask2 = 0x0f0f;
|
|
|
|
const uint16_t kmask3 = 0xc0c0;
|
|
|
|
|
2023-07-20 14:18:43 +02:00
|
|
|
const int ix = tiisg/8; // 0...3
|
|
|
|
const int it = tiisg%8; // 0...7
|
2023-12-07 21:26:54 +01:00
|
|
|
const int iq = it/4; // 0 or 1
|
2023-07-20 14:18:43 +02:00
|
|
|
const int ir = it%4; // 0...3
|
Metal implementation for all k_quants (#1807)
* metal : improve q4_K
28.3 -> 26.0 ms/token by avoiding a branch in the
calculation of the scales.
* metal : small improvement for Q4_K
* metal : still optimizing Q4_K
This commit pushes it down to 25.3 ms / token.
The crazy idea of using 6 bits for the scales is really costly on
Metal: if I remove the bit fiddling necessary to make the block
scales, time goes almost to the Q4_0 23 ms/token.
Before pushing the k-quants upstream I had a Q4_K variant that
had used 8-bit scales. It wasn't more accurate, used 0.125 bits more per weight,
was running slightly slower on the CPU (due to the larger model size
and being memory bound there), and the difference was entirely
negligible under CUDA. So, I decided to publish the version with 6-bit
scales. Perhaps I should re-consider and change to 8-bit scales?
* metal : some more optimizations
Q2_K: 25.4 ms/token
Q6_K: 27.3 ms/token
Q4_0: 22.8 ms/token
Q4_1: 23.1 ms/token
* metal : Q3_K support
Something is not quite right yet.
* metal : Q5_K support
Initial version achieves 31.2 ms/token, 210 GB/s
* metal : still not able to figure out why q3_K does not work
* Minor
* metal : yet another failed attempt to make q3_K work
* metal : optimize Q5_K
31.2 ms -> 27.8 ms.
250 GB/s.
* metal : q3_K still not working
Adding a heavily commented q3_K metal kernel to explain
my obviously faulty logic. Perhaps someone could spot the issue?
* metal : q3_K finally working
Not optimized at all.
What was the issue? The scales are not 4-bytes aligned,
and I was accessing them with a uint32_t pointer.
When I tried that on CUDA, I got an error (illegal memory access)
and added a memcpy to a local array of 3 uint32_t's.
But on Metal it told me there is no memcpy, so I tried
accessing directly. There is no error, just garbage results.
At some point I did try accessing the scales with an uint16_t
pointer (the scales are for sure 2-byte aligned), but was
still getting garbage. I guess, there must have been another bug.
No access to scales is via a uint16_t pointer and, after starting
from scratch from the C dequantize function, it finally works.
* metal : Q3_K 1st optimization pass
* metal : Q3_K second optimization pass - 29.6 ms/token
* metal : Q3_K cleanup
* metal : fixed accidentally broken Q2_K
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-12 21:39:21 +02:00
|
|
|
|
2023-07-20 14:18:43 +02:00
|
|
|
const int nb = ne00/QK_K;
|
|
|
|
const int r0 = tgpig.x;
|
|
|
|
const int r1 = tgpig.y;
|
2023-12-07 21:26:54 +01:00
|
|
|
const int im = tgpig.z;
|
2023-09-03 10:06:22 +02:00
|
|
|
//const int first_row = (r0 * N_SIMDGROUP + sgitg) * N_DST;
|
|
|
|
const int first_row = r0 * N_DST;
|
2023-07-20 14:18:43 +02:00
|
|
|
const int ib_row = first_row * nb;
|
2023-12-07 21:26:54 +01:00
|
|
|
|
|
|
|
const uint i12 = im%ne12;
|
|
|
|
const uint i13 = im/ne12;
|
|
|
|
|
|
|
|
const uint offset0 = (i12/r2)*(nb*ne01) + (i13/r3)*(nb*ne01*ne02);
|
|
|
|
|
2023-08-16 22:07:04 +02:00
|
|
|
device const block_q4_K * x = (device const block_q4_K *) src0 + ib_row + offset0;
|
2023-12-07 21:26:54 +01:00
|
|
|
device const float * y = (device const float *) src1 + r1*ne10 + im*ne00*ne1;
|
|
|
|
|
2023-07-20 14:18:43 +02:00
|
|
|
float yl[16];
|
|
|
|
float yh[16];
|
|
|
|
float sumf[N_DST]={0.f}, all_sum;
|
Metal implementation for all k_quants (#1807)
* metal : improve q4_K
28.3 -> 26.0 ms/token by avoiding a branch in the
calculation of the scales.
* metal : small improvement for Q4_K
* metal : still optimizing Q4_K
This commit pushes it down to 25.3 ms / token.
The crazy idea of using 6 bits for the scales is really costly on
Metal: if I remove the bit fiddling necessary to make the block
scales, time goes almost to the Q4_0 23 ms/token.
Before pushing the k-quants upstream I had a Q4_K variant that
had used 8-bit scales. It wasn't more accurate, used 0.125 bits more per weight,
was running slightly slower on the CPU (due to the larger model size
and being memory bound there), and the difference was entirely
negligible under CUDA. So, I decided to publish the version with 6-bit
scales. Perhaps I should re-consider and change to 8-bit scales?
* metal : some more optimizations
Q2_K: 25.4 ms/token
Q6_K: 27.3 ms/token
Q4_0: 22.8 ms/token
Q4_1: 23.1 ms/token
* metal : Q3_K support
Something is not quite right yet.
* metal : Q5_K support
Initial version achieves 31.2 ms/token, 210 GB/s
* metal : still not able to figure out why q3_K does not work
* Minor
* metal : yet another failed attempt to make q3_K work
* metal : optimize Q5_K
31.2 ms -> 27.8 ms.
250 GB/s.
* metal : q3_K still not working
Adding a heavily commented q3_K metal kernel to explain
my obviously faulty logic. Perhaps someone could spot the issue?
* metal : q3_K finally working
Not optimized at all.
What was the issue? The scales are not 4-bytes aligned,
and I was accessing them with a uint32_t pointer.
When I tried that on CUDA, I got an error (illegal memory access)
and added a memcpy to a local array of 3 uint32_t's.
But on Metal it told me there is no memcpy, so I tried
accessing directly. There is no error, just garbage results.
At some point I did try accessing the scales with an uint16_t
pointer (the scales are for sure 2-byte aligned), but was
still getting garbage. I guess, there must have been another bug.
No access to scales is via a uint16_t pointer and, after starting
from scratch from the C dequantize function, it finally works.
* metal : Q3_K 1st optimization pass
* metal : Q3_K second optimization pass - 29.6 ms/token
* metal : Q3_K cleanup
* metal : fixed accidentally broken Q2_K
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-12 21:39:21 +02:00
|
|
|
|
2023-07-20 14:18:43 +02:00
|
|
|
const int step = sizeof(block_q4_K) * nb / 2;
|
2023-06-08 09:08:23 +02:00
|
|
|
|
2023-12-07 21:26:54 +01:00
|
|
|
device const float * y4 = y + ix * QK_K + 64 * iq + 8 * ir;
|
Metal implementation for all k_quants (#1807)
* metal : improve q4_K
28.3 -> 26.0 ms/token by avoiding a branch in the
calculation of the scales.
* metal : small improvement for Q4_K
* metal : still optimizing Q4_K
This commit pushes it down to 25.3 ms / token.
The crazy idea of using 6 bits for the scales is really costly on
Metal: if I remove the bit fiddling necessary to make the block
scales, time goes almost to the Q4_0 23 ms/token.
Before pushing the k-quants upstream I had a Q4_K variant that
had used 8-bit scales. It wasn't more accurate, used 0.125 bits more per weight,
was running slightly slower on the CPU (due to the larger model size
and being memory bound there), and the difference was entirely
negligible under CUDA. So, I decided to publish the version with 6-bit
scales. Perhaps I should re-consider and change to 8-bit scales?
* metal : some more optimizations
Q2_K: 25.4 ms/token
Q6_K: 27.3 ms/token
Q4_0: 22.8 ms/token
Q4_1: 23.1 ms/token
* metal : Q3_K support
Something is not quite right yet.
* metal : Q5_K support
Initial version achieves 31.2 ms/token, 210 GB/s
* metal : still not able to figure out why q3_K does not work
* Minor
* metal : yet another failed attempt to make q3_K work
* metal : optimize Q5_K
31.2 ms -> 27.8 ms.
250 GB/s.
* metal : q3_K still not working
Adding a heavily commented q3_K metal kernel to explain
my obviously faulty logic. Perhaps someone could spot the issue?
* metal : q3_K finally working
Not optimized at all.
What was the issue? The scales are not 4-bytes aligned,
and I was accessing them with a uint32_t pointer.
When I tried that on CUDA, I got an error (illegal memory access)
and added a memcpy to a local array of 3 uint32_t's.
But on Metal it told me there is no memcpy, so I tried
accessing directly. There is no error, just garbage results.
At some point I did try accessing the scales with an uint16_t
pointer (the scales are for sure 2-byte aligned), but was
still getting garbage. I guess, there must have been another bug.
No access to scales is via a uint16_t pointer and, after starting
from scratch from the C dequantize function, it finally works.
* metal : Q3_K 1st optimization pass
* metal : Q3_K second optimization pass - 29.6 ms/token
* metal : Q3_K cleanup
* metal : fixed accidentally broken Q2_K
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-12 21:39:21 +02:00
|
|
|
|
2023-07-20 14:18:43 +02:00
|
|
|
uint16_t sc16[4];
|
|
|
|
thread const uint8_t * sc8 = (thread const uint8_t *)sc16;
|
2023-06-08 09:08:23 +02:00
|
|
|
|
2023-07-20 14:18:43 +02:00
|
|
|
for (int ib = ix; ib < nb; ib += 4) {
|
2023-06-08 09:08:23 +02:00
|
|
|
|
2023-07-20 14:18:43 +02:00
|
|
|
float4 sumy = {0.f, 0.f, 0.f, 0.f};
|
|
|
|
for (int i = 0; i < 8; ++i) {
|
|
|
|
yl[i+0] = y4[i+ 0]; sumy[0] += yl[i+0];
|
|
|
|
yl[i+8] = y4[i+ 32]; sumy[1] += yl[i+8];
|
|
|
|
yh[i+0] = y4[i+128]; sumy[2] += yh[i+0];
|
|
|
|
yh[i+8] = y4[i+160]; sumy[3] += yh[i+8];
|
|
|
|
}
|
2023-06-08 09:08:23 +02:00
|
|
|
|
2023-12-07 21:26:54 +01:00
|
|
|
device const uint16_t * sc = (device const uint16_t *)x[ib].scales + iq;
|
|
|
|
device const uint16_t * q1 = (device const uint16_t *)x[ib].qs + 16 * iq + 4 * ir;
|
2023-07-20 14:18:43 +02:00
|
|
|
device const half * dh = &x[ib].d;
|
2023-06-08 09:08:23 +02:00
|
|
|
|
2023-07-20 14:18:43 +02:00
|
|
|
for (int row = 0; row < N_DST; row++) {
|
Metal implementation for all k_quants (#1807)
* metal : improve q4_K
28.3 -> 26.0 ms/token by avoiding a branch in the
calculation of the scales.
* metal : small improvement for Q4_K
* metal : still optimizing Q4_K
This commit pushes it down to 25.3 ms / token.
The crazy idea of using 6 bits for the scales is really costly on
Metal: if I remove the bit fiddling necessary to make the block
scales, time goes almost to the Q4_0 23 ms/token.
Before pushing the k-quants upstream I had a Q4_K variant that
had used 8-bit scales. It wasn't more accurate, used 0.125 bits more per weight,
was running slightly slower on the CPU (due to the larger model size
and being memory bound there), and the difference was entirely
negligible under CUDA. So, I decided to publish the version with 6-bit
scales. Perhaps I should re-consider and change to 8-bit scales?
* metal : some more optimizations
Q2_K: 25.4 ms/token
Q6_K: 27.3 ms/token
Q4_0: 22.8 ms/token
Q4_1: 23.1 ms/token
* metal : Q3_K support
Something is not quite right yet.
* metal : Q5_K support
Initial version achieves 31.2 ms/token, 210 GB/s
* metal : still not able to figure out why q3_K does not work
* Minor
* metal : yet another failed attempt to make q3_K work
* metal : optimize Q5_K
31.2 ms -> 27.8 ms.
250 GB/s.
* metal : q3_K still not working
Adding a heavily commented q3_K metal kernel to explain
my obviously faulty logic. Perhaps someone could spot the issue?
* metal : q3_K finally working
Not optimized at all.
What was the issue? The scales are not 4-bytes aligned,
and I was accessing them with a uint32_t pointer.
When I tried that on CUDA, I got an error (illegal memory access)
and added a memcpy to a local array of 3 uint32_t's.
But on Metal it told me there is no memcpy, so I tried
accessing directly. There is no error, just garbage results.
At some point I did try accessing the scales with an uint16_t
pointer (the scales are for sure 2-byte aligned), but was
still getting garbage. I guess, there must have been another bug.
No access to scales is via a uint16_t pointer and, after starting
from scratch from the C dequantize function, it finally works.
* metal : Q3_K 1st optimization pass
* metal : Q3_K second optimization pass - 29.6 ms/token
* metal : Q3_K cleanup
* metal : fixed accidentally broken Q2_K
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-12 21:39:21 +02:00
|
|
|
|
2023-07-20 14:18:43 +02:00
|
|
|
sc16[0] = sc[0] & kmask1;
|
|
|
|
sc16[1] = sc[2] & kmask1;
|
|
|
|
sc16[2] = ((sc[4] >> 0) & kmask2) | ((sc[0] & kmask3) >> 2);
|
|
|
|
sc16[3] = ((sc[4] >> 4) & kmask2) | ((sc[2] & kmask3) >> 2);
|
|
|
|
|
|
|
|
device const uint16_t * q2 = q1 + 32;
|
|
|
|
|
|
|
|
float4 acc1 = {0.f, 0.f, 0.f, 0.f};
|
|
|
|
float4 acc2 = {0.f, 0.f, 0.f, 0.f};
|
|
|
|
for (int i = 0; i < 8; i += 2) {
|
|
|
|
acc1[0] += yl[i+0] * (q1[i/2] & 0x000F);
|
|
|
|
acc1[1] += yl[i+1] * (q1[i/2] & 0x0F00);
|
|
|
|
acc1[2] += yl[i+8] * (q1[i/2] & 0x00F0);
|
|
|
|
acc1[3] += yl[i+9] * (q1[i/2] & 0xF000);
|
|
|
|
acc2[0] += yh[i+0] * (q2[i/2] & 0x000F);
|
|
|
|
acc2[1] += yh[i+1] * (q2[i/2] & 0x0F00);
|
|
|
|
acc2[2] += yh[i+8] * (q2[i/2] & 0x00F0);
|
|
|
|
acc2[3] += yh[i+9] * (q2[i/2] & 0xF000);
|
|
|
|
}
|
Metal implementation for all k_quants (#1807)
* metal : improve q4_K
28.3 -> 26.0 ms/token by avoiding a branch in the
calculation of the scales.
* metal : small improvement for Q4_K
* metal : still optimizing Q4_K
This commit pushes it down to 25.3 ms / token.
The crazy idea of using 6 bits for the scales is really costly on
Metal: if I remove the bit fiddling necessary to make the block
scales, time goes almost to the Q4_0 23 ms/token.
Before pushing the k-quants upstream I had a Q4_K variant that
had used 8-bit scales. It wasn't more accurate, used 0.125 bits more per weight,
was running slightly slower on the CPU (due to the larger model size
and being memory bound there), and the difference was entirely
negligible under CUDA. So, I decided to publish the version with 6-bit
scales. Perhaps I should re-consider and change to 8-bit scales?
* metal : some more optimizations
Q2_K: 25.4 ms/token
Q6_K: 27.3 ms/token
Q4_0: 22.8 ms/token
Q4_1: 23.1 ms/token
* metal : Q3_K support
Something is not quite right yet.
* metal : Q5_K support
Initial version achieves 31.2 ms/token, 210 GB/s
* metal : still not able to figure out why q3_K does not work
* Minor
* metal : yet another failed attempt to make q3_K work
* metal : optimize Q5_K
31.2 ms -> 27.8 ms.
250 GB/s.
* metal : q3_K still not working
Adding a heavily commented q3_K metal kernel to explain
my obviously faulty logic. Perhaps someone could spot the issue?
* metal : q3_K finally working
Not optimized at all.
What was the issue? The scales are not 4-bytes aligned,
and I was accessing them with a uint32_t pointer.
When I tried that on CUDA, I got an error (illegal memory access)
and added a memcpy to a local array of 3 uint32_t's.
But on Metal it told me there is no memcpy, so I tried
accessing directly. There is no error, just garbage results.
At some point I did try accessing the scales with an uint16_t
pointer (the scales are for sure 2-byte aligned), but was
still getting garbage. I guess, there must have been another bug.
No access to scales is via a uint16_t pointer and, after starting
from scratch from the C dequantize function, it finally works.
* metal : Q3_K 1st optimization pass
* metal : Q3_K second optimization pass - 29.6 ms/token
* metal : Q3_K cleanup
* metal : fixed accidentally broken Q2_K
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-12 21:39:21 +02:00
|
|
|
|
2023-07-20 14:18:43 +02:00
|
|
|
float dall = dh[0];
|
|
|
|
float dmin = dh[1];
|
|
|
|
sumf[row] += dall * ((acc1[0] + 1.f/256.f * acc1[1]) * sc8[0] +
|
|
|
|
(acc1[2] + 1.f/256.f * acc1[3]) * sc8[1] * 1.f/16.f +
|
|
|
|
(acc2[0] + 1.f/256.f * acc2[1]) * sc8[4] +
|
|
|
|
(acc2[2] + 1.f/256.f * acc2[3]) * sc8[5] * 1.f/16.f) -
|
|
|
|
dmin * (sumy[0] * sc8[2] + sumy[1] * sc8[3] + sumy[2] * sc8[6] + sumy[3] * sc8[7]);
|
|
|
|
|
|
|
|
q1 += step;
|
|
|
|
sc += step;
|
|
|
|
dh += step;
|
2023-06-08 09:08:23 +02:00
|
|
|
}
|
|
|
|
|
2023-07-20 14:18:43 +02:00
|
|
|
y4 += 4 * QK_K;
|
|
|
|
}
|
|
|
|
|
|
|
|
for (int row = 0; row < N_DST; ++row) {
|
|
|
|
all_sum = simd_sum(sumf[row]);
|
|
|
|
if (tiisg == 0) {
|
2023-12-07 21:26:54 +01:00
|
|
|
dst[r1*ne0 + im*ne0*ne1 + first_row + row] = all_sum;
|
2023-07-20 14:18:43 +02:00
|
|
|
}
|
2023-06-08 09:08:23 +02:00
|
|
|
}
|
2023-07-20 14:18:43 +02:00
|
|
|
}
|
k-quants : support for super-block size of 64 (#2001)
* k_quants: WIP super-blocks with 64 weights
* k_quants: WIP super-blocks with 64 weights
Q6_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q4_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q2_K scalar and AVX2 works. Q2_K is way too slow (it is actually slower
than the scalar implementation)
* k_quants: WIP super-blocks with 64 weights
Q3_K scalar and AVX2 works.
* k_quants: WIP super-blocks with 64 weights
Q5_K scalar and AVX2 works, and with that all
k_quants are done on AVX2 and scalar
* k_quants: WIP super-blocks with 64 weights
Q6_K working on CUDA. Cannot make it run quite as gast as
with super-blocks with 256 weigths: 8% slower on 4080,
20% slower on the 1660 (but there we fit 1 less layer on the
GPU because pf the larger model size), so some fraction of
these 20% is due to that,
* k_quants: WIP super-blocks with 64 weights
Q4_K working on CUDA. ~10% slower on GTX-1660,
16% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q2_K working on CUDA. ~3% slower on GTX-1660,
10% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q3_K working on CUDA.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on CUDA, and with this CUDA is done.
* k_quants: WIP super-blocks with 64 weights
Q6_K working on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Q4_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q2_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q3_K working on ARM_NEON, but quite a bit slower than 256 weights.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on ARM_NEON, but quite a bit slower than 256 weights.
With that, we have full support for ARM_NEON, although
performance is not quite there.
* k_quants: WIP super-blocks with 64 weights
Slightly more efficient Q3_K and Q5_K
* k_quants: WIP super-blocks with 64 weights
Another small improvement for Q3_K and Q5_K on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Yet another speedup for Q5_K on ARM_NEON.
We are now within 10% of the QK_K = 256 version.
* k_quants: WIP super-blocks with 64 weights
* We are able to pass preprocessor macros to the Metal
compiler
* Q6_K works and is actually slightly more efficient than
the QK_K = 256 version (25.2 ms vs 25.8 ms)
* k_quants: WIP super-blocks with 64 weights
Q4_K works on Metal and is actually slightly faster
than QK_K = 256 (21.95 ms vs 24.0 ms).
* k_quants: WIP super-blocks with 64 weights
Q2_K works on Metal and is very slightly faster
than QK_K = 256 (23.8 ms vs 24.2 ms).
* k_quants: WIP super-blocks with 64 weights
Q3_K works on Metal and is slightly faster
than QK_K = 256 (26.6 ms vs 28.3 ms).
* k_quants: WIP super-blocks with 64 weights
Q5_K works on Metal and is slightly faster
than QK_K = 256 (23.7 ms vs 26.3 ms).
* k_quants: call them _K, not _k, also on Metal
* k_quants: correctly define QK_K in llama.cpp
* Fixed bug in q4_K quantization added with the 64-block addition
* Simplify via lambda
* k_quants: swicth Q3_K to 4-bit scales when QK_K = 64
Otherwise there isn't much benefit from this
quantization type. There is some very slight loss
in accuracy, but we reduce size by ~7%.
E.g., for OpenLLaMA-3B, Q3_K_S perplexity is
8.6131 with 8-bit scales and 8.6352 with 4-bit,
while file size decreases from 1.53G to 1.44G.
* k_quants: switch Q4_K to 4-bit scales when QK_K = 64
Here the loss in accuracy is greater than for Q3_K,
but the Q4_K points still move further to the left on
the perplexity vs size curve.
* k_quants: forgot to add the Metal changes in last commit
* k_quants: change Q5_K to be type 0 when QK_K = 64
Still needs AVX2 implementation
* k_quants: AVX2 implementation for new 64-weight Q5_K
* k_quants: 10% faster ARM_NEON Q5_K dot product
* k_quants: fixed issue caused by merging with master
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-26 18:43:07 +02:00
|
|
|
#else
|
2023-12-13 13:04:25 +01:00
|
|
|
void kernel_mul_mv_q4_K_f32_impl(
|
2023-07-20 14:18:43 +02:00
|
|
|
device const void * src0,
|
|
|
|
device const float * src1,
|
|
|
|
device float * dst,
|
|
|
|
constant int64_t & ne00,
|
2023-12-13 13:04:25 +01:00
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant int64_t & ne10,
|
|
|
|
constant int64_t & ne12,
|
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
|
|
|
constant uint & r2,
|
|
|
|
constant uint & r3,
|
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
2024-01-02 09:57:44 +01:00
|
|
|
uint tiisg[[thread_index_in_simdgroup]],
|
|
|
|
uint sgitg[[simdgroup_index_in_threadgroup]]) {
|
k-quants : support for super-block size of 64 (#2001)
* k_quants: WIP super-blocks with 64 weights
* k_quants: WIP super-blocks with 64 weights
Q6_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q4_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q2_K scalar and AVX2 works. Q2_K is way too slow (it is actually slower
than the scalar implementation)
* k_quants: WIP super-blocks with 64 weights
Q3_K scalar and AVX2 works.
* k_quants: WIP super-blocks with 64 weights
Q5_K scalar and AVX2 works, and with that all
k_quants are done on AVX2 and scalar
* k_quants: WIP super-blocks with 64 weights
Q6_K working on CUDA. Cannot make it run quite as gast as
with super-blocks with 256 weigths: 8% slower on 4080,
20% slower on the 1660 (but there we fit 1 less layer on the
GPU because pf the larger model size), so some fraction of
these 20% is due to that,
* k_quants: WIP super-blocks with 64 weights
Q4_K working on CUDA. ~10% slower on GTX-1660,
16% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q2_K working on CUDA. ~3% slower on GTX-1660,
10% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q3_K working on CUDA.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on CUDA, and with this CUDA is done.
* k_quants: WIP super-blocks with 64 weights
Q6_K working on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Q4_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q2_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q3_K working on ARM_NEON, but quite a bit slower than 256 weights.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on ARM_NEON, but quite a bit slower than 256 weights.
With that, we have full support for ARM_NEON, although
performance is not quite there.
* k_quants: WIP super-blocks with 64 weights
Slightly more efficient Q3_K and Q5_K
* k_quants: WIP super-blocks with 64 weights
Another small improvement for Q3_K and Q5_K on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Yet another speedup for Q5_K on ARM_NEON.
We are now within 10% of the QK_K = 256 version.
* k_quants: WIP super-blocks with 64 weights
* We are able to pass preprocessor macros to the Metal
compiler
* Q6_K works and is actually slightly more efficient than
the QK_K = 256 version (25.2 ms vs 25.8 ms)
* k_quants: WIP super-blocks with 64 weights
Q4_K works on Metal and is actually slightly faster
than QK_K = 256 (21.95 ms vs 24.0 ms).
* k_quants: WIP super-blocks with 64 weights
Q2_K works on Metal and is very slightly faster
than QK_K = 256 (23.8 ms vs 24.2 ms).
* k_quants: WIP super-blocks with 64 weights
Q3_K works on Metal and is slightly faster
than QK_K = 256 (26.6 ms vs 28.3 ms).
* k_quants: WIP super-blocks with 64 weights
Q5_K works on Metal and is slightly faster
than QK_K = 256 (23.7 ms vs 26.3 ms).
* k_quants: call them _K, not _k, also on Metal
* k_quants: correctly define QK_K in llama.cpp
* Fixed bug in q4_K quantization added with the 64-block addition
* Simplify via lambda
* k_quants: swicth Q3_K to 4-bit scales when QK_K = 64
Otherwise there isn't much benefit from this
quantization type. There is some very slight loss
in accuracy, but we reduce size by ~7%.
E.g., for OpenLLaMA-3B, Q3_K_S perplexity is
8.6131 with 8-bit scales and 8.6352 with 4-bit,
while file size decreases from 1.53G to 1.44G.
* k_quants: switch Q4_K to 4-bit scales when QK_K = 64
Here the loss in accuracy is greater than for Q3_K,
but the Q4_K points still move further to the left on
the perplexity vs size curve.
* k_quants: forgot to add the Metal changes in last commit
* k_quants: change Q5_K to be type 0 when QK_K = 64
Still needs AVX2 implementation
* k_quants: AVX2 implementation for new 64-weight Q5_K
* k_quants: 10% faster ARM_NEON Q5_K dot product
* k_quants: fixed issue caused by merging with master
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-26 18:43:07 +02:00
|
|
|
|
2023-07-20 14:18:43 +02:00
|
|
|
const int ix = tiisg/4; // 0...7
|
|
|
|
const int it = tiisg%4; // 0...3
|
k-quants : support for super-block size of 64 (#2001)
* k_quants: WIP super-blocks with 64 weights
* k_quants: WIP super-blocks with 64 weights
Q6_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q4_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q2_K scalar and AVX2 works. Q2_K is way too slow (it is actually slower
than the scalar implementation)
* k_quants: WIP super-blocks with 64 weights
Q3_K scalar and AVX2 works.
* k_quants: WIP super-blocks with 64 weights
Q5_K scalar and AVX2 works, and with that all
k_quants are done on AVX2 and scalar
* k_quants: WIP super-blocks with 64 weights
Q6_K working on CUDA. Cannot make it run quite as gast as
with super-blocks with 256 weigths: 8% slower on 4080,
20% slower on the 1660 (but there we fit 1 less layer on the
GPU because pf the larger model size), so some fraction of
these 20% is due to that,
* k_quants: WIP super-blocks with 64 weights
Q4_K working on CUDA. ~10% slower on GTX-1660,
16% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q2_K working on CUDA. ~3% slower on GTX-1660,
10% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q3_K working on CUDA.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on CUDA, and with this CUDA is done.
* k_quants: WIP super-blocks with 64 weights
Q6_K working on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Q4_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q2_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q3_K working on ARM_NEON, but quite a bit slower than 256 weights.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on ARM_NEON, but quite a bit slower than 256 weights.
With that, we have full support for ARM_NEON, although
performance is not quite there.
* k_quants: WIP super-blocks with 64 weights
Slightly more efficient Q3_K and Q5_K
* k_quants: WIP super-blocks with 64 weights
Another small improvement for Q3_K and Q5_K on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Yet another speedup for Q5_K on ARM_NEON.
We are now within 10% of the QK_K = 256 version.
* k_quants: WIP super-blocks with 64 weights
* We are able to pass preprocessor macros to the Metal
compiler
* Q6_K works and is actually slightly more efficient than
the QK_K = 256 version (25.2 ms vs 25.8 ms)
* k_quants: WIP super-blocks with 64 weights
Q4_K works on Metal and is actually slightly faster
than QK_K = 256 (21.95 ms vs 24.0 ms).
* k_quants: WIP super-blocks with 64 weights
Q2_K works on Metal and is very slightly faster
than QK_K = 256 (23.8 ms vs 24.2 ms).
* k_quants: WIP super-blocks with 64 weights
Q3_K works on Metal and is slightly faster
than QK_K = 256 (26.6 ms vs 28.3 ms).
* k_quants: WIP super-blocks with 64 weights
Q5_K works on Metal and is slightly faster
than QK_K = 256 (23.7 ms vs 26.3 ms).
* k_quants: call them _K, not _k, also on Metal
* k_quants: correctly define QK_K in llama.cpp
* Fixed bug in q4_K quantization added with the 64-block addition
* Simplify via lambda
* k_quants: swicth Q3_K to 4-bit scales when QK_K = 64
Otherwise there isn't much benefit from this
quantization type. There is some very slight loss
in accuracy, but we reduce size by ~7%.
E.g., for OpenLLaMA-3B, Q3_K_S perplexity is
8.6131 with 8-bit scales and 8.6352 with 4-bit,
while file size decreases from 1.53G to 1.44G.
* k_quants: switch Q4_K to 4-bit scales when QK_K = 64
Here the loss in accuracy is greater than for Q3_K,
but the Q4_K points still move further to the left on
the perplexity vs size curve.
* k_quants: forgot to add the Metal changes in last commit
* k_quants: change Q5_K to be type 0 when QK_K = 64
Still needs AVX2 implementation
* k_quants: AVX2 implementation for new 64-weight Q5_K
* k_quants: 10% faster ARM_NEON Q5_K dot product
* k_quants: fixed issue caused by merging with master
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-26 18:43:07 +02:00
|
|
|
|
2023-07-20 14:18:43 +02:00
|
|
|
const int nb = ne00/QK_K;
|
|
|
|
const int r0 = tgpig.x;
|
|
|
|
const int r1 = tgpig.y;
|
2023-12-07 21:26:54 +01:00
|
|
|
const int im = tgpig.z;
|
2024-01-02 09:57:44 +01:00
|
|
|
const int first_row = r0 * N_DST;
|
2023-07-20 14:18:43 +02:00
|
|
|
const int ib_row = first_row * nb;
|
2023-12-07 21:26:54 +01:00
|
|
|
|
|
|
|
const uint i12 = im%ne12;
|
|
|
|
const uint i13 = im/ne12;
|
|
|
|
|
|
|
|
const uint offset0 = (i12/r2)*(nb*ne01) + (i13/r3)*(nb*ne01*ne02);
|
|
|
|
|
2023-08-16 22:07:04 +02:00
|
|
|
device const block_q4_K * x = (device const block_q4_K *) src0 + ib_row + offset0;
|
2023-12-07 21:26:54 +01:00
|
|
|
device const float * y = (device const float *) src1 + r1*ne10 + im*ne00*ne1;
|
|
|
|
|
2023-07-20 14:18:43 +02:00
|
|
|
float yl[8];
|
|
|
|
float yh[8];
|
|
|
|
float sumf[N_DST]={0.f}, all_sum;
|
k-quants : support for super-block size of 64 (#2001)
* k_quants: WIP super-blocks with 64 weights
* k_quants: WIP super-blocks with 64 weights
Q6_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q4_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q2_K scalar and AVX2 works. Q2_K is way too slow (it is actually slower
than the scalar implementation)
* k_quants: WIP super-blocks with 64 weights
Q3_K scalar and AVX2 works.
* k_quants: WIP super-blocks with 64 weights
Q5_K scalar and AVX2 works, and with that all
k_quants are done on AVX2 and scalar
* k_quants: WIP super-blocks with 64 weights
Q6_K working on CUDA. Cannot make it run quite as gast as
with super-blocks with 256 weigths: 8% slower on 4080,
20% slower on the 1660 (but there we fit 1 less layer on the
GPU because pf the larger model size), so some fraction of
these 20% is due to that,
* k_quants: WIP super-blocks with 64 weights
Q4_K working on CUDA. ~10% slower on GTX-1660,
16% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q2_K working on CUDA. ~3% slower on GTX-1660,
10% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q3_K working on CUDA.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on CUDA, and with this CUDA is done.
* k_quants: WIP super-blocks with 64 weights
Q6_K working on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Q4_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q2_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q3_K working on ARM_NEON, but quite a bit slower than 256 weights.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on ARM_NEON, but quite a bit slower than 256 weights.
With that, we have full support for ARM_NEON, although
performance is not quite there.
* k_quants: WIP super-blocks with 64 weights
Slightly more efficient Q3_K and Q5_K
* k_quants: WIP super-blocks with 64 weights
Another small improvement for Q3_K and Q5_K on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Yet another speedup for Q5_K on ARM_NEON.
We are now within 10% of the QK_K = 256 version.
* k_quants: WIP super-blocks with 64 weights
* We are able to pass preprocessor macros to the Metal
compiler
* Q6_K works and is actually slightly more efficient than
the QK_K = 256 version (25.2 ms vs 25.8 ms)
* k_quants: WIP super-blocks with 64 weights
Q4_K works on Metal and is actually slightly faster
than QK_K = 256 (21.95 ms vs 24.0 ms).
* k_quants: WIP super-blocks with 64 weights
Q2_K works on Metal and is very slightly faster
than QK_K = 256 (23.8 ms vs 24.2 ms).
* k_quants: WIP super-blocks with 64 weights
Q3_K works on Metal and is slightly faster
than QK_K = 256 (26.6 ms vs 28.3 ms).
* k_quants: WIP super-blocks with 64 weights
Q5_K works on Metal and is slightly faster
than QK_K = 256 (23.7 ms vs 26.3 ms).
* k_quants: call them _K, not _k, also on Metal
* k_quants: correctly define QK_K in llama.cpp
* Fixed bug in q4_K quantization added with the 64-block addition
* Simplify via lambda
* k_quants: swicth Q3_K to 4-bit scales when QK_K = 64
Otherwise there isn't much benefit from this
quantization type. There is some very slight loss
in accuracy, but we reduce size by ~7%.
E.g., for OpenLLaMA-3B, Q3_K_S perplexity is
8.6131 with 8-bit scales and 8.6352 with 4-bit,
while file size decreases from 1.53G to 1.44G.
* k_quants: switch Q4_K to 4-bit scales when QK_K = 64
Here the loss in accuracy is greater than for Q3_K,
but the Q4_K points still move further to the left on
the perplexity vs size curve.
* k_quants: forgot to add the Metal changes in last commit
* k_quants: change Q5_K to be type 0 when QK_K = 64
Still needs AVX2 implementation
* k_quants: AVX2 implementation for new 64-weight Q5_K
* k_quants: 10% faster ARM_NEON Q5_K dot product
* k_quants: fixed issue caused by merging with master
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-26 18:43:07 +02:00
|
|
|
|
2023-07-20 14:18:43 +02:00
|
|
|
const int step = sizeof(block_q4_K) * nb / 2;
|
k-quants : support for super-block size of 64 (#2001)
* k_quants: WIP super-blocks with 64 weights
* k_quants: WIP super-blocks with 64 weights
Q6_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q4_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q2_K scalar and AVX2 works. Q2_K is way too slow (it is actually slower
than the scalar implementation)
* k_quants: WIP super-blocks with 64 weights
Q3_K scalar and AVX2 works.
* k_quants: WIP super-blocks with 64 weights
Q5_K scalar and AVX2 works, and with that all
k_quants are done on AVX2 and scalar
* k_quants: WIP super-blocks with 64 weights
Q6_K working on CUDA. Cannot make it run quite as gast as
with super-blocks with 256 weigths: 8% slower on 4080,
20% slower on the 1660 (but there we fit 1 less layer on the
GPU because pf the larger model size), so some fraction of
these 20% is due to that,
* k_quants: WIP super-blocks with 64 weights
Q4_K working on CUDA. ~10% slower on GTX-1660,
16% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q2_K working on CUDA. ~3% slower on GTX-1660,
10% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q3_K working on CUDA.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on CUDA, and with this CUDA is done.
* k_quants: WIP super-blocks with 64 weights
Q6_K working on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Q4_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q2_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q3_K working on ARM_NEON, but quite a bit slower than 256 weights.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on ARM_NEON, but quite a bit slower than 256 weights.
With that, we have full support for ARM_NEON, although
performance is not quite there.
* k_quants: WIP super-blocks with 64 weights
Slightly more efficient Q3_K and Q5_K
* k_quants: WIP super-blocks with 64 weights
Another small improvement for Q3_K and Q5_K on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Yet another speedup for Q5_K on ARM_NEON.
We are now within 10% of the QK_K = 256 version.
* k_quants: WIP super-blocks with 64 weights
* We are able to pass preprocessor macros to the Metal
compiler
* Q6_K works and is actually slightly more efficient than
the QK_K = 256 version (25.2 ms vs 25.8 ms)
* k_quants: WIP super-blocks with 64 weights
Q4_K works on Metal and is actually slightly faster
than QK_K = 256 (21.95 ms vs 24.0 ms).
* k_quants: WIP super-blocks with 64 weights
Q2_K works on Metal and is very slightly faster
than QK_K = 256 (23.8 ms vs 24.2 ms).
* k_quants: WIP super-blocks with 64 weights
Q3_K works on Metal and is slightly faster
than QK_K = 256 (26.6 ms vs 28.3 ms).
* k_quants: WIP super-blocks with 64 weights
Q5_K works on Metal and is slightly faster
than QK_K = 256 (23.7 ms vs 26.3 ms).
* k_quants: call them _K, not _k, also on Metal
* k_quants: correctly define QK_K in llama.cpp
* Fixed bug in q4_K quantization added with the 64-block addition
* Simplify via lambda
* k_quants: swicth Q3_K to 4-bit scales when QK_K = 64
Otherwise there isn't much benefit from this
quantization type. There is some very slight loss
in accuracy, but we reduce size by ~7%.
E.g., for OpenLLaMA-3B, Q3_K_S perplexity is
8.6131 with 8-bit scales and 8.6352 with 4-bit,
while file size decreases from 1.53G to 1.44G.
* k_quants: switch Q4_K to 4-bit scales when QK_K = 64
Here the loss in accuracy is greater than for Q3_K,
but the Q4_K points still move further to the left on
the perplexity vs size curve.
* k_quants: forgot to add the Metal changes in last commit
* k_quants: change Q5_K to be type 0 when QK_K = 64
Still needs AVX2 implementation
* k_quants: AVX2 implementation for new 64-weight Q5_K
* k_quants: 10% faster ARM_NEON Q5_K dot product
* k_quants: fixed issue caused by merging with master
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-26 18:43:07 +02:00
|
|
|
|
2023-07-20 14:18:43 +02:00
|
|
|
device const float * y4 = y + ix * QK_K + 8 * it;
|
k-quants : support for super-block size of 64 (#2001)
* k_quants: WIP super-blocks with 64 weights
* k_quants: WIP super-blocks with 64 weights
Q6_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q4_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q2_K scalar and AVX2 works. Q2_K is way too slow (it is actually slower
than the scalar implementation)
* k_quants: WIP super-blocks with 64 weights
Q3_K scalar and AVX2 works.
* k_quants: WIP super-blocks with 64 weights
Q5_K scalar and AVX2 works, and with that all
k_quants are done on AVX2 and scalar
* k_quants: WIP super-blocks with 64 weights
Q6_K working on CUDA. Cannot make it run quite as gast as
with super-blocks with 256 weigths: 8% slower on 4080,
20% slower on the 1660 (but there we fit 1 less layer on the
GPU because pf the larger model size), so some fraction of
these 20% is due to that,
* k_quants: WIP super-blocks with 64 weights
Q4_K working on CUDA. ~10% slower on GTX-1660,
16% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q2_K working on CUDA. ~3% slower on GTX-1660,
10% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q3_K working on CUDA.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on CUDA, and with this CUDA is done.
* k_quants: WIP super-blocks with 64 weights
Q6_K working on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Q4_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q2_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q3_K working on ARM_NEON, but quite a bit slower than 256 weights.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on ARM_NEON, but quite a bit slower than 256 weights.
With that, we have full support for ARM_NEON, although
performance is not quite there.
* k_quants: WIP super-blocks with 64 weights
Slightly more efficient Q3_K and Q5_K
* k_quants: WIP super-blocks with 64 weights
Another small improvement for Q3_K and Q5_K on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Yet another speedup for Q5_K on ARM_NEON.
We are now within 10% of the QK_K = 256 version.
* k_quants: WIP super-blocks with 64 weights
* We are able to pass preprocessor macros to the Metal
compiler
* Q6_K works and is actually slightly more efficient than
the QK_K = 256 version (25.2 ms vs 25.8 ms)
* k_quants: WIP super-blocks with 64 weights
Q4_K works on Metal and is actually slightly faster
than QK_K = 256 (21.95 ms vs 24.0 ms).
* k_quants: WIP super-blocks with 64 weights
Q2_K works on Metal and is very slightly faster
than QK_K = 256 (23.8 ms vs 24.2 ms).
* k_quants: WIP super-blocks with 64 weights
Q3_K works on Metal and is slightly faster
than QK_K = 256 (26.6 ms vs 28.3 ms).
* k_quants: WIP super-blocks with 64 weights
Q5_K works on Metal and is slightly faster
than QK_K = 256 (23.7 ms vs 26.3 ms).
* k_quants: call them _K, not _k, also on Metal
* k_quants: correctly define QK_K in llama.cpp
* Fixed bug in q4_K quantization added with the 64-block addition
* Simplify via lambda
* k_quants: swicth Q3_K to 4-bit scales when QK_K = 64
Otherwise there isn't much benefit from this
quantization type. There is some very slight loss
in accuracy, but we reduce size by ~7%.
E.g., for OpenLLaMA-3B, Q3_K_S perplexity is
8.6131 with 8-bit scales and 8.6352 with 4-bit,
while file size decreases from 1.53G to 1.44G.
* k_quants: switch Q4_K to 4-bit scales when QK_K = 64
Here the loss in accuracy is greater than for Q3_K,
but the Q4_K points still move further to the left on
the perplexity vs size curve.
* k_quants: forgot to add the Metal changes in last commit
* k_quants: change Q5_K to be type 0 when QK_K = 64
Still needs AVX2 implementation
* k_quants: AVX2 implementation for new 64-weight Q5_K
* k_quants: 10% faster ARM_NEON Q5_K dot product
* k_quants: fixed issue caused by merging with master
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-26 18:43:07 +02:00
|
|
|
|
2023-07-20 14:18:43 +02:00
|
|
|
uint16_t sc16[4];
|
k-quants : support for super-block size of 64 (#2001)
* k_quants: WIP super-blocks with 64 weights
* k_quants: WIP super-blocks with 64 weights
Q6_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q4_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q2_K scalar and AVX2 works. Q2_K is way too slow (it is actually slower
than the scalar implementation)
* k_quants: WIP super-blocks with 64 weights
Q3_K scalar and AVX2 works.
* k_quants: WIP super-blocks with 64 weights
Q5_K scalar and AVX2 works, and with that all
k_quants are done on AVX2 and scalar
* k_quants: WIP super-blocks with 64 weights
Q6_K working on CUDA. Cannot make it run quite as gast as
with super-blocks with 256 weigths: 8% slower on 4080,
20% slower on the 1660 (but there we fit 1 less layer on the
GPU because pf the larger model size), so some fraction of
these 20% is due to that,
* k_quants: WIP super-blocks with 64 weights
Q4_K working on CUDA. ~10% slower on GTX-1660,
16% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q2_K working on CUDA. ~3% slower on GTX-1660,
10% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q3_K working on CUDA.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on CUDA, and with this CUDA is done.
* k_quants: WIP super-blocks with 64 weights
Q6_K working on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Q4_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q2_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q3_K working on ARM_NEON, but quite a bit slower than 256 weights.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on ARM_NEON, but quite a bit slower than 256 weights.
With that, we have full support for ARM_NEON, although
performance is not quite there.
* k_quants: WIP super-blocks with 64 weights
Slightly more efficient Q3_K and Q5_K
* k_quants: WIP super-blocks with 64 weights
Another small improvement for Q3_K and Q5_K on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Yet another speedup for Q5_K on ARM_NEON.
We are now within 10% of the QK_K = 256 version.
* k_quants: WIP super-blocks with 64 weights
* We are able to pass preprocessor macros to the Metal
compiler
* Q6_K works and is actually slightly more efficient than
the QK_K = 256 version (25.2 ms vs 25.8 ms)
* k_quants: WIP super-blocks with 64 weights
Q4_K works on Metal and is actually slightly faster
than QK_K = 256 (21.95 ms vs 24.0 ms).
* k_quants: WIP super-blocks with 64 weights
Q2_K works on Metal and is very slightly faster
than QK_K = 256 (23.8 ms vs 24.2 ms).
* k_quants: WIP super-blocks with 64 weights
Q3_K works on Metal and is slightly faster
than QK_K = 256 (26.6 ms vs 28.3 ms).
* k_quants: WIP super-blocks with 64 weights
Q5_K works on Metal and is slightly faster
than QK_K = 256 (23.7 ms vs 26.3 ms).
* k_quants: call them _K, not _k, also on Metal
* k_quants: correctly define QK_K in llama.cpp
* Fixed bug in q4_K quantization added with the 64-block addition
* Simplify via lambda
* k_quants: swicth Q3_K to 4-bit scales when QK_K = 64
Otherwise there isn't much benefit from this
quantization type. There is some very slight loss
in accuracy, but we reduce size by ~7%.
E.g., for OpenLLaMA-3B, Q3_K_S perplexity is
8.6131 with 8-bit scales and 8.6352 with 4-bit,
while file size decreases from 1.53G to 1.44G.
* k_quants: switch Q4_K to 4-bit scales when QK_K = 64
Here the loss in accuracy is greater than for Q3_K,
but the Q4_K points still move further to the left on
the perplexity vs size curve.
* k_quants: forgot to add the Metal changes in last commit
* k_quants: change Q5_K to be type 0 when QK_K = 64
Still needs AVX2 implementation
* k_quants: AVX2 implementation for new 64-weight Q5_K
* k_quants: 10% faster ARM_NEON Q5_K dot product
* k_quants: fixed issue caused by merging with master
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-26 18:43:07 +02:00
|
|
|
|
2023-07-20 14:18:43 +02:00
|
|
|
for (int ib = ix; ib < nb; ib += 8) {
|
|
|
|
|
|
|
|
float2 sumy = {0.f, 0.f};
|
|
|
|
for (int i = 0; i < 8; ++i) {
|
|
|
|
yl[i] = y4[i+ 0]; sumy[0] += yl[i];
|
|
|
|
yh[i] = y4[i+32]; sumy[1] += yh[i];
|
k-quants : support for super-block size of 64 (#2001)
* k_quants: WIP super-blocks with 64 weights
* k_quants: WIP super-blocks with 64 weights
Q6_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q4_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q2_K scalar and AVX2 works. Q2_K is way too slow (it is actually slower
than the scalar implementation)
* k_quants: WIP super-blocks with 64 weights
Q3_K scalar and AVX2 works.
* k_quants: WIP super-blocks with 64 weights
Q5_K scalar and AVX2 works, and with that all
k_quants are done on AVX2 and scalar
* k_quants: WIP super-blocks with 64 weights
Q6_K working on CUDA. Cannot make it run quite as gast as
with super-blocks with 256 weigths: 8% slower on 4080,
20% slower on the 1660 (but there we fit 1 less layer on the
GPU because pf the larger model size), so some fraction of
these 20% is due to that,
* k_quants: WIP super-blocks with 64 weights
Q4_K working on CUDA. ~10% slower on GTX-1660,
16% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q2_K working on CUDA. ~3% slower on GTX-1660,
10% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q3_K working on CUDA.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on CUDA, and with this CUDA is done.
* k_quants: WIP super-blocks with 64 weights
Q6_K working on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Q4_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q2_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q3_K working on ARM_NEON, but quite a bit slower than 256 weights.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on ARM_NEON, but quite a bit slower than 256 weights.
With that, we have full support for ARM_NEON, although
performance is not quite there.
* k_quants: WIP super-blocks with 64 weights
Slightly more efficient Q3_K and Q5_K
* k_quants: WIP super-blocks with 64 weights
Another small improvement for Q3_K and Q5_K on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Yet another speedup for Q5_K on ARM_NEON.
We are now within 10% of the QK_K = 256 version.
* k_quants: WIP super-blocks with 64 weights
* We are able to pass preprocessor macros to the Metal
compiler
* Q6_K works and is actually slightly more efficient than
the QK_K = 256 version (25.2 ms vs 25.8 ms)
* k_quants: WIP super-blocks with 64 weights
Q4_K works on Metal and is actually slightly faster
than QK_K = 256 (21.95 ms vs 24.0 ms).
* k_quants: WIP super-blocks with 64 weights
Q2_K works on Metal and is very slightly faster
than QK_K = 256 (23.8 ms vs 24.2 ms).
* k_quants: WIP super-blocks with 64 weights
Q3_K works on Metal and is slightly faster
than QK_K = 256 (26.6 ms vs 28.3 ms).
* k_quants: WIP super-blocks with 64 weights
Q5_K works on Metal and is slightly faster
than QK_K = 256 (23.7 ms vs 26.3 ms).
* k_quants: call them _K, not _k, also on Metal
* k_quants: correctly define QK_K in llama.cpp
* Fixed bug in q4_K quantization added with the 64-block addition
* Simplify via lambda
* k_quants: swicth Q3_K to 4-bit scales when QK_K = 64
Otherwise there isn't much benefit from this
quantization type. There is some very slight loss
in accuracy, but we reduce size by ~7%.
E.g., for OpenLLaMA-3B, Q3_K_S perplexity is
8.6131 with 8-bit scales and 8.6352 with 4-bit,
while file size decreases from 1.53G to 1.44G.
* k_quants: switch Q4_K to 4-bit scales when QK_K = 64
Here the loss in accuracy is greater than for Q3_K,
but the Q4_K points still move further to the left on
the perplexity vs size curve.
* k_quants: forgot to add the Metal changes in last commit
* k_quants: change Q5_K to be type 0 when QK_K = 64
Still needs AVX2 implementation
* k_quants: AVX2 implementation for new 64-weight Q5_K
* k_quants: 10% faster ARM_NEON Q5_K dot product
* k_quants: fixed issue caused by merging with master
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-26 18:43:07 +02:00
|
|
|
}
|
Metal implementation for all k_quants (#1807)
* metal : improve q4_K
28.3 -> 26.0 ms/token by avoiding a branch in the
calculation of the scales.
* metal : small improvement for Q4_K
* metal : still optimizing Q4_K
This commit pushes it down to 25.3 ms / token.
The crazy idea of using 6 bits for the scales is really costly on
Metal: if I remove the bit fiddling necessary to make the block
scales, time goes almost to the Q4_0 23 ms/token.
Before pushing the k-quants upstream I had a Q4_K variant that
had used 8-bit scales. It wasn't more accurate, used 0.125 bits more per weight,
was running slightly slower on the CPU (due to the larger model size
and being memory bound there), and the difference was entirely
negligible under CUDA. So, I decided to publish the version with 6-bit
scales. Perhaps I should re-consider and change to 8-bit scales?
* metal : some more optimizations
Q2_K: 25.4 ms/token
Q6_K: 27.3 ms/token
Q4_0: 22.8 ms/token
Q4_1: 23.1 ms/token
* metal : Q3_K support
Something is not quite right yet.
* metal : Q5_K support
Initial version achieves 31.2 ms/token, 210 GB/s
* metal : still not able to figure out why q3_K does not work
* Minor
* metal : yet another failed attempt to make q3_K work
* metal : optimize Q5_K
31.2 ms -> 27.8 ms.
250 GB/s.
* metal : q3_K still not working
Adding a heavily commented q3_K metal kernel to explain
my obviously faulty logic. Perhaps someone could spot the issue?
* metal : q3_K finally working
Not optimized at all.
What was the issue? The scales are not 4-bytes aligned,
and I was accessing them with a uint32_t pointer.
When I tried that on CUDA, I got an error (illegal memory access)
and added a memcpy to a local array of 3 uint32_t's.
But on Metal it told me there is no memcpy, so I tried
accessing directly. There is no error, just garbage results.
At some point I did try accessing the scales with an uint16_t
pointer (the scales are for sure 2-byte aligned), but was
still getting garbage. I guess, there must have been another bug.
No access to scales is via a uint16_t pointer and, after starting
from scratch from the C dequantize function, it finally works.
* metal : Q3_K 1st optimization pass
* metal : Q3_K second optimization pass - 29.6 ms/token
* metal : Q3_K cleanup
* metal : fixed accidentally broken Q2_K
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-12 21:39:21 +02:00
|
|
|
|
2023-07-20 14:18:43 +02:00
|
|
|
device const uint16_t * sc = (device const uint16_t *)x[ib].scales;
|
|
|
|
device const uint16_t * qs = (device const uint16_t *)x[ib].qs + 4 * it;
|
|
|
|
device const half * dh = x[ib].d;
|
2023-06-08 09:08:23 +02:00
|
|
|
|
2023-07-20 14:18:43 +02:00
|
|
|
for (int row = 0; row < N_DST; row++) {
|
|
|
|
|
|
|
|
sc16[0] = sc[0] & 0x000f;
|
|
|
|
sc16[1] = sc[0] & 0x0f00;
|
|
|
|
sc16[2] = sc[0] & 0x00f0;
|
|
|
|
sc16[3] = sc[0] & 0xf000;
|
|
|
|
|
|
|
|
float2 acc1 = {0.f, 0.f};
|
|
|
|
float2 acc2 = {0.f, 0.f};
|
|
|
|
for (int i = 0; i < 8; i += 2) {
|
|
|
|
acc1[0] += yl[i+0] * (qs[i/2] & 0x000F);
|
|
|
|
acc1[1] += yl[i+1] * (qs[i/2] & 0x0F00);
|
|
|
|
acc2[0] += yh[i+0] * (qs[i/2] & 0x00F0);
|
|
|
|
acc2[1] += yh[i+1] * (qs[i/2] & 0xF000);
|
|
|
|
}
|
|
|
|
|
|
|
|
float dall = dh[0];
|
|
|
|
float dmin = dh[1];
|
|
|
|
sumf[row] += dall * ((acc1[0] + 1.f/256.f * acc1[1]) * sc16[0] +
|
|
|
|
(acc2[0] + 1.f/256.f * acc2[1]) * sc16[1] * 1.f/4096.f) -
|
|
|
|
dmin * 1.f/16.f * (sumy[0] * sc16[2] + sumy[1] * sc16[3] * 1.f/256.f);
|
|
|
|
|
|
|
|
qs += step;
|
|
|
|
sc += step;
|
|
|
|
dh += step;
|
|
|
|
}
|
|
|
|
|
|
|
|
y4 += 8 * QK_K;
|
2023-06-08 09:08:23 +02:00
|
|
|
}
|
|
|
|
|
2023-07-20 14:18:43 +02:00
|
|
|
for (int row = 0; row < N_DST; ++row) {
|
|
|
|
all_sum = simd_sum(sumf[row]);
|
|
|
|
if (tiisg == 0) {
|
2024-01-02 09:57:44 +01:00
|
|
|
dst[r1*ne0 + im*ne0*ne1 + first_row + row] = all_sum;
|
2023-07-20 14:18:43 +02:00
|
|
|
}
|
|
|
|
}
|
2023-06-08 09:08:23 +02:00
|
|
|
}
|
2023-07-20 14:18:43 +02:00
|
|
|
#endif
|
2023-06-08 18:46:22 +02:00
|
|
|
|
2023-12-13 13:04:25 +01:00
|
|
|
[[host_name("kernel_mul_mv_q4_K_f32")]]
|
|
|
|
kernel void kernel_mul_mv_q4_K_f32(
|
Metal implementation for all k_quants (#1807)
* metal : improve q4_K
28.3 -> 26.0 ms/token by avoiding a branch in the
calculation of the scales.
* metal : small improvement for Q4_K
* metal : still optimizing Q4_K
This commit pushes it down to 25.3 ms / token.
The crazy idea of using 6 bits for the scales is really costly on
Metal: if I remove the bit fiddling necessary to make the block
scales, time goes almost to the Q4_0 23 ms/token.
Before pushing the k-quants upstream I had a Q4_K variant that
had used 8-bit scales. It wasn't more accurate, used 0.125 bits more per weight,
was running slightly slower on the CPU (due to the larger model size
and being memory bound there), and the difference was entirely
negligible under CUDA. So, I decided to publish the version with 6-bit
scales. Perhaps I should re-consider and change to 8-bit scales?
* metal : some more optimizations
Q2_K: 25.4 ms/token
Q6_K: 27.3 ms/token
Q4_0: 22.8 ms/token
Q4_1: 23.1 ms/token
* metal : Q3_K support
Something is not quite right yet.
* metal : Q5_K support
Initial version achieves 31.2 ms/token, 210 GB/s
* metal : still not able to figure out why q3_K does not work
* Minor
* metal : yet another failed attempt to make q3_K work
* metal : optimize Q5_K
31.2 ms -> 27.8 ms.
250 GB/s.
* metal : q3_K still not working
Adding a heavily commented q3_K metal kernel to explain
my obviously faulty logic. Perhaps someone could spot the issue?
* metal : q3_K finally working
Not optimized at all.
What was the issue? The scales are not 4-bytes aligned,
and I was accessing them with a uint32_t pointer.
When I tried that on CUDA, I got an error (illegal memory access)
and added a memcpy to a local array of 3 uint32_t's.
But on Metal it told me there is no memcpy, so I tried
accessing directly. There is no error, just garbage results.
At some point I did try accessing the scales with an uint16_t
pointer (the scales are for sure 2-byte aligned), but was
still getting garbage. I guess, there must have been another bug.
No access to scales is via a uint16_t pointer and, after starting
from scratch from the C dequantize function, it finally works.
* metal : Q3_K 1st optimization pass
* metal : Q3_K second optimization pass - 29.6 ms/token
* metal : Q3_K cleanup
* metal : fixed accidentally broken Q2_K
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-12 21:39:21 +02:00
|
|
|
device const void * src0,
|
|
|
|
device const float * src1,
|
|
|
|
device float * dst,
|
|
|
|
constant int64_t & ne00,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant uint64_t & nb00,
|
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant uint64_t & nb02,
|
|
|
|
constant int64_t & ne10,
|
|
|
|
constant int64_t & ne11,
|
|
|
|
constant int64_t & ne12,
|
|
|
|
constant uint64_t & nb10,
|
|
|
|
constant uint64_t & nb11,
|
|
|
|
constant uint64_t & nb12,
|
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
|
|
|
constant uint & r2,
|
|
|
|
constant uint & r3,
|
2023-08-16 22:07:04 +02:00
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
2023-07-20 17:19:45 +02:00
|
|
|
uint tiisg[[thread_index_in_simdgroup]],
|
|
|
|
uint sgitg[[simdgroup_index_in_threadgroup]]) {
|
Metal implementation for all k_quants (#1807)
* metal : improve q4_K
28.3 -> 26.0 ms/token by avoiding a branch in the
calculation of the scales.
* metal : small improvement for Q4_K
* metal : still optimizing Q4_K
This commit pushes it down to 25.3 ms / token.
The crazy idea of using 6 bits for the scales is really costly on
Metal: if I remove the bit fiddling necessary to make the block
scales, time goes almost to the Q4_0 23 ms/token.
Before pushing the k-quants upstream I had a Q4_K variant that
had used 8-bit scales. It wasn't more accurate, used 0.125 bits more per weight,
was running slightly slower on the CPU (due to the larger model size
and being memory bound there), and the difference was entirely
negligible under CUDA. So, I decided to publish the version with 6-bit
scales. Perhaps I should re-consider and change to 8-bit scales?
* metal : some more optimizations
Q2_K: 25.4 ms/token
Q6_K: 27.3 ms/token
Q4_0: 22.8 ms/token
Q4_1: 23.1 ms/token
* metal : Q3_K support
Something is not quite right yet.
* metal : Q5_K support
Initial version achieves 31.2 ms/token, 210 GB/s
* metal : still not able to figure out why q3_K does not work
* Minor
* metal : yet another failed attempt to make q3_K work
* metal : optimize Q5_K
31.2 ms -> 27.8 ms.
250 GB/s.
* metal : q3_K still not working
Adding a heavily commented q3_K metal kernel to explain
my obviously faulty logic. Perhaps someone could spot the issue?
* metal : q3_K finally working
Not optimized at all.
What was the issue? The scales are not 4-bytes aligned,
and I was accessing them with a uint32_t pointer.
When I tried that on CUDA, I got an error (illegal memory access)
and added a memcpy to a local array of 3 uint32_t's.
But on Metal it told me there is no memcpy, so I tried
accessing directly. There is no error, just garbage results.
At some point I did try accessing the scales with an uint16_t
pointer (the scales are for sure 2-byte aligned), but was
still getting garbage. I guess, there must have been another bug.
No access to scales is via a uint16_t pointer and, after starting
from scratch from the C dequantize function, it finally works.
* metal : Q3_K 1st optimization pass
* metal : Q3_K second optimization pass - 29.6 ms/token
* metal : Q3_K cleanup
* metal : fixed accidentally broken Q2_K
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-12 21:39:21 +02:00
|
|
|
|
2023-12-13 13:04:25 +01:00
|
|
|
kernel_mul_mv_q4_K_f32_impl(src0, src1, dst, ne00, ne01, ne02, ne10, ne12, ne0, ne1, r2, r3, tgpig, tiisg, sgitg);
|
|
|
|
}
|
|
|
|
|
|
|
|
void kernel_mul_mv_q5_K_f32_impl(
|
|
|
|
device const void * src0,
|
|
|
|
device const float * src1,
|
|
|
|
device float * dst,
|
|
|
|
constant int64_t & ne00,
|
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant int64_t & ne10,
|
|
|
|
constant int64_t & ne12,
|
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
|
|
|
constant uint & r2,
|
|
|
|
constant uint & r3,
|
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint tiisg[[thread_index_in_simdgroup]],
|
|
|
|
uint sgitg[[simdgroup_index_in_threadgroup]]) {
|
|
|
|
|
Metal implementation for all k_quants (#1807)
* metal : improve q4_K
28.3 -> 26.0 ms/token by avoiding a branch in the
calculation of the scales.
* metal : small improvement for Q4_K
* metal : still optimizing Q4_K
This commit pushes it down to 25.3 ms / token.
The crazy idea of using 6 bits for the scales is really costly on
Metal: if I remove the bit fiddling necessary to make the block
scales, time goes almost to the Q4_0 23 ms/token.
Before pushing the k-quants upstream I had a Q4_K variant that
had used 8-bit scales. It wasn't more accurate, used 0.125 bits more per weight,
was running slightly slower on the CPU (due to the larger model size
and being memory bound there), and the difference was entirely
negligible under CUDA. So, I decided to publish the version with 6-bit
scales. Perhaps I should re-consider and change to 8-bit scales?
* metal : some more optimizations
Q2_K: 25.4 ms/token
Q6_K: 27.3 ms/token
Q4_0: 22.8 ms/token
Q4_1: 23.1 ms/token
* metal : Q3_K support
Something is not quite right yet.
* metal : Q5_K support
Initial version achieves 31.2 ms/token, 210 GB/s
* metal : still not able to figure out why q3_K does not work
* Minor
* metal : yet another failed attempt to make q3_K work
* metal : optimize Q5_K
31.2 ms -> 27.8 ms.
250 GB/s.
* metal : q3_K still not working
Adding a heavily commented q3_K metal kernel to explain
my obviously faulty logic. Perhaps someone could spot the issue?
* metal : q3_K finally working
Not optimized at all.
What was the issue? The scales are not 4-bytes aligned,
and I was accessing them with a uint32_t pointer.
When I tried that on CUDA, I got an error (illegal memory access)
and added a memcpy to a local array of 3 uint32_t's.
But on Metal it told me there is no memcpy, so I tried
accessing directly. There is no error, just garbage results.
At some point I did try accessing the scales with an uint16_t
pointer (the scales are for sure 2-byte aligned), but was
still getting garbage. I guess, there must have been another bug.
No access to scales is via a uint16_t pointer and, after starting
from scratch from the C dequantize function, it finally works.
* metal : Q3_K 1st optimization pass
* metal : Q3_K second optimization pass - 29.6 ms/token
* metal : Q3_K cleanup
* metal : fixed accidentally broken Q2_K
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-12 21:39:21 +02:00
|
|
|
const int nb = ne00/QK_K;
|
|
|
|
|
|
|
|
const int64_t r0 = tgpig.x;
|
|
|
|
const int64_t r1 = tgpig.y;
|
2023-12-07 21:26:54 +01:00
|
|
|
const int im = tgpig.z;
|
Metal implementation for all k_quants (#1807)
* metal : improve q4_K
28.3 -> 26.0 ms/token by avoiding a branch in the
calculation of the scales.
* metal : small improvement for Q4_K
* metal : still optimizing Q4_K
This commit pushes it down to 25.3 ms / token.
The crazy idea of using 6 bits for the scales is really costly on
Metal: if I remove the bit fiddling necessary to make the block
scales, time goes almost to the Q4_0 23 ms/token.
Before pushing the k-quants upstream I had a Q4_K variant that
had used 8-bit scales. It wasn't more accurate, used 0.125 bits more per weight,
was running slightly slower on the CPU (due to the larger model size
and being memory bound there), and the difference was entirely
negligible under CUDA. So, I decided to publish the version with 6-bit
scales. Perhaps I should re-consider and change to 8-bit scales?
* metal : some more optimizations
Q2_K: 25.4 ms/token
Q6_K: 27.3 ms/token
Q4_0: 22.8 ms/token
Q4_1: 23.1 ms/token
* metal : Q3_K support
Something is not quite right yet.
* metal : Q5_K support
Initial version achieves 31.2 ms/token, 210 GB/s
* metal : still not able to figure out why q3_K does not work
* Minor
* metal : yet another failed attempt to make q3_K work
* metal : optimize Q5_K
31.2 ms -> 27.8 ms.
250 GB/s.
* metal : q3_K still not working
Adding a heavily commented q3_K metal kernel to explain
my obviously faulty logic. Perhaps someone could spot the issue?
* metal : q3_K finally working
Not optimized at all.
What was the issue? The scales are not 4-bytes aligned,
and I was accessing them with a uint32_t pointer.
When I tried that on CUDA, I got an error (illegal memory access)
and added a memcpy to a local array of 3 uint32_t's.
But on Metal it told me there is no memcpy, so I tried
accessing directly. There is no error, just garbage results.
At some point I did try accessing the scales with an uint16_t
pointer (the scales are for sure 2-byte aligned), but was
still getting garbage. I guess, there must have been another bug.
No access to scales is via a uint16_t pointer and, after starting
from scratch from the C dequantize function, it finally works.
* metal : Q3_K 1st optimization pass
* metal : Q3_K second optimization pass - 29.6 ms/token
* metal : Q3_K cleanup
* metal : fixed accidentally broken Q2_K
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-12 21:39:21 +02:00
|
|
|
|
2023-07-20 17:19:45 +02:00
|
|
|
const int first_row = (r0 * N_SIMDGROUP + sgitg) * 2;
|
2023-12-07 21:26:54 +01:00
|
|
|
|
|
|
|
const uint i12 = im%ne12;
|
|
|
|
const uint i13 = im/ne12;
|
|
|
|
|
|
|
|
const uint offset0 = (i12/r2)*(nb*ne01) + (i13/r3)*(nb*ne01*ne02);
|
|
|
|
|
2023-08-16 22:07:04 +02:00
|
|
|
device const block_q5_K * x = (device const block_q5_K *) src0 + first_row*nb + offset0;
|
2023-12-07 21:26:54 +01:00
|
|
|
device const float * yy = (device const float *) src1 + r1*ne10 + im*ne00*ne1;
|
Metal implementation for all k_quants (#1807)
* metal : improve q4_K
28.3 -> 26.0 ms/token by avoiding a branch in the
calculation of the scales.
* metal : small improvement for Q4_K
* metal : still optimizing Q4_K
This commit pushes it down to 25.3 ms / token.
The crazy idea of using 6 bits for the scales is really costly on
Metal: if I remove the bit fiddling necessary to make the block
scales, time goes almost to the Q4_0 23 ms/token.
Before pushing the k-quants upstream I had a Q4_K variant that
had used 8-bit scales. It wasn't more accurate, used 0.125 bits more per weight,
was running slightly slower on the CPU (due to the larger model size
and being memory bound there), and the difference was entirely
negligible under CUDA. So, I decided to publish the version with 6-bit
scales. Perhaps I should re-consider and change to 8-bit scales?
* metal : some more optimizations
Q2_K: 25.4 ms/token
Q6_K: 27.3 ms/token
Q4_0: 22.8 ms/token
Q4_1: 23.1 ms/token
* metal : Q3_K support
Something is not quite right yet.
* metal : Q5_K support
Initial version achieves 31.2 ms/token, 210 GB/s
* metal : still not able to figure out why q3_K does not work
* Minor
* metal : yet another failed attempt to make q3_K work
* metal : optimize Q5_K
31.2 ms -> 27.8 ms.
250 GB/s.
* metal : q3_K still not working
Adding a heavily commented q3_K metal kernel to explain
my obviously faulty logic. Perhaps someone could spot the issue?
* metal : q3_K finally working
Not optimized at all.
What was the issue? The scales are not 4-bytes aligned,
and I was accessing them with a uint32_t pointer.
When I tried that on CUDA, I got an error (illegal memory access)
and added a memcpy to a local array of 3 uint32_t's.
But on Metal it told me there is no memcpy, so I tried
accessing directly. There is no error, just garbage results.
At some point I did try accessing the scales with an uint16_t
pointer (the scales are for sure 2-byte aligned), but was
still getting garbage. I guess, there must have been another bug.
No access to scales is via a uint16_t pointer and, after starting
from scratch from the C dequantize function, it finally works.
* metal : Q3_K 1st optimization pass
* metal : Q3_K second optimization pass - 29.6 ms/token
* metal : Q3_K cleanup
* metal : fixed accidentally broken Q2_K
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-12 21:39:21 +02:00
|
|
|
|
2023-07-20 17:19:45 +02:00
|
|
|
float sumf[2]={0.f};
|
Metal implementation for all k_quants (#1807)
* metal : improve q4_K
28.3 -> 26.0 ms/token by avoiding a branch in the
calculation of the scales.
* metal : small improvement for Q4_K
* metal : still optimizing Q4_K
This commit pushes it down to 25.3 ms / token.
The crazy idea of using 6 bits for the scales is really costly on
Metal: if I remove the bit fiddling necessary to make the block
scales, time goes almost to the Q4_0 23 ms/token.
Before pushing the k-quants upstream I had a Q4_K variant that
had used 8-bit scales. It wasn't more accurate, used 0.125 bits more per weight,
was running slightly slower on the CPU (due to the larger model size
and being memory bound there), and the difference was entirely
negligible under CUDA. So, I decided to publish the version with 6-bit
scales. Perhaps I should re-consider and change to 8-bit scales?
* metal : some more optimizations
Q2_K: 25.4 ms/token
Q6_K: 27.3 ms/token
Q4_0: 22.8 ms/token
Q4_1: 23.1 ms/token
* metal : Q3_K support
Something is not quite right yet.
* metal : Q5_K support
Initial version achieves 31.2 ms/token, 210 GB/s
* metal : still not able to figure out why q3_K does not work
* Minor
* metal : yet another failed attempt to make q3_K work
* metal : optimize Q5_K
31.2 ms -> 27.8 ms.
250 GB/s.
* metal : q3_K still not working
Adding a heavily commented q3_K metal kernel to explain
my obviously faulty logic. Perhaps someone could spot the issue?
* metal : q3_K finally working
Not optimized at all.
What was the issue? The scales are not 4-bytes aligned,
and I was accessing them with a uint32_t pointer.
When I tried that on CUDA, I got an error (illegal memory access)
and added a memcpy to a local array of 3 uint32_t's.
But on Metal it told me there is no memcpy, so I tried
accessing directly. There is no error, just garbage results.
At some point I did try accessing the scales with an uint16_t
pointer (the scales are for sure 2-byte aligned), but was
still getting garbage. I guess, there must have been another bug.
No access to scales is via a uint16_t pointer and, after starting
from scratch from the C dequantize function, it finally works.
* metal : Q3_K 1st optimization pass
* metal : Q3_K second optimization pass - 29.6 ms/token
* metal : Q3_K cleanup
* metal : fixed accidentally broken Q2_K
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-12 21:39:21 +02:00
|
|
|
|
2023-07-20 17:19:45 +02:00
|
|
|
const int step = sizeof(block_q5_K) * nb;
|
k-quants : support for super-block size of 64 (#2001)
* k_quants: WIP super-blocks with 64 weights
* k_quants: WIP super-blocks with 64 weights
Q6_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q4_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q2_K scalar and AVX2 works. Q2_K is way too slow (it is actually slower
than the scalar implementation)
* k_quants: WIP super-blocks with 64 weights
Q3_K scalar and AVX2 works.
* k_quants: WIP super-blocks with 64 weights
Q5_K scalar and AVX2 works, and with that all
k_quants are done on AVX2 and scalar
* k_quants: WIP super-blocks with 64 weights
Q6_K working on CUDA. Cannot make it run quite as gast as
with super-blocks with 256 weigths: 8% slower on 4080,
20% slower on the 1660 (but there we fit 1 less layer on the
GPU because pf the larger model size), so some fraction of
these 20% is due to that,
* k_quants: WIP super-blocks with 64 weights
Q4_K working on CUDA. ~10% slower on GTX-1660,
16% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q2_K working on CUDA. ~3% slower on GTX-1660,
10% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q3_K working on CUDA.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on CUDA, and with this CUDA is done.
* k_quants: WIP super-blocks with 64 weights
Q6_K working on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Q4_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q2_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q3_K working on ARM_NEON, but quite a bit slower than 256 weights.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on ARM_NEON, but quite a bit slower than 256 weights.
With that, we have full support for ARM_NEON, although
performance is not quite there.
* k_quants: WIP super-blocks with 64 weights
Slightly more efficient Q3_K and Q5_K
* k_quants: WIP super-blocks with 64 weights
Another small improvement for Q3_K and Q5_K on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Yet another speedup for Q5_K on ARM_NEON.
We are now within 10% of the QK_K = 256 version.
* k_quants: WIP super-blocks with 64 weights
* We are able to pass preprocessor macros to the Metal
compiler
* Q6_K works and is actually slightly more efficient than
the QK_K = 256 version (25.2 ms vs 25.8 ms)
* k_quants: WIP super-blocks with 64 weights
Q4_K works on Metal and is actually slightly faster
than QK_K = 256 (21.95 ms vs 24.0 ms).
* k_quants: WIP super-blocks with 64 weights
Q2_K works on Metal and is very slightly faster
than QK_K = 256 (23.8 ms vs 24.2 ms).
* k_quants: WIP super-blocks with 64 weights
Q3_K works on Metal and is slightly faster
than QK_K = 256 (26.6 ms vs 28.3 ms).
* k_quants: WIP super-blocks with 64 weights
Q5_K works on Metal and is slightly faster
than QK_K = 256 (23.7 ms vs 26.3 ms).
* k_quants: call them _K, not _k, also on Metal
* k_quants: correctly define QK_K in llama.cpp
* Fixed bug in q4_K quantization added with the 64-block addition
* Simplify via lambda
* k_quants: swicth Q3_K to 4-bit scales when QK_K = 64
Otherwise there isn't much benefit from this
quantization type. There is some very slight loss
in accuracy, but we reduce size by ~7%.
E.g., for OpenLLaMA-3B, Q3_K_S perplexity is
8.6131 with 8-bit scales and 8.6352 with 4-bit,
while file size decreases from 1.53G to 1.44G.
* k_quants: switch Q4_K to 4-bit scales when QK_K = 64
Here the loss in accuracy is greater than for Q3_K,
but the Q4_K points still move further to the left on
the perplexity vs size curve.
* k_quants: forgot to add the Metal changes in last commit
* k_quants: change Q5_K to be type 0 when QK_K = 64
Still needs AVX2 implementation
* k_quants: AVX2 implementation for new 64-weight Q5_K
* k_quants: 10% faster ARM_NEON Q5_K dot product
* k_quants: fixed issue caused by merging with master
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-26 18:43:07 +02:00
|
|
|
|
|
|
|
#if QK_K == 256
|
2023-07-20 17:19:45 +02:00
|
|
|
#
|
|
|
|
float yl[16], yh[16];
|
k-quants : support for super-block size of 64 (#2001)
* k_quants: WIP super-blocks with 64 weights
* k_quants: WIP super-blocks with 64 weights
Q6_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q4_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q2_K scalar and AVX2 works. Q2_K is way too slow (it is actually slower
than the scalar implementation)
* k_quants: WIP super-blocks with 64 weights
Q3_K scalar and AVX2 works.
* k_quants: WIP super-blocks with 64 weights
Q5_K scalar and AVX2 works, and with that all
k_quants are done on AVX2 and scalar
* k_quants: WIP super-blocks with 64 weights
Q6_K working on CUDA. Cannot make it run quite as gast as
with super-blocks with 256 weigths: 8% slower on 4080,
20% slower on the 1660 (but there we fit 1 less layer on the
GPU because pf the larger model size), so some fraction of
these 20% is due to that,
* k_quants: WIP super-blocks with 64 weights
Q4_K working on CUDA. ~10% slower on GTX-1660,
16% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q2_K working on CUDA. ~3% slower on GTX-1660,
10% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q3_K working on CUDA.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on CUDA, and with this CUDA is done.
* k_quants: WIP super-blocks with 64 weights
Q6_K working on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Q4_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q2_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q3_K working on ARM_NEON, but quite a bit slower than 256 weights.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on ARM_NEON, but quite a bit slower than 256 weights.
With that, we have full support for ARM_NEON, although
performance is not quite there.
* k_quants: WIP super-blocks with 64 weights
Slightly more efficient Q3_K and Q5_K
* k_quants: WIP super-blocks with 64 weights
Another small improvement for Q3_K and Q5_K on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Yet another speedup for Q5_K on ARM_NEON.
We are now within 10% of the QK_K = 256 version.
* k_quants: WIP super-blocks with 64 weights
* We are able to pass preprocessor macros to the Metal
compiler
* Q6_K works and is actually slightly more efficient than
the QK_K = 256 version (25.2 ms vs 25.8 ms)
* k_quants: WIP super-blocks with 64 weights
Q4_K works on Metal and is actually slightly faster
than QK_K = 256 (21.95 ms vs 24.0 ms).
* k_quants: WIP super-blocks with 64 weights
Q2_K works on Metal and is very slightly faster
than QK_K = 256 (23.8 ms vs 24.2 ms).
* k_quants: WIP super-blocks with 64 weights
Q3_K works on Metal and is slightly faster
than QK_K = 256 (26.6 ms vs 28.3 ms).
* k_quants: WIP super-blocks with 64 weights
Q5_K works on Metal and is slightly faster
than QK_K = 256 (23.7 ms vs 26.3 ms).
* k_quants: call them _K, not _k, also on Metal
* k_quants: correctly define QK_K in llama.cpp
* Fixed bug in q4_K quantization added with the 64-block addition
* Simplify via lambda
* k_quants: swicth Q3_K to 4-bit scales when QK_K = 64
Otherwise there isn't much benefit from this
quantization type. There is some very slight loss
in accuracy, but we reduce size by ~7%.
E.g., for OpenLLaMA-3B, Q3_K_S perplexity is
8.6131 with 8-bit scales and 8.6352 with 4-bit,
while file size decreases from 1.53G to 1.44G.
* k_quants: switch Q4_K to 4-bit scales when QK_K = 64
Here the loss in accuracy is greater than for Q3_K,
but the Q4_K points still move further to the left on
the perplexity vs size curve.
* k_quants: forgot to add the Metal changes in last commit
* k_quants: change Q5_K to be type 0 when QK_K = 64
Still needs AVX2 implementation
* k_quants: AVX2 implementation for new 64-weight Q5_K
* k_quants: 10% faster ARM_NEON Q5_K dot product
* k_quants: fixed issue caused by merging with master
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-26 18:43:07 +02:00
|
|
|
|
|
|
|
const uint16_t kmask1 = 0x3f3f;
|
|
|
|
const uint16_t kmask2 = 0x0f0f;
|
|
|
|
const uint16_t kmask3 = 0xc0c0;
|
|
|
|
|
2023-07-20 17:19:45 +02:00
|
|
|
const int tid = tiisg/4;
|
|
|
|
const int ix = tiisg%4;
|
2023-12-07 21:26:54 +01:00
|
|
|
const int iq = tid/4;
|
2023-07-20 17:19:45 +02:00
|
|
|
const int ir = tid%4;
|
|
|
|
const int n = 8;
|
Metal implementation for all k_quants (#1807)
* metal : improve q4_K
28.3 -> 26.0 ms/token by avoiding a branch in the
calculation of the scales.
* metal : small improvement for Q4_K
* metal : still optimizing Q4_K
This commit pushes it down to 25.3 ms / token.
The crazy idea of using 6 bits for the scales is really costly on
Metal: if I remove the bit fiddling necessary to make the block
scales, time goes almost to the Q4_0 23 ms/token.
Before pushing the k-quants upstream I had a Q4_K variant that
had used 8-bit scales. It wasn't more accurate, used 0.125 bits more per weight,
was running slightly slower on the CPU (due to the larger model size
and being memory bound there), and the difference was entirely
negligible under CUDA. So, I decided to publish the version with 6-bit
scales. Perhaps I should re-consider and change to 8-bit scales?
* metal : some more optimizations
Q2_K: 25.4 ms/token
Q6_K: 27.3 ms/token
Q4_0: 22.8 ms/token
Q4_1: 23.1 ms/token
* metal : Q3_K support
Something is not quite right yet.
* metal : Q5_K support
Initial version achieves 31.2 ms/token, 210 GB/s
* metal : still not able to figure out why q3_K does not work
* Minor
* metal : yet another failed attempt to make q3_K work
* metal : optimize Q5_K
31.2 ms -> 27.8 ms.
250 GB/s.
* metal : q3_K still not working
Adding a heavily commented q3_K metal kernel to explain
my obviously faulty logic. Perhaps someone could spot the issue?
* metal : q3_K finally working
Not optimized at all.
What was the issue? The scales are not 4-bytes aligned,
and I was accessing them with a uint32_t pointer.
When I tried that on CUDA, I got an error (illegal memory access)
and added a memcpy to a local array of 3 uint32_t's.
But on Metal it told me there is no memcpy, so I tried
accessing directly. There is no error, just garbage results.
At some point I did try accessing the scales with an uint16_t
pointer (the scales are for sure 2-byte aligned), but was
still getting garbage. I guess, there must have been another bug.
No access to scales is via a uint16_t pointer and, after starting
from scratch from the C dequantize function, it finally works.
* metal : Q3_K 1st optimization pass
* metal : Q3_K second optimization pass - 29.6 ms/token
* metal : Q3_K cleanup
* metal : fixed accidentally broken Q2_K
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-12 21:39:21 +02:00
|
|
|
|
2023-07-20 17:19:45 +02:00
|
|
|
const int l0 = n*ir;
|
2023-12-07 21:26:54 +01:00
|
|
|
const int q_offset = 32*iq + l0;
|
|
|
|
const int y_offset = 64*iq + l0;
|
Metal implementation for all k_quants (#1807)
* metal : improve q4_K
28.3 -> 26.0 ms/token by avoiding a branch in the
calculation of the scales.
* metal : small improvement for Q4_K
* metal : still optimizing Q4_K
This commit pushes it down to 25.3 ms / token.
The crazy idea of using 6 bits for the scales is really costly on
Metal: if I remove the bit fiddling necessary to make the block
scales, time goes almost to the Q4_0 23 ms/token.
Before pushing the k-quants upstream I had a Q4_K variant that
had used 8-bit scales. It wasn't more accurate, used 0.125 bits more per weight,
was running slightly slower on the CPU (due to the larger model size
and being memory bound there), and the difference was entirely
negligible under CUDA. So, I decided to publish the version with 6-bit
scales. Perhaps I should re-consider and change to 8-bit scales?
* metal : some more optimizations
Q2_K: 25.4 ms/token
Q6_K: 27.3 ms/token
Q4_0: 22.8 ms/token
Q4_1: 23.1 ms/token
* metal : Q3_K support
Something is not quite right yet.
* metal : Q5_K support
Initial version achieves 31.2 ms/token, 210 GB/s
* metal : still not able to figure out why q3_K does not work
* Minor
* metal : yet another failed attempt to make q3_K work
* metal : optimize Q5_K
31.2 ms -> 27.8 ms.
250 GB/s.
* metal : q3_K still not working
Adding a heavily commented q3_K metal kernel to explain
my obviously faulty logic. Perhaps someone could spot the issue?
* metal : q3_K finally working
Not optimized at all.
What was the issue? The scales are not 4-bytes aligned,
and I was accessing them with a uint32_t pointer.
When I tried that on CUDA, I got an error (illegal memory access)
and added a memcpy to a local array of 3 uint32_t's.
But on Metal it told me there is no memcpy, so I tried
accessing directly. There is no error, just garbage results.
At some point I did try accessing the scales with an uint16_t
pointer (the scales are for sure 2-byte aligned), but was
still getting garbage. I guess, there must have been another bug.
No access to scales is via a uint16_t pointer and, after starting
from scratch from the C dequantize function, it finally works.
* metal : Q3_K 1st optimization pass
* metal : Q3_K second optimization pass - 29.6 ms/token
* metal : Q3_K cleanup
* metal : fixed accidentally broken Q2_K
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-12 21:39:21 +02:00
|
|
|
|
2023-12-07 21:26:54 +01:00
|
|
|
const uint8_t hm1 = 1u << (2*iq);
|
Metal implementation for all k_quants (#1807)
* metal : improve q4_K
28.3 -> 26.0 ms/token by avoiding a branch in the
calculation of the scales.
* metal : small improvement for Q4_K
* metal : still optimizing Q4_K
This commit pushes it down to 25.3 ms / token.
The crazy idea of using 6 bits for the scales is really costly on
Metal: if I remove the bit fiddling necessary to make the block
scales, time goes almost to the Q4_0 23 ms/token.
Before pushing the k-quants upstream I had a Q4_K variant that
had used 8-bit scales. It wasn't more accurate, used 0.125 bits more per weight,
was running slightly slower on the CPU (due to the larger model size
and being memory bound there), and the difference was entirely
negligible under CUDA. So, I decided to publish the version with 6-bit
scales. Perhaps I should re-consider and change to 8-bit scales?
* metal : some more optimizations
Q2_K: 25.4 ms/token
Q6_K: 27.3 ms/token
Q4_0: 22.8 ms/token
Q4_1: 23.1 ms/token
* metal : Q3_K support
Something is not quite right yet.
* metal : Q5_K support
Initial version achieves 31.2 ms/token, 210 GB/s
* metal : still not able to figure out why q3_K does not work
* Minor
* metal : yet another failed attempt to make q3_K work
* metal : optimize Q5_K
31.2 ms -> 27.8 ms.
250 GB/s.
* metal : q3_K still not working
Adding a heavily commented q3_K metal kernel to explain
my obviously faulty logic. Perhaps someone could spot the issue?
* metal : q3_K finally working
Not optimized at all.
What was the issue? The scales are not 4-bytes aligned,
and I was accessing them with a uint32_t pointer.
When I tried that on CUDA, I got an error (illegal memory access)
and added a memcpy to a local array of 3 uint32_t's.
But on Metal it told me there is no memcpy, so I tried
accessing directly. There is no error, just garbage results.
At some point I did try accessing the scales with an uint16_t
pointer (the scales are for sure 2-byte aligned), but was
still getting garbage. I guess, there must have been another bug.
No access to scales is via a uint16_t pointer and, after starting
from scratch from the C dequantize function, it finally works.
* metal : Q3_K 1st optimization pass
* metal : Q3_K second optimization pass - 29.6 ms/token
* metal : Q3_K cleanup
* metal : fixed accidentally broken Q2_K
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-12 21:39:21 +02:00
|
|
|
const uint8_t hm2 = hm1 << 1;
|
|
|
|
const uint8_t hm3 = hm1 << 4;
|
|
|
|
const uint8_t hm4 = hm2 << 4;
|
|
|
|
|
2023-07-20 17:19:45 +02:00
|
|
|
uint16_t sc16[4];
|
|
|
|
thread const uint8_t * sc8 = (thread const uint8_t *)sc16;
|
Metal implementation for all k_quants (#1807)
* metal : improve q4_K
28.3 -> 26.0 ms/token by avoiding a branch in the
calculation of the scales.
* metal : small improvement for Q4_K
* metal : still optimizing Q4_K
This commit pushes it down to 25.3 ms / token.
The crazy idea of using 6 bits for the scales is really costly on
Metal: if I remove the bit fiddling necessary to make the block
scales, time goes almost to the Q4_0 23 ms/token.
Before pushing the k-quants upstream I had a Q4_K variant that
had used 8-bit scales. It wasn't more accurate, used 0.125 bits more per weight,
was running slightly slower on the CPU (due to the larger model size
and being memory bound there), and the difference was entirely
negligible under CUDA. So, I decided to publish the version with 6-bit
scales. Perhaps I should re-consider and change to 8-bit scales?
* metal : some more optimizations
Q2_K: 25.4 ms/token
Q6_K: 27.3 ms/token
Q4_0: 22.8 ms/token
Q4_1: 23.1 ms/token
* metal : Q3_K support
Something is not quite right yet.
* metal : Q5_K support
Initial version achieves 31.2 ms/token, 210 GB/s
* metal : still not able to figure out why q3_K does not work
* Minor
* metal : yet another failed attempt to make q3_K work
* metal : optimize Q5_K
31.2 ms -> 27.8 ms.
250 GB/s.
* metal : q3_K still not working
Adding a heavily commented q3_K metal kernel to explain
my obviously faulty logic. Perhaps someone could spot the issue?
* metal : q3_K finally working
Not optimized at all.
What was the issue? The scales are not 4-bytes aligned,
and I was accessing them with a uint32_t pointer.
When I tried that on CUDA, I got an error (illegal memory access)
and added a memcpy to a local array of 3 uint32_t's.
But on Metal it told me there is no memcpy, so I tried
accessing directly. There is no error, just garbage results.
At some point I did try accessing the scales with an uint16_t
pointer (the scales are for sure 2-byte aligned), but was
still getting garbage. I guess, there must have been another bug.
No access to scales is via a uint16_t pointer and, after starting
from scratch from the C dequantize function, it finally works.
* metal : Q3_K 1st optimization pass
* metal : Q3_K second optimization pass - 29.6 ms/token
* metal : Q3_K cleanup
* metal : fixed accidentally broken Q2_K
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-12 21:39:21 +02:00
|
|
|
|
2023-07-20 17:19:45 +02:00
|
|
|
device const float * y1 = yy + ix*QK_K + y_offset;
|
Metal implementation for all k_quants (#1807)
* metal : improve q4_K
28.3 -> 26.0 ms/token by avoiding a branch in the
calculation of the scales.
* metal : small improvement for Q4_K
* metal : still optimizing Q4_K
This commit pushes it down to 25.3 ms / token.
The crazy idea of using 6 bits for the scales is really costly on
Metal: if I remove the bit fiddling necessary to make the block
scales, time goes almost to the Q4_0 23 ms/token.
Before pushing the k-quants upstream I had a Q4_K variant that
had used 8-bit scales. It wasn't more accurate, used 0.125 bits more per weight,
was running slightly slower on the CPU (due to the larger model size
and being memory bound there), and the difference was entirely
negligible under CUDA. So, I decided to publish the version with 6-bit
scales. Perhaps I should re-consider and change to 8-bit scales?
* metal : some more optimizations
Q2_K: 25.4 ms/token
Q6_K: 27.3 ms/token
Q4_0: 22.8 ms/token
Q4_1: 23.1 ms/token
* metal : Q3_K support
Something is not quite right yet.
* metal : Q5_K support
Initial version achieves 31.2 ms/token, 210 GB/s
* metal : still not able to figure out why q3_K does not work
* Minor
* metal : yet another failed attempt to make q3_K work
* metal : optimize Q5_K
31.2 ms -> 27.8 ms.
250 GB/s.
* metal : q3_K still not working
Adding a heavily commented q3_K metal kernel to explain
my obviously faulty logic. Perhaps someone could spot the issue?
* metal : q3_K finally working
Not optimized at all.
What was the issue? The scales are not 4-bytes aligned,
and I was accessing them with a uint32_t pointer.
When I tried that on CUDA, I got an error (illegal memory access)
and added a memcpy to a local array of 3 uint32_t's.
But on Metal it told me there is no memcpy, so I tried
accessing directly. There is no error, just garbage results.
At some point I did try accessing the scales with an uint16_t
pointer (the scales are for sure 2-byte aligned), but was
still getting garbage. I guess, there must have been another bug.
No access to scales is via a uint16_t pointer and, after starting
from scratch from the C dequantize function, it finally works.
* metal : Q3_K 1st optimization pass
* metal : Q3_K second optimization pass - 29.6 ms/token
* metal : Q3_K cleanup
* metal : fixed accidentally broken Q2_K
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-12 21:39:21 +02:00
|
|
|
|
2023-07-20 17:19:45 +02:00
|
|
|
for (int i = ix; i < nb; i += 4) {
|
Metal implementation for all k_quants (#1807)
* metal : improve q4_K
28.3 -> 26.0 ms/token by avoiding a branch in the
calculation of the scales.
* metal : small improvement for Q4_K
* metal : still optimizing Q4_K
This commit pushes it down to 25.3 ms / token.
The crazy idea of using 6 bits for the scales is really costly on
Metal: if I remove the bit fiddling necessary to make the block
scales, time goes almost to the Q4_0 23 ms/token.
Before pushing the k-quants upstream I had a Q4_K variant that
had used 8-bit scales. It wasn't more accurate, used 0.125 bits more per weight,
was running slightly slower on the CPU (due to the larger model size
and being memory bound there), and the difference was entirely
negligible under CUDA. So, I decided to publish the version with 6-bit
scales. Perhaps I should re-consider and change to 8-bit scales?
* metal : some more optimizations
Q2_K: 25.4 ms/token
Q6_K: 27.3 ms/token
Q4_0: 22.8 ms/token
Q4_1: 23.1 ms/token
* metal : Q3_K support
Something is not quite right yet.
* metal : Q5_K support
Initial version achieves 31.2 ms/token, 210 GB/s
* metal : still not able to figure out why q3_K does not work
* Minor
* metal : yet another failed attempt to make q3_K work
* metal : optimize Q5_K
31.2 ms -> 27.8 ms.
250 GB/s.
* metal : q3_K still not working
Adding a heavily commented q3_K metal kernel to explain
my obviously faulty logic. Perhaps someone could spot the issue?
* metal : q3_K finally working
Not optimized at all.
What was the issue? The scales are not 4-bytes aligned,
and I was accessing them with a uint32_t pointer.
When I tried that on CUDA, I got an error (illegal memory access)
and added a memcpy to a local array of 3 uint32_t's.
But on Metal it told me there is no memcpy, so I tried
accessing directly. There is no error, just garbage results.
At some point I did try accessing the scales with an uint16_t
pointer (the scales are for sure 2-byte aligned), but was
still getting garbage. I guess, there must have been another bug.
No access to scales is via a uint16_t pointer and, after starting
from scratch from the C dequantize function, it finally works.
* metal : Q3_K 1st optimization pass
* metal : Q3_K second optimization pass - 29.6 ms/token
* metal : Q3_K cleanup
* metal : fixed accidentally broken Q2_K
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-12 21:39:21 +02:00
|
|
|
|
2023-07-20 17:19:45 +02:00
|
|
|
device const uint8_t * q1 = x[i].qs + q_offset;
|
|
|
|
device const uint8_t * qh = x[i].qh + l0;
|
|
|
|
device const half * dh = &x[i].d;
|
2023-12-07 21:26:54 +01:00
|
|
|
device const uint16_t * a = (device const uint16_t *)x[i].scales + iq;
|
Metal implementation for all k_quants (#1807)
* metal : improve q4_K
28.3 -> 26.0 ms/token by avoiding a branch in the
calculation of the scales.
* metal : small improvement for Q4_K
* metal : still optimizing Q4_K
This commit pushes it down to 25.3 ms / token.
The crazy idea of using 6 bits for the scales is really costly on
Metal: if I remove the bit fiddling necessary to make the block
scales, time goes almost to the Q4_0 23 ms/token.
Before pushing the k-quants upstream I had a Q4_K variant that
had used 8-bit scales. It wasn't more accurate, used 0.125 bits more per weight,
was running slightly slower on the CPU (due to the larger model size
and being memory bound there), and the difference was entirely
negligible under CUDA. So, I decided to publish the version with 6-bit
scales. Perhaps I should re-consider and change to 8-bit scales?
* metal : some more optimizations
Q2_K: 25.4 ms/token
Q6_K: 27.3 ms/token
Q4_0: 22.8 ms/token
Q4_1: 23.1 ms/token
* metal : Q3_K support
Something is not quite right yet.
* metal : Q5_K support
Initial version achieves 31.2 ms/token, 210 GB/s
* metal : still not able to figure out why q3_K does not work
* Minor
* metal : yet another failed attempt to make q3_K work
* metal : optimize Q5_K
31.2 ms -> 27.8 ms.
250 GB/s.
* metal : q3_K still not working
Adding a heavily commented q3_K metal kernel to explain
my obviously faulty logic. Perhaps someone could spot the issue?
* metal : q3_K finally working
Not optimized at all.
What was the issue? The scales are not 4-bytes aligned,
and I was accessing them with a uint32_t pointer.
When I tried that on CUDA, I got an error (illegal memory access)
and added a memcpy to a local array of 3 uint32_t's.
But on Metal it told me there is no memcpy, so I tried
accessing directly. There is no error, just garbage results.
At some point I did try accessing the scales with an uint16_t
pointer (the scales are for sure 2-byte aligned), but was
still getting garbage. I guess, there must have been another bug.
No access to scales is via a uint16_t pointer and, after starting
from scratch from the C dequantize function, it finally works.
* metal : Q3_K 1st optimization pass
* metal : Q3_K second optimization pass - 29.6 ms/token
* metal : Q3_K cleanup
* metal : fixed accidentally broken Q2_K
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-12 21:39:21 +02:00
|
|
|
|
2023-07-20 17:19:45 +02:00
|
|
|
device const float * y2 = y1 + 128;
|
|
|
|
float4 sumy = {0.f, 0.f, 0.f, 0.f};
|
|
|
|
for (int l = 0; l < 8; ++l) {
|
|
|
|
yl[l+0] = y1[l+ 0]; sumy[0] += yl[l+0];
|
|
|
|
yl[l+8] = y1[l+32]; sumy[1] += yl[l+8];
|
|
|
|
yh[l+0] = y2[l+ 0]; sumy[2] += yh[l+0];
|
|
|
|
yh[l+8] = y2[l+32]; sumy[3] += yh[l+8];
|
|
|
|
}
|
Metal implementation for all k_quants (#1807)
* metal : improve q4_K
28.3 -> 26.0 ms/token by avoiding a branch in the
calculation of the scales.
* metal : small improvement for Q4_K
* metal : still optimizing Q4_K
This commit pushes it down to 25.3 ms / token.
The crazy idea of using 6 bits for the scales is really costly on
Metal: if I remove the bit fiddling necessary to make the block
scales, time goes almost to the Q4_0 23 ms/token.
Before pushing the k-quants upstream I had a Q4_K variant that
had used 8-bit scales. It wasn't more accurate, used 0.125 bits more per weight,
was running slightly slower on the CPU (due to the larger model size
and being memory bound there), and the difference was entirely
negligible under CUDA. So, I decided to publish the version with 6-bit
scales. Perhaps I should re-consider and change to 8-bit scales?
* metal : some more optimizations
Q2_K: 25.4 ms/token
Q6_K: 27.3 ms/token
Q4_0: 22.8 ms/token
Q4_1: 23.1 ms/token
* metal : Q3_K support
Something is not quite right yet.
* metal : Q5_K support
Initial version achieves 31.2 ms/token, 210 GB/s
* metal : still not able to figure out why q3_K does not work
* Minor
* metal : yet another failed attempt to make q3_K work
* metal : optimize Q5_K
31.2 ms -> 27.8 ms.
250 GB/s.
* metal : q3_K still not working
Adding a heavily commented q3_K metal kernel to explain
my obviously faulty logic. Perhaps someone could spot the issue?
* metal : q3_K finally working
Not optimized at all.
What was the issue? The scales are not 4-bytes aligned,
and I was accessing them with a uint32_t pointer.
When I tried that on CUDA, I got an error (illegal memory access)
and added a memcpy to a local array of 3 uint32_t's.
But on Metal it told me there is no memcpy, so I tried
accessing directly. There is no error, just garbage results.
At some point I did try accessing the scales with an uint16_t
pointer (the scales are for sure 2-byte aligned), but was
still getting garbage. I guess, there must have been another bug.
No access to scales is via a uint16_t pointer and, after starting
from scratch from the C dequantize function, it finally works.
* metal : Q3_K 1st optimization pass
* metal : Q3_K second optimization pass - 29.6 ms/token
* metal : Q3_K cleanup
* metal : fixed accidentally broken Q2_K
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-12 21:39:21 +02:00
|
|
|
|
2023-07-20 17:19:45 +02:00
|
|
|
for (int row = 0; row < 2; ++row) {
|
|
|
|
|
|
|
|
device const uint8_t * q2 = q1 + 64;
|
|
|
|
|
|
|
|
sc16[0] = a[0] & kmask1;
|
|
|
|
sc16[1] = a[2] & kmask1;
|
|
|
|
sc16[2] = ((a[4] >> 0) & kmask2) | ((a[0] & kmask3) >> 2);
|
|
|
|
sc16[3] = ((a[4] >> 4) & kmask2) | ((a[2] & kmask3) >> 2);
|
|
|
|
|
2023-09-08 18:01:04 +02:00
|
|
|
float4 acc1 = {0.f};
|
|
|
|
float4 acc2 = {0.f};
|
2023-07-20 17:19:45 +02:00
|
|
|
for (int l = 0; l < n; ++l) {
|
|
|
|
uint8_t h = qh[l];
|
2023-09-08 18:01:04 +02:00
|
|
|
acc1[0] += yl[l+0] * (q1[l] & 0x0F);
|
|
|
|
acc1[1] += yl[l+8] * (q1[l] & 0xF0);
|
|
|
|
acc1[2] += yh[l+0] * (q2[l] & 0x0F);
|
|
|
|
acc1[3] += yh[l+8] * (q2[l] & 0xF0);
|
|
|
|
acc2[0] += h & hm1 ? yl[l+0] : 0.f;
|
|
|
|
acc2[1] += h & hm2 ? yl[l+8] : 0.f;
|
|
|
|
acc2[2] += h & hm3 ? yh[l+0] : 0.f;
|
|
|
|
acc2[3] += h & hm4 ? yh[l+8] : 0.f;
|
2023-07-20 17:19:45 +02:00
|
|
|
}
|
|
|
|
const float dall = dh[0];
|
|
|
|
const float dmin = dh[1];
|
2023-09-08 18:01:04 +02:00
|
|
|
sumf[row] += dall * (sc8[0] * (acc1[0] + 16.f*acc2[0]) +
|
|
|
|
sc8[1] * (acc1[1]/16.f + 16.f*acc2[1]) +
|
|
|
|
sc8[4] * (acc1[2] + 16.f*acc2[2]) +
|
|
|
|
sc8[5] * (acc1[3]/16.f + 16.f*acc2[3])) -
|
2023-07-20 17:19:45 +02:00
|
|
|
dmin * (sumy[0] * sc8[2] + sumy[1] * sc8[3] + sumy[2] * sc8[6] + sumy[3] * sc8[7]);
|
Metal implementation for all k_quants (#1807)
* metal : improve q4_K
28.3 -> 26.0 ms/token by avoiding a branch in the
calculation of the scales.
* metal : small improvement for Q4_K
* metal : still optimizing Q4_K
This commit pushes it down to 25.3 ms / token.
The crazy idea of using 6 bits for the scales is really costly on
Metal: if I remove the bit fiddling necessary to make the block
scales, time goes almost to the Q4_0 23 ms/token.
Before pushing the k-quants upstream I had a Q4_K variant that
had used 8-bit scales. It wasn't more accurate, used 0.125 bits more per weight,
was running slightly slower on the CPU (due to the larger model size
and being memory bound there), and the difference was entirely
negligible under CUDA. So, I decided to publish the version with 6-bit
scales. Perhaps I should re-consider and change to 8-bit scales?
* metal : some more optimizations
Q2_K: 25.4 ms/token
Q6_K: 27.3 ms/token
Q4_0: 22.8 ms/token
Q4_1: 23.1 ms/token
* metal : Q3_K support
Something is not quite right yet.
* metal : Q5_K support
Initial version achieves 31.2 ms/token, 210 GB/s
* metal : still not able to figure out why q3_K does not work
* Minor
* metal : yet another failed attempt to make q3_K work
* metal : optimize Q5_K
31.2 ms -> 27.8 ms.
250 GB/s.
* metal : q3_K still not working
Adding a heavily commented q3_K metal kernel to explain
my obviously faulty logic. Perhaps someone could spot the issue?
* metal : q3_K finally working
Not optimized at all.
What was the issue? The scales are not 4-bytes aligned,
and I was accessing them with a uint32_t pointer.
When I tried that on CUDA, I got an error (illegal memory access)
and added a memcpy to a local array of 3 uint32_t's.
But on Metal it told me there is no memcpy, so I tried
accessing directly. There is no error, just garbage results.
At some point I did try accessing the scales with an uint16_t
pointer (the scales are for sure 2-byte aligned), but was
still getting garbage. I guess, there must have been another bug.
No access to scales is via a uint16_t pointer and, after starting
from scratch from the C dequantize function, it finally works.
* metal : Q3_K 1st optimization pass
* metal : Q3_K second optimization pass - 29.6 ms/token
* metal : Q3_K cleanup
* metal : fixed accidentally broken Q2_K
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-12 21:39:21 +02:00
|
|
|
|
2023-07-20 17:19:45 +02:00
|
|
|
q1 += step;
|
|
|
|
qh += step;
|
|
|
|
dh += step/2;
|
|
|
|
a += step/2;
|
Metal implementation for all k_quants (#1807)
* metal : improve q4_K
28.3 -> 26.0 ms/token by avoiding a branch in the
calculation of the scales.
* metal : small improvement for Q4_K
* metal : still optimizing Q4_K
This commit pushes it down to 25.3 ms / token.
The crazy idea of using 6 bits for the scales is really costly on
Metal: if I remove the bit fiddling necessary to make the block
scales, time goes almost to the Q4_0 23 ms/token.
Before pushing the k-quants upstream I had a Q4_K variant that
had used 8-bit scales. It wasn't more accurate, used 0.125 bits more per weight,
was running slightly slower on the CPU (due to the larger model size
and being memory bound there), and the difference was entirely
negligible under CUDA. So, I decided to publish the version with 6-bit
scales. Perhaps I should re-consider and change to 8-bit scales?
* metal : some more optimizations
Q2_K: 25.4 ms/token
Q6_K: 27.3 ms/token
Q4_0: 22.8 ms/token
Q4_1: 23.1 ms/token
* metal : Q3_K support
Something is not quite right yet.
* metal : Q5_K support
Initial version achieves 31.2 ms/token, 210 GB/s
* metal : still not able to figure out why q3_K does not work
* Minor
* metal : yet another failed attempt to make q3_K work
* metal : optimize Q5_K
31.2 ms -> 27.8 ms.
250 GB/s.
* metal : q3_K still not working
Adding a heavily commented q3_K metal kernel to explain
my obviously faulty logic. Perhaps someone could spot the issue?
* metal : q3_K finally working
Not optimized at all.
What was the issue? The scales are not 4-bytes aligned,
and I was accessing them with a uint32_t pointer.
When I tried that on CUDA, I got an error (illegal memory access)
and added a memcpy to a local array of 3 uint32_t's.
But on Metal it told me there is no memcpy, so I tried
accessing directly. There is no error, just garbage results.
At some point I did try accessing the scales with an uint16_t
pointer (the scales are for sure 2-byte aligned), but was
still getting garbage. I guess, there must have been another bug.
No access to scales is via a uint16_t pointer and, after starting
from scratch from the C dequantize function, it finally works.
* metal : Q3_K 1st optimization pass
* metal : Q3_K second optimization pass - 29.6 ms/token
* metal : Q3_K cleanup
* metal : fixed accidentally broken Q2_K
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-12 21:39:21 +02:00
|
|
|
|
|
|
|
}
|
2023-07-20 17:19:45 +02:00
|
|
|
|
|
|
|
y1 += 4 * QK_K;
|
Metal implementation for all k_quants (#1807)
* metal : improve q4_K
28.3 -> 26.0 ms/token by avoiding a branch in the
calculation of the scales.
* metal : small improvement for Q4_K
* metal : still optimizing Q4_K
This commit pushes it down to 25.3 ms / token.
The crazy idea of using 6 bits for the scales is really costly on
Metal: if I remove the bit fiddling necessary to make the block
scales, time goes almost to the Q4_0 23 ms/token.
Before pushing the k-quants upstream I had a Q4_K variant that
had used 8-bit scales. It wasn't more accurate, used 0.125 bits more per weight,
was running slightly slower on the CPU (due to the larger model size
and being memory bound there), and the difference was entirely
negligible under CUDA. So, I decided to publish the version with 6-bit
scales. Perhaps I should re-consider and change to 8-bit scales?
* metal : some more optimizations
Q2_K: 25.4 ms/token
Q6_K: 27.3 ms/token
Q4_0: 22.8 ms/token
Q4_1: 23.1 ms/token
* metal : Q3_K support
Something is not quite right yet.
* metal : Q5_K support
Initial version achieves 31.2 ms/token, 210 GB/s
* metal : still not able to figure out why q3_K does not work
* Minor
* metal : yet another failed attempt to make q3_K work
* metal : optimize Q5_K
31.2 ms -> 27.8 ms.
250 GB/s.
* metal : q3_K still not working
Adding a heavily commented q3_K metal kernel to explain
my obviously faulty logic. Perhaps someone could spot the issue?
* metal : q3_K finally working
Not optimized at all.
What was the issue? The scales are not 4-bytes aligned,
and I was accessing them with a uint32_t pointer.
When I tried that on CUDA, I got an error (illegal memory access)
and added a memcpy to a local array of 3 uint32_t's.
But on Metal it told me there is no memcpy, so I tried
accessing directly. There is no error, just garbage results.
At some point I did try accessing the scales with an uint16_t
pointer (the scales are for sure 2-byte aligned), but was
still getting garbage. I guess, there must have been another bug.
No access to scales is via a uint16_t pointer and, after starting
from scratch from the C dequantize function, it finally works.
* metal : Q3_K 1st optimization pass
* metal : Q3_K second optimization pass - 29.6 ms/token
* metal : Q3_K cleanup
* metal : fixed accidentally broken Q2_K
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-12 21:39:21 +02:00
|
|
|
|
|
|
|
}
|
k-quants : support for super-block size of 64 (#2001)
* k_quants: WIP super-blocks with 64 weights
* k_quants: WIP super-blocks with 64 weights
Q6_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q4_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q2_K scalar and AVX2 works. Q2_K is way too slow (it is actually slower
than the scalar implementation)
* k_quants: WIP super-blocks with 64 weights
Q3_K scalar and AVX2 works.
* k_quants: WIP super-blocks with 64 weights
Q5_K scalar and AVX2 works, and with that all
k_quants are done on AVX2 and scalar
* k_quants: WIP super-blocks with 64 weights
Q6_K working on CUDA. Cannot make it run quite as gast as
with super-blocks with 256 weigths: 8% slower on 4080,
20% slower on the 1660 (but there we fit 1 less layer on the
GPU because pf the larger model size), so some fraction of
these 20% is due to that,
* k_quants: WIP super-blocks with 64 weights
Q4_K working on CUDA. ~10% slower on GTX-1660,
16% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q2_K working on CUDA. ~3% slower on GTX-1660,
10% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q3_K working on CUDA.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on CUDA, and with this CUDA is done.
* k_quants: WIP super-blocks with 64 weights
Q6_K working on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Q4_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q2_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q3_K working on ARM_NEON, but quite a bit slower than 256 weights.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on ARM_NEON, but quite a bit slower than 256 weights.
With that, we have full support for ARM_NEON, although
performance is not quite there.
* k_quants: WIP super-blocks with 64 weights
Slightly more efficient Q3_K and Q5_K
* k_quants: WIP super-blocks with 64 weights
Another small improvement for Q3_K and Q5_K on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Yet another speedup for Q5_K on ARM_NEON.
We are now within 10% of the QK_K = 256 version.
* k_quants: WIP super-blocks with 64 weights
* We are able to pass preprocessor macros to the Metal
compiler
* Q6_K works and is actually slightly more efficient than
the QK_K = 256 version (25.2 ms vs 25.8 ms)
* k_quants: WIP super-blocks with 64 weights
Q4_K works on Metal and is actually slightly faster
than QK_K = 256 (21.95 ms vs 24.0 ms).
* k_quants: WIP super-blocks with 64 weights
Q2_K works on Metal and is very slightly faster
than QK_K = 256 (23.8 ms vs 24.2 ms).
* k_quants: WIP super-blocks with 64 weights
Q3_K works on Metal and is slightly faster
than QK_K = 256 (26.6 ms vs 28.3 ms).
* k_quants: WIP super-blocks with 64 weights
Q5_K works on Metal and is slightly faster
than QK_K = 256 (23.7 ms vs 26.3 ms).
* k_quants: call them _K, not _k, also on Metal
* k_quants: correctly define QK_K in llama.cpp
* Fixed bug in q4_K quantization added with the 64-block addition
* Simplify via lambda
* k_quants: swicth Q3_K to 4-bit scales when QK_K = 64
Otherwise there isn't much benefit from this
quantization type. There is some very slight loss
in accuracy, but we reduce size by ~7%.
E.g., for OpenLLaMA-3B, Q3_K_S perplexity is
8.6131 with 8-bit scales and 8.6352 with 4-bit,
while file size decreases from 1.53G to 1.44G.
* k_quants: switch Q4_K to 4-bit scales when QK_K = 64
Here the loss in accuracy is greater than for Q3_K,
but the Q4_K points still move further to the left on
the perplexity vs size curve.
* k_quants: forgot to add the Metal changes in last commit
* k_quants: change Q5_K to be type 0 when QK_K = 64
Still needs AVX2 implementation
* k_quants: AVX2 implementation for new 64-weight Q5_K
* k_quants: 10% faster ARM_NEON Q5_K dot product
* k_quants: fixed issue caused by merging with master
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-26 18:43:07 +02:00
|
|
|
#else
|
2023-07-20 17:19:45 +02:00
|
|
|
float yl[8], yh[8];
|
k-quants : support for super-block size of 64 (#2001)
* k_quants: WIP super-blocks with 64 weights
* k_quants: WIP super-blocks with 64 weights
Q6_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q4_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q2_K scalar and AVX2 works. Q2_K is way too slow (it is actually slower
than the scalar implementation)
* k_quants: WIP super-blocks with 64 weights
Q3_K scalar and AVX2 works.
* k_quants: WIP super-blocks with 64 weights
Q5_K scalar and AVX2 works, and with that all
k_quants are done on AVX2 and scalar
* k_quants: WIP super-blocks with 64 weights
Q6_K working on CUDA. Cannot make it run quite as gast as
with super-blocks with 256 weigths: 8% slower on 4080,
20% slower on the 1660 (but there we fit 1 less layer on the
GPU because pf the larger model size), so some fraction of
these 20% is due to that,
* k_quants: WIP super-blocks with 64 weights
Q4_K working on CUDA. ~10% slower on GTX-1660,
16% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q2_K working on CUDA. ~3% slower on GTX-1660,
10% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q3_K working on CUDA.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on CUDA, and with this CUDA is done.
* k_quants: WIP super-blocks with 64 weights
Q6_K working on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Q4_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q2_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q3_K working on ARM_NEON, but quite a bit slower than 256 weights.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on ARM_NEON, but quite a bit slower than 256 weights.
With that, we have full support for ARM_NEON, although
performance is not quite there.
* k_quants: WIP super-blocks with 64 weights
Slightly more efficient Q3_K and Q5_K
* k_quants: WIP super-blocks with 64 weights
Another small improvement for Q3_K and Q5_K on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Yet another speedup for Q5_K on ARM_NEON.
We are now within 10% of the QK_K = 256 version.
* k_quants: WIP super-blocks with 64 weights
* We are able to pass preprocessor macros to the Metal
compiler
* Q6_K works and is actually slightly more efficient than
the QK_K = 256 version (25.2 ms vs 25.8 ms)
* k_quants: WIP super-blocks with 64 weights
Q4_K works on Metal and is actually slightly faster
than QK_K = 256 (21.95 ms vs 24.0 ms).
* k_quants: WIP super-blocks with 64 weights
Q2_K works on Metal and is very slightly faster
than QK_K = 256 (23.8 ms vs 24.2 ms).
* k_quants: WIP super-blocks with 64 weights
Q3_K works on Metal and is slightly faster
than QK_K = 256 (26.6 ms vs 28.3 ms).
* k_quants: WIP super-blocks with 64 weights
Q5_K works on Metal and is slightly faster
than QK_K = 256 (23.7 ms vs 26.3 ms).
* k_quants: call them _K, not _k, also on Metal
* k_quants: correctly define QK_K in llama.cpp
* Fixed bug in q4_K quantization added with the 64-block addition
* Simplify via lambda
* k_quants: swicth Q3_K to 4-bit scales when QK_K = 64
Otherwise there isn't much benefit from this
quantization type. There is some very slight loss
in accuracy, but we reduce size by ~7%.
E.g., for OpenLLaMA-3B, Q3_K_S perplexity is
8.6131 with 8-bit scales and 8.6352 with 4-bit,
while file size decreases from 1.53G to 1.44G.
* k_quants: switch Q4_K to 4-bit scales when QK_K = 64
Here the loss in accuracy is greater than for Q3_K,
but the Q4_K points still move further to the left on
the perplexity vs size curve.
* k_quants: forgot to add the Metal changes in last commit
* k_quants: change Q5_K to be type 0 when QK_K = 64
Still needs AVX2 implementation
* k_quants: AVX2 implementation for new 64-weight Q5_K
* k_quants: 10% faster ARM_NEON Q5_K dot product
* k_quants: fixed issue caused by merging with master
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-26 18:43:07 +02:00
|
|
|
|
2023-07-20 17:19:45 +02:00
|
|
|
const int il = 4 * (tiisg/8); // 0, 4, 8, 12
|
|
|
|
const int ix = tiisg%8;
|
2023-12-07 21:26:54 +01:00
|
|
|
const int iq = il/8; // 0, 0, 1, 1
|
2023-07-20 17:19:45 +02:00
|
|
|
const int in = il%8; // 0, 4, 0, 4
|
k-quants : support for super-block size of 64 (#2001)
* k_quants: WIP super-blocks with 64 weights
* k_quants: WIP super-blocks with 64 weights
Q6_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q4_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q2_K scalar and AVX2 works. Q2_K is way too slow (it is actually slower
than the scalar implementation)
* k_quants: WIP super-blocks with 64 weights
Q3_K scalar and AVX2 works.
* k_quants: WIP super-blocks with 64 weights
Q5_K scalar and AVX2 works, and with that all
k_quants are done on AVX2 and scalar
* k_quants: WIP super-blocks with 64 weights
Q6_K working on CUDA. Cannot make it run quite as gast as
with super-blocks with 256 weigths: 8% slower on 4080,
20% slower on the 1660 (but there we fit 1 less layer on the
GPU because pf the larger model size), so some fraction of
these 20% is due to that,
* k_quants: WIP super-blocks with 64 weights
Q4_K working on CUDA. ~10% slower on GTX-1660,
16% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q2_K working on CUDA. ~3% slower on GTX-1660,
10% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q3_K working on CUDA.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on CUDA, and with this CUDA is done.
* k_quants: WIP super-blocks with 64 weights
Q6_K working on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Q4_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q2_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q3_K working on ARM_NEON, but quite a bit slower than 256 weights.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on ARM_NEON, but quite a bit slower than 256 weights.
With that, we have full support for ARM_NEON, although
performance is not quite there.
* k_quants: WIP super-blocks with 64 weights
Slightly more efficient Q3_K and Q5_K
* k_quants: WIP super-blocks with 64 weights
Another small improvement for Q3_K and Q5_K on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Yet another speedup for Q5_K on ARM_NEON.
We are now within 10% of the QK_K = 256 version.
* k_quants: WIP super-blocks with 64 weights
* We are able to pass preprocessor macros to the Metal
compiler
* Q6_K works and is actually slightly more efficient than
the QK_K = 256 version (25.2 ms vs 25.8 ms)
* k_quants: WIP super-blocks with 64 weights
Q4_K works on Metal and is actually slightly faster
than QK_K = 256 (21.95 ms vs 24.0 ms).
* k_quants: WIP super-blocks with 64 weights
Q2_K works on Metal and is very slightly faster
than QK_K = 256 (23.8 ms vs 24.2 ms).
* k_quants: WIP super-blocks with 64 weights
Q3_K works on Metal and is slightly faster
than QK_K = 256 (26.6 ms vs 28.3 ms).
* k_quants: WIP super-blocks with 64 weights
Q5_K works on Metal and is slightly faster
than QK_K = 256 (23.7 ms vs 26.3 ms).
* k_quants: call them _K, not _k, also on Metal
* k_quants: correctly define QK_K in llama.cpp
* Fixed bug in q4_K quantization added with the 64-block addition
* Simplify via lambda
* k_quants: swicth Q3_K to 4-bit scales when QK_K = 64
Otherwise there isn't much benefit from this
quantization type. There is some very slight loss
in accuracy, but we reduce size by ~7%.
E.g., for OpenLLaMA-3B, Q3_K_S perplexity is
8.6131 with 8-bit scales and 8.6352 with 4-bit,
while file size decreases from 1.53G to 1.44G.
* k_quants: switch Q4_K to 4-bit scales when QK_K = 64
Here the loss in accuracy is greater than for Q3_K,
but the Q4_K points still move further to the left on
the perplexity vs size curve.
* k_quants: forgot to add the Metal changes in last commit
* k_quants: change Q5_K to be type 0 when QK_K = 64
Still needs AVX2 implementation
* k_quants: AVX2 implementation for new 64-weight Q5_K
* k_quants: 10% faster ARM_NEON Q5_K dot product
* k_quants: fixed issue caused by merging with master
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-26 18:43:07 +02:00
|
|
|
|
2023-07-20 17:19:45 +02:00
|
|
|
device const float * y = yy + ix*QK_K + il;
|
|
|
|
|
|
|
|
for (int i = ix; i < nb; i += 8) {
|
|
|
|
|
|
|
|
for (int l = 0; l < 4; ++l) {
|
|
|
|
yl[l+0] = y[l+ 0];
|
|
|
|
yl[l+4] = y[l+16];
|
|
|
|
yh[l+0] = y[l+32];
|
|
|
|
yh[l+4] = y[l+48];
|
|
|
|
}
|
|
|
|
|
|
|
|
device const half * dh = &x[i].d;
|
k-quants : support for super-block size of 64 (#2001)
* k_quants: WIP super-blocks with 64 weights
* k_quants: WIP super-blocks with 64 weights
Q6_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q4_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q2_K scalar and AVX2 works. Q2_K is way too slow (it is actually slower
than the scalar implementation)
* k_quants: WIP super-blocks with 64 weights
Q3_K scalar and AVX2 works.
* k_quants: WIP super-blocks with 64 weights
Q5_K scalar and AVX2 works, and with that all
k_quants are done on AVX2 and scalar
* k_quants: WIP super-blocks with 64 weights
Q6_K working on CUDA. Cannot make it run quite as gast as
with super-blocks with 256 weigths: 8% slower on 4080,
20% slower on the 1660 (but there we fit 1 less layer on the
GPU because pf the larger model size), so some fraction of
these 20% is due to that,
* k_quants: WIP super-blocks with 64 weights
Q4_K working on CUDA. ~10% slower on GTX-1660,
16% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q2_K working on CUDA. ~3% slower on GTX-1660,
10% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q3_K working on CUDA.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on CUDA, and with this CUDA is done.
* k_quants: WIP super-blocks with 64 weights
Q6_K working on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Q4_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q2_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q3_K working on ARM_NEON, but quite a bit slower than 256 weights.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on ARM_NEON, but quite a bit slower than 256 weights.
With that, we have full support for ARM_NEON, although
performance is not quite there.
* k_quants: WIP super-blocks with 64 weights
Slightly more efficient Q3_K and Q5_K
* k_quants: WIP super-blocks with 64 weights
Another small improvement for Q3_K and Q5_K on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Yet another speedup for Q5_K on ARM_NEON.
We are now within 10% of the QK_K = 256 version.
* k_quants: WIP super-blocks with 64 weights
* We are able to pass preprocessor macros to the Metal
compiler
* Q6_K works and is actually slightly more efficient than
the QK_K = 256 version (25.2 ms vs 25.8 ms)
* k_quants: WIP super-blocks with 64 weights
Q4_K works on Metal and is actually slightly faster
than QK_K = 256 (21.95 ms vs 24.0 ms).
* k_quants: WIP super-blocks with 64 weights
Q2_K works on Metal and is very slightly faster
than QK_K = 256 (23.8 ms vs 24.2 ms).
* k_quants: WIP super-blocks with 64 weights
Q3_K works on Metal and is slightly faster
than QK_K = 256 (26.6 ms vs 28.3 ms).
* k_quants: WIP super-blocks with 64 weights
Q5_K works on Metal and is slightly faster
than QK_K = 256 (23.7 ms vs 26.3 ms).
* k_quants: call them _K, not _k, also on Metal
* k_quants: correctly define QK_K in llama.cpp
* Fixed bug in q4_K quantization added with the 64-block addition
* Simplify via lambda
* k_quants: swicth Q3_K to 4-bit scales when QK_K = 64
Otherwise there isn't much benefit from this
quantization type. There is some very slight loss
in accuracy, but we reduce size by ~7%.
E.g., for OpenLLaMA-3B, Q3_K_S perplexity is
8.6131 with 8-bit scales and 8.6352 with 4-bit,
while file size decreases from 1.53G to 1.44G.
* k_quants: switch Q4_K to 4-bit scales when QK_K = 64
Here the loss in accuracy is greater than for Q3_K,
but the Q4_K points still move further to the left on
the perplexity vs size curve.
* k_quants: forgot to add the Metal changes in last commit
* k_quants: change Q5_K to be type 0 when QK_K = 64
Still needs AVX2 implementation
* k_quants: AVX2 implementation for new 64-weight Q5_K
* k_quants: 10% faster ARM_NEON Q5_K dot product
* k_quants: fixed issue caused by merging with master
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-26 18:43:07 +02:00
|
|
|
device const uint8_t * q = x[i].qs + il;
|
|
|
|
device const uint8_t * h = x[i].qh + in;
|
|
|
|
device const int8_t * s = x[i].scales;
|
|
|
|
|
2023-07-20 17:19:45 +02:00
|
|
|
for (int row = 0; row < 2; ++row) {
|
|
|
|
|
|
|
|
const float d = dh[0];
|
|
|
|
|
|
|
|
float2 acc = {0.f, 0.f};
|
|
|
|
for (int l = 0; l < 4; ++l) {
|
2023-12-07 21:26:54 +01:00
|
|
|
const uint8_t hl = h[l] >> iq;
|
2023-07-20 17:19:45 +02:00
|
|
|
acc[0] += yl[l+0] * s[0] * ((int16_t)(q[l+ 0] & 0x0F) - (hl & 0x01 ? 0 : 16))
|
|
|
|
+ yl[l+4] * s[1] * ((int16_t)(q[l+16] & 0x0F) - (hl & 0x04 ? 0 : 16));
|
|
|
|
acc[1] += yh[l+0] * s[2] * ((int16_t)(q[l+ 0] & 0xF0) - (hl & 0x10 ? 0 : 256))
|
|
|
|
+ yh[l+4] * s[3] * ((int16_t)(q[l+16] & 0xF0) - (hl & 0x40 ? 0 : 256));
|
|
|
|
}
|
|
|
|
sumf[row] += d * (acc[0] + 1.f/16.f * acc[1]);
|
|
|
|
|
|
|
|
q += step;
|
|
|
|
h += step;
|
|
|
|
s += step;
|
|
|
|
dh += step/2;
|
|
|
|
|
k-quants : support for super-block size of 64 (#2001)
* k_quants: WIP super-blocks with 64 weights
* k_quants: WIP super-blocks with 64 weights
Q6_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q4_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q2_K scalar and AVX2 works. Q2_K is way too slow (it is actually slower
than the scalar implementation)
* k_quants: WIP super-blocks with 64 weights
Q3_K scalar and AVX2 works.
* k_quants: WIP super-blocks with 64 weights
Q5_K scalar and AVX2 works, and with that all
k_quants are done on AVX2 and scalar
* k_quants: WIP super-blocks with 64 weights
Q6_K working on CUDA. Cannot make it run quite as gast as
with super-blocks with 256 weigths: 8% slower on 4080,
20% slower on the 1660 (but there we fit 1 less layer on the
GPU because pf the larger model size), so some fraction of
these 20% is due to that,
* k_quants: WIP super-blocks with 64 weights
Q4_K working on CUDA. ~10% slower on GTX-1660,
16% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q2_K working on CUDA. ~3% slower on GTX-1660,
10% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q3_K working on CUDA.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on CUDA, and with this CUDA is done.
* k_quants: WIP super-blocks with 64 weights
Q6_K working on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Q4_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q2_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q3_K working on ARM_NEON, but quite a bit slower than 256 weights.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on ARM_NEON, but quite a bit slower than 256 weights.
With that, we have full support for ARM_NEON, although
performance is not quite there.
* k_quants: WIP super-blocks with 64 weights
Slightly more efficient Q3_K and Q5_K
* k_quants: WIP super-blocks with 64 weights
Another small improvement for Q3_K and Q5_K on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Yet another speedup for Q5_K on ARM_NEON.
We are now within 10% of the QK_K = 256 version.
* k_quants: WIP super-blocks with 64 weights
* We are able to pass preprocessor macros to the Metal
compiler
* Q6_K works and is actually slightly more efficient than
the QK_K = 256 version (25.2 ms vs 25.8 ms)
* k_quants: WIP super-blocks with 64 weights
Q4_K works on Metal and is actually slightly faster
than QK_K = 256 (21.95 ms vs 24.0 ms).
* k_quants: WIP super-blocks with 64 weights
Q2_K works on Metal and is very slightly faster
than QK_K = 256 (23.8 ms vs 24.2 ms).
* k_quants: WIP super-blocks with 64 weights
Q3_K works on Metal and is slightly faster
than QK_K = 256 (26.6 ms vs 28.3 ms).
* k_quants: WIP super-blocks with 64 weights
Q5_K works on Metal and is slightly faster
than QK_K = 256 (23.7 ms vs 26.3 ms).
* k_quants: call them _K, not _k, also on Metal
* k_quants: correctly define QK_K in llama.cpp
* Fixed bug in q4_K quantization added with the 64-block addition
* Simplify via lambda
* k_quants: swicth Q3_K to 4-bit scales when QK_K = 64
Otherwise there isn't much benefit from this
quantization type. There is some very slight loss
in accuracy, but we reduce size by ~7%.
E.g., for OpenLLaMA-3B, Q3_K_S perplexity is
8.6131 with 8-bit scales and 8.6352 with 4-bit,
while file size decreases from 1.53G to 1.44G.
* k_quants: switch Q4_K to 4-bit scales when QK_K = 64
Here the loss in accuracy is greater than for Q3_K,
but the Q4_K points still move further to the left on
the perplexity vs size curve.
* k_quants: forgot to add the Metal changes in last commit
* k_quants: change Q5_K to be type 0 when QK_K = 64
Still needs AVX2 implementation
* k_quants: AVX2 implementation for new 64-weight Q5_K
* k_quants: 10% faster ARM_NEON Q5_K dot product
* k_quants: fixed issue caused by merging with master
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-26 18:43:07 +02:00
|
|
|
}
|
2023-07-20 17:19:45 +02:00
|
|
|
|
|
|
|
y += 8 * QK_K;
|
k-quants : support for super-block size of 64 (#2001)
* k_quants: WIP super-blocks with 64 weights
* k_quants: WIP super-blocks with 64 weights
Q6_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q4_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q2_K scalar and AVX2 works. Q2_K is way too slow (it is actually slower
than the scalar implementation)
* k_quants: WIP super-blocks with 64 weights
Q3_K scalar and AVX2 works.
* k_quants: WIP super-blocks with 64 weights
Q5_K scalar and AVX2 works, and with that all
k_quants are done on AVX2 and scalar
* k_quants: WIP super-blocks with 64 weights
Q6_K working on CUDA. Cannot make it run quite as gast as
with super-blocks with 256 weigths: 8% slower on 4080,
20% slower on the 1660 (but there we fit 1 less layer on the
GPU because pf the larger model size), so some fraction of
these 20% is due to that,
* k_quants: WIP super-blocks with 64 weights
Q4_K working on CUDA. ~10% slower on GTX-1660,
16% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q2_K working on CUDA. ~3% slower on GTX-1660,
10% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q3_K working on CUDA.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on CUDA, and with this CUDA is done.
* k_quants: WIP super-blocks with 64 weights
Q6_K working on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Q4_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q2_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q3_K working on ARM_NEON, but quite a bit slower than 256 weights.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on ARM_NEON, but quite a bit slower than 256 weights.
With that, we have full support for ARM_NEON, although
performance is not quite there.
* k_quants: WIP super-blocks with 64 weights
Slightly more efficient Q3_K and Q5_K
* k_quants: WIP super-blocks with 64 weights
Another small improvement for Q3_K and Q5_K on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Yet another speedup for Q5_K on ARM_NEON.
We are now within 10% of the QK_K = 256 version.
* k_quants: WIP super-blocks with 64 weights
* We are able to pass preprocessor macros to the Metal
compiler
* Q6_K works and is actually slightly more efficient than
the QK_K = 256 version (25.2 ms vs 25.8 ms)
* k_quants: WIP super-blocks with 64 weights
Q4_K works on Metal and is actually slightly faster
than QK_K = 256 (21.95 ms vs 24.0 ms).
* k_quants: WIP super-blocks with 64 weights
Q2_K works on Metal and is very slightly faster
than QK_K = 256 (23.8 ms vs 24.2 ms).
* k_quants: WIP super-blocks with 64 weights
Q3_K works on Metal and is slightly faster
than QK_K = 256 (26.6 ms vs 28.3 ms).
* k_quants: WIP super-blocks with 64 weights
Q5_K works on Metal and is slightly faster
than QK_K = 256 (23.7 ms vs 26.3 ms).
* k_quants: call them _K, not _k, also on Metal
* k_quants: correctly define QK_K in llama.cpp
* Fixed bug in q4_K quantization added with the 64-block addition
* Simplify via lambda
* k_quants: swicth Q3_K to 4-bit scales when QK_K = 64
Otherwise there isn't much benefit from this
quantization type. There is some very slight loss
in accuracy, but we reduce size by ~7%.
E.g., for OpenLLaMA-3B, Q3_K_S perplexity is
8.6131 with 8-bit scales and 8.6352 with 4-bit,
while file size decreases from 1.53G to 1.44G.
* k_quants: switch Q4_K to 4-bit scales when QK_K = 64
Here the loss in accuracy is greater than for Q3_K,
but the Q4_K points still move further to the left on
the perplexity vs size curve.
* k_quants: forgot to add the Metal changes in last commit
* k_quants: change Q5_K to be type 0 when QK_K = 64
Still needs AVX2 implementation
* k_quants: AVX2 implementation for new 64-weight Q5_K
* k_quants: 10% faster ARM_NEON Q5_K dot product
* k_quants: fixed issue caused by merging with master
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-26 18:43:07 +02:00
|
|
|
}
|
|
|
|
#endif
|
Metal implementation for all k_quants (#1807)
* metal : improve q4_K
28.3 -> 26.0 ms/token by avoiding a branch in the
calculation of the scales.
* metal : small improvement for Q4_K
* metal : still optimizing Q4_K
This commit pushes it down to 25.3 ms / token.
The crazy idea of using 6 bits for the scales is really costly on
Metal: if I remove the bit fiddling necessary to make the block
scales, time goes almost to the Q4_0 23 ms/token.
Before pushing the k-quants upstream I had a Q4_K variant that
had used 8-bit scales. It wasn't more accurate, used 0.125 bits more per weight,
was running slightly slower on the CPU (due to the larger model size
and being memory bound there), and the difference was entirely
negligible under CUDA. So, I decided to publish the version with 6-bit
scales. Perhaps I should re-consider and change to 8-bit scales?
* metal : some more optimizations
Q2_K: 25.4 ms/token
Q6_K: 27.3 ms/token
Q4_0: 22.8 ms/token
Q4_1: 23.1 ms/token
* metal : Q3_K support
Something is not quite right yet.
* metal : Q5_K support
Initial version achieves 31.2 ms/token, 210 GB/s
* metal : still not able to figure out why q3_K does not work
* Minor
* metal : yet another failed attempt to make q3_K work
* metal : optimize Q5_K
31.2 ms -> 27.8 ms.
250 GB/s.
* metal : q3_K still not working
Adding a heavily commented q3_K metal kernel to explain
my obviously faulty logic. Perhaps someone could spot the issue?
* metal : q3_K finally working
Not optimized at all.
What was the issue? The scales are not 4-bytes aligned,
and I was accessing them with a uint32_t pointer.
When I tried that on CUDA, I got an error (illegal memory access)
and added a memcpy to a local array of 3 uint32_t's.
But on Metal it told me there is no memcpy, so I tried
accessing directly. There is no error, just garbage results.
At some point I did try accessing the scales with an uint16_t
pointer (the scales are for sure 2-byte aligned), but was
still getting garbage. I guess, there must have been another bug.
No access to scales is via a uint16_t pointer and, after starting
from scratch from the C dequantize function, it finally works.
* metal : Q3_K 1st optimization pass
* metal : Q3_K second optimization pass - 29.6 ms/token
* metal : Q3_K cleanup
* metal : fixed accidentally broken Q2_K
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-12 21:39:21 +02:00
|
|
|
|
2023-07-20 17:19:45 +02:00
|
|
|
for (int row = 0; row < 2; ++row) {
|
|
|
|
const float tot = simd_sum(sumf[row]);
|
|
|
|
if (tiisg == 0) {
|
2023-12-07 21:26:54 +01:00
|
|
|
dst[r1*ne0 + im*ne0*ne1 + first_row + row] = tot;
|
2023-07-20 17:19:45 +02:00
|
|
|
}
|
Metal implementation for all k_quants (#1807)
* metal : improve q4_K
28.3 -> 26.0 ms/token by avoiding a branch in the
calculation of the scales.
* metal : small improvement for Q4_K
* metal : still optimizing Q4_K
This commit pushes it down to 25.3 ms / token.
The crazy idea of using 6 bits for the scales is really costly on
Metal: if I remove the bit fiddling necessary to make the block
scales, time goes almost to the Q4_0 23 ms/token.
Before pushing the k-quants upstream I had a Q4_K variant that
had used 8-bit scales. It wasn't more accurate, used 0.125 bits more per weight,
was running slightly slower on the CPU (due to the larger model size
and being memory bound there), and the difference was entirely
negligible under CUDA. So, I decided to publish the version with 6-bit
scales. Perhaps I should re-consider and change to 8-bit scales?
* metal : some more optimizations
Q2_K: 25.4 ms/token
Q6_K: 27.3 ms/token
Q4_0: 22.8 ms/token
Q4_1: 23.1 ms/token
* metal : Q3_K support
Something is not quite right yet.
* metal : Q5_K support
Initial version achieves 31.2 ms/token, 210 GB/s
* metal : still not able to figure out why q3_K does not work
* Minor
* metal : yet another failed attempt to make q3_K work
* metal : optimize Q5_K
31.2 ms -> 27.8 ms.
250 GB/s.
* metal : q3_K still not working
Adding a heavily commented q3_K metal kernel to explain
my obviously faulty logic. Perhaps someone could spot the issue?
* metal : q3_K finally working
Not optimized at all.
What was the issue? The scales are not 4-bytes aligned,
and I was accessing them with a uint32_t pointer.
When I tried that on CUDA, I got an error (illegal memory access)
and added a memcpy to a local array of 3 uint32_t's.
But on Metal it told me there is no memcpy, so I tried
accessing directly. There is no error, just garbage results.
At some point I did try accessing the scales with an uint16_t
pointer (the scales are for sure 2-byte aligned), but was
still getting garbage. I guess, there must have been another bug.
No access to scales is via a uint16_t pointer and, after starting
from scratch from the C dequantize function, it finally works.
* metal : Q3_K 1st optimization pass
* metal : Q3_K second optimization pass - 29.6 ms/token
* metal : Q3_K cleanup
* metal : fixed accidentally broken Q2_K
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-12 21:39:21 +02:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2023-12-13 13:04:25 +01:00
|
|
|
[[host_name("kernel_mul_mv_q5_K_f32")]]
|
|
|
|
kernel void kernel_mul_mv_q5_K_f32(
|
2023-06-08 18:46:22 +02:00
|
|
|
device const void * src0,
|
|
|
|
device const float * src1,
|
|
|
|
device float * dst,
|
|
|
|
constant int64_t & ne00,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant uint64_t & nb00,
|
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant uint64_t & nb02,
|
|
|
|
constant int64_t & ne10,
|
|
|
|
constant int64_t & ne11,
|
|
|
|
constant int64_t & ne12,
|
|
|
|
constant uint64_t & nb10,
|
|
|
|
constant uint64_t & nb11,
|
|
|
|
constant uint64_t & nb12,
|
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
|
|
|
constant uint & r2,
|
|
|
|
constant uint & r3,
|
2023-08-16 22:07:04 +02:00
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
2023-12-13 13:04:25 +01:00
|
|
|
uint tiisg[[thread_index_in_simdgroup]],
|
|
|
|
uint sgitg[[simdgroup_index_in_threadgroup]]) {
|
|
|
|
|
|
|
|
kernel_mul_mv_q5_K_f32_impl(src0, src1, dst, ne00, ne01, ne02, ne10, ne12, ne0, ne1, r2, r3, tgpig, tiisg, sgitg);
|
|
|
|
}
|
|
|
|
|
|
|
|
void kernel_mul_mv_q6_K_f32_impl(
|
|
|
|
device const void * src0,
|
|
|
|
device const float * src1,
|
|
|
|
device float * dst,
|
|
|
|
constant int64_t & ne00,
|
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant int64_t & ne10,
|
|
|
|
constant int64_t & ne12,
|
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
|
|
|
constant uint & r2,
|
|
|
|
constant uint & r3,
|
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint tiisg[[thread_index_in_simdgroup]],
|
|
|
|
uint sgitg[[simdgroup_index_in_threadgroup]]) {
|
2023-06-08 18:46:22 +02:00
|
|
|
|
|
|
|
const uint8_t kmask1 = 0x03;
|
|
|
|
const uint8_t kmask2 = 0x0C;
|
|
|
|
const uint8_t kmask3 = 0x30;
|
|
|
|
const uint8_t kmask4 = 0xC0;
|
|
|
|
|
|
|
|
const int nb = ne00/QK_K;
|
|
|
|
|
|
|
|
const int64_t r0 = tgpig.x;
|
|
|
|
const int64_t r1 = tgpig.y;
|
2023-12-07 21:26:54 +01:00
|
|
|
const int im = tgpig.z;
|
2023-06-08 18:46:22 +02:00
|
|
|
|
2023-07-20 17:19:45 +02:00
|
|
|
const int row = 2 * r0 + sgitg;
|
2023-12-07 21:26:54 +01:00
|
|
|
|
|
|
|
const uint i12 = im%ne12;
|
|
|
|
const uint i13 = im/ne12;
|
|
|
|
|
|
|
|
const uint offset0 = (i12/r2)*(nb*ne01) + (i13/r3)*(nb*ne01*ne02);
|
|
|
|
|
2023-08-16 22:07:04 +02:00
|
|
|
device const block_q6_K * x = (device const block_q6_K *) src0 + row * nb + offset0;
|
2023-12-07 21:26:54 +01:00
|
|
|
device const float * yy = (device const float *) src1 + r1*ne10 + im*ne00*ne1;
|
2023-06-08 18:46:22 +02:00
|
|
|
|
k-quants : support for super-block size of 64 (#2001)
* k_quants: WIP super-blocks with 64 weights
* k_quants: WIP super-blocks with 64 weights
Q6_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q4_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q2_K scalar and AVX2 works. Q2_K is way too slow (it is actually slower
than the scalar implementation)
* k_quants: WIP super-blocks with 64 weights
Q3_K scalar and AVX2 works.
* k_quants: WIP super-blocks with 64 weights
Q5_K scalar and AVX2 works, and with that all
k_quants are done on AVX2 and scalar
* k_quants: WIP super-blocks with 64 weights
Q6_K working on CUDA. Cannot make it run quite as gast as
with super-blocks with 256 weigths: 8% slower on 4080,
20% slower on the 1660 (but there we fit 1 less layer on the
GPU because pf the larger model size), so some fraction of
these 20% is due to that,
* k_quants: WIP super-blocks with 64 weights
Q4_K working on CUDA. ~10% slower on GTX-1660,
16% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q2_K working on CUDA. ~3% slower on GTX-1660,
10% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q3_K working on CUDA.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on CUDA, and with this CUDA is done.
* k_quants: WIP super-blocks with 64 weights
Q6_K working on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Q4_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q2_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q3_K working on ARM_NEON, but quite a bit slower than 256 weights.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on ARM_NEON, but quite a bit slower than 256 weights.
With that, we have full support for ARM_NEON, although
performance is not quite there.
* k_quants: WIP super-blocks with 64 weights
Slightly more efficient Q3_K and Q5_K
* k_quants: WIP super-blocks with 64 weights
Another small improvement for Q3_K and Q5_K on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Yet another speedup for Q5_K on ARM_NEON.
We are now within 10% of the QK_K = 256 version.
* k_quants: WIP super-blocks with 64 weights
* We are able to pass preprocessor macros to the Metal
compiler
* Q6_K works and is actually slightly more efficient than
the QK_K = 256 version (25.2 ms vs 25.8 ms)
* k_quants: WIP super-blocks with 64 weights
Q4_K works on Metal and is actually slightly faster
than QK_K = 256 (21.95 ms vs 24.0 ms).
* k_quants: WIP super-blocks with 64 weights
Q2_K works on Metal and is very slightly faster
than QK_K = 256 (23.8 ms vs 24.2 ms).
* k_quants: WIP super-blocks with 64 weights
Q3_K works on Metal and is slightly faster
than QK_K = 256 (26.6 ms vs 28.3 ms).
* k_quants: WIP super-blocks with 64 weights
Q5_K works on Metal and is slightly faster
than QK_K = 256 (23.7 ms vs 26.3 ms).
* k_quants: call them _K, not _k, also on Metal
* k_quants: correctly define QK_K in llama.cpp
* Fixed bug in q4_K quantization added with the 64-block addition
* Simplify via lambda
* k_quants: swicth Q3_K to 4-bit scales when QK_K = 64
Otherwise there isn't much benefit from this
quantization type. There is some very slight loss
in accuracy, but we reduce size by ~7%.
E.g., for OpenLLaMA-3B, Q3_K_S perplexity is
8.6131 with 8-bit scales and 8.6352 with 4-bit,
while file size decreases from 1.53G to 1.44G.
* k_quants: switch Q4_K to 4-bit scales when QK_K = 64
Here the loss in accuracy is greater than for Q3_K,
but the Q4_K points still move further to the left on
the perplexity vs size curve.
* k_quants: forgot to add the Metal changes in last commit
* k_quants: change Q5_K to be type 0 when QK_K = 64
Still needs AVX2 implementation
* k_quants: AVX2 implementation for new 64-weight Q5_K
* k_quants: 10% faster ARM_NEON Q5_K dot product
* k_quants: fixed issue caused by merging with master
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-26 18:43:07 +02:00
|
|
|
float sumf = 0;
|
|
|
|
|
|
|
|
#if QK_K == 256
|
2023-07-20 17:19:45 +02:00
|
|
|
const int tid = tiisg/2;
|
|
|
|
const int ix = tiisg%2;
|
|
|
|
const int ip = tid/8; // 0 or 1
|
|
|
|
const int il = tid%8;
|
2023-06-08 18:46:22 +02:00
|
|
|
const int n = 4;
|
Metal implementation for all k_quants (#1807)
* metal : improve q4_K
28.3 -> 26.0 ms/token by avoiding a branch in the
calculation of the scales.
* metal : small improvement for Q4_K
* metal : still optimizing Q4_K
This commit pushes it down to 25.3 ms / token.
The crazy idea of using 6 bits for the scales is really costly on
Metal: if I remove the bit fiddling necessary to make the block
scales, time goes almost to the Q4_0 23 ms/token.
Before pushing the k-quants upstream I had a Q4_K variant that
had used 8-bit scales. It wasn't more accurate, used 0.125 bits more per weight,
was running slightly slower on the CPU (due to the larger model size
and being memory bound there), and the difference was entirely
negligible under CUDA. So, I decided to publish the version with 6-bit
scales. Perhaps I should re-consider and change to 8-bit scales?
* metal : some more optimizations
Q2_K: 25.4 ms/token
Q6_K: 27.3 ms/token
Q4_0: 22.8 ms/token
Q4_1: 23.1 ms/token
* metal : Q3_K support
Something is not quite right yet.
* metal : Q5_K support
Initial version achieves 31.2 ms/token, 210 GB/s
* metal : still not able to figure out why q3_K does not work
* Minor
* metal : yet another failed attempt to make q3_K work
* metal : optimize Q5_K
31.2 ms -> 27.8 ms.
250 GB/s.
* metal : q3_K still not working
Adding a heavily commented q3_K metal kernel to explain
my obviously faulty logic. Perhaps someone could spot the issue?
* metal : q3_K finally working
Not optimized at all.
What was the issue? The scales are not 4-bytes aligned,
and I was accessing them with a uint32_t pointer.
When I tried that on CUDA, I got an error (illegal memory access)
and added a memcpy to a local array of 3 uint32_t's.
But on Metal it told me there is no memcpy, so I tried
accessing directly. There is no error, just garbage results.
At some point I did try accessing the scales with an uint16_t
pointer (the scales are for sure 2-byte aligned), but was
still getting garbage. I guess, there must have been another bug.
No access to scales is via a uint16_t pointer and, after starting
from scratch from the C dequantize function, it finally works.
* metal : Q3_K 1st optimization pass
* metal : Q3_K second optimization pass - 29.6 ms/token
* metal : Q3_K cleanup
* metal : fixed accidentally broken Q2_K
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-12 21:39:21 +02:00
|
|
|
const int l0 = n*il;
|
|
|
|
const int is = 8*ip + l0/16;
|
|
|
|
|
|
|
|
const int y_offset = 128*ip + l0;
|
|
|
|
const int q_offset_l = 64*ip + l0;
|
|
|
|
const int q_offset_h = 32*ip + l0;
|
2023-06-08 18:46:22 +02:00
|
|
|
|
2023-07-20 17:19:45 +02:00
|
|
|
for (int i = ix; i < nb; i += 2) {
|
2023-06-08 18:46:22 +02:00
|
|
|
|
2023-07-20 17:19:45 +02:00
|
|
|
device const uint8_t * q1 = x[i].ql + q_offset_l;
|
|
|
|
device const uint8_t * q2 = q1 + 32;
|
Metal implementation for all k_quants (#1807)
* metal : improve q4_K
28.3 -> 26.0 ms/token by avoiding a branch in the
calculation of the scales.
* metal : small improvement for Q4_K
* metal : still optimizing Q4_K
This commit pushes it down to 25.3 ms / token.
The crazy idea of using 6 bits for the scales is really costly on
Metal: if I remove the bit fiddling necessary to make the block
scales, time goes almost to the Q4_0 23 ms/token.
Before pushing the k-quants upstream I had a Q4_K variant that
had used 8-bit scales. It wasn't more accurate, used 0.125 bits more per weight,
was running slightly slower on the CPU (due to the larger model size
and being memory bound there), and the difference was entirely
negligible under CUDA. So, I decided to publish the version with 6-bit
scales. Perhaps I should re-consider and change to 8-bit scales?
* metal : some more optimizations
Q2_K: 25.4 ms/token
Q6_K: 27.3 ms/token
Q4_0: 22.8 ms/token
Q4_1: 23.1 ms/token
* metal : Q3_K support
Something is not quite right yet.
* metal : Q5_K support
Initial version achieves 31.2 ms/token, 210 GB/s
* metal : still not able to figure out why q3_K does not work
* Minor
* metal : yet another failed attempt to make q3_K work
* metal : optimize Q5_K
31.2 ms -> 27.8 ms.
250 GB/s.
* metal : q3_K still not working
Adding a heavily commented q3_K metal kernel to explain
my obviously faulty logic. Perhaps someone could spot the issue?
* metal : q3_K finally working
Not optimized at all.
What was the issue? The scales are not 4-bytes aligned,
and I was accessing them with a uint32_t pointer.
When I tried that on CUDA, I got an error (illegal memory access)
and added a memcpy to a local array of 3 uint32_t's.
But on Metal it told me there is no memcpy, so I tried
accessing directly. There is no error, just garbage results.
At some point I did try accessing the scales with an uint16_t
pointer (the scales are for sure 2-byte aligned), but was
still getting garbage. I guess, there must have been another bug.
No access to scales is via a uint16_t pointer and, after starting
from scratch from the C dequantize function, it finally works.
* metal : Q3_K 1st optimization pass
* metal : Q3_K second optimization pass - 29.6 ms/token
* metal : Q3_K cleanup
* metal : fixed accidentally broken Q2_K
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-12 21:39:21 +02:00
|
|
|
device const uint8_t * qh = x[i].qh + q_offset_h;
|
2023-06-08 18:46:22 +02:00
|
|
|
device const int8_t * sc = x[i].scales + is;
|
|
|
|
|
Metal implementation for all k_quants (#1807)
* metal : improve q4_K
28.3 -> 26.0 ms/token by avoiding a branch in the
calculation of the scales.
* metal : small improvement for Q4_K
* metal : still optimizing Q4_K
This commit pushes it down to 25.3 ms / token.
The crazy idea of using 6 bits for the scales is really costly on
Metal: if I remove the bit fiddling necessary to make the block
scales, time goes almost to the Q4_0 23 ms/token.
Before pushing the k-quants upstream I had a Q4_K variant that
had used 8-bit scales. It wasn't more accurate, used 0.125 bits more per weight,
was running slightly slower on the CPU (due to the larger model size
and being memory bound there), and the difference was entirely
negligible under CUDA. So, I decided to publish the version with 6-bit
scales. Perhaps I should re-consider and change to 8-bit scales?
* metal : some more optimizations
Q2_K: 25.4 ms/token
Q6_K: 27.3 ms/token
Q4_0: 22.8 ms/token
Q4_1: 23.1 ms/token
* metal : Q3_K support
Something is not quite right yet.
* metal : Q5_K support
Initial version achieves 31.2 ms/token, 210 GB/s
* metal : still not able to figure out why q3_K does not work
* Minor
* metal : yet another failed attempt to make q3_K work
* metal : optimize Q5_K
31.2 ms -> 27.8 ms.
250 GB/s.
* metal : q3_K still not working
Adding a heavily commented q3_K metal kernel to explain
my obviously faulty logic. Perhaps someone could spot the issue?
* metal : q3_K finally working
Not optimized at all.
What was the issue? The scales are not 4-bytes aligned,
and I was accessing them with a uint32_t pointer.
When I tried that on CUDA, I got an error (illegal memory access)
and added a memcpy to a local array of 3 uint32_t's.
But on Metal it told me there is no memcpy, so I tried
accessing directly. There is no error, just garbage results.
At some point I did try accessing the scales with an uint16_t
pointer (the scales are for sure 2-byte aligned), but was
still getting garbage. I guess, there must have been another bug.
No access to scales is via a uint16_t pointer and, after starting
from scratch from the C dequantize function, it finally works.
* metal : Q3_K 1st optimization pass
* metal : Q3_K second optimization pass - 29.6 ms/token
* metal : Q3_K cleanup
* metal : fixed accidentally broken Q2_K
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-12 21:39:21 +02:00
|
|
|
device const float * y = yy + i * QK_K + y_offset;
|
2023-06-08 18:46:22 +02:00
|
|
|
|
|
|
|
const float dall = x[i].d;
|
|
|
|
|
|
|
|
float4 sums = {0.f, 0.f, 0.f, 0.f};
|
|
|
|
for (int l = 0; l < n; ++l) {
|
2023-07-20 17:19:45 +02:00
|
|
|
sums[0] += y[l+ 0] * ((int8_t)((q1[l] & 0xF) | ((qh[l] & kmask1) << 4)) - 32);
|
|
|
|
sums[1] += y[l+32] * ((int8_t)((q2[l] & 0xF) | ((qh[l] & kmask2) << 2)) - 32);
|
|
|
|
sums[2] += y[l+64] * ((int8_t)((q1[l] >> 4) | ((qh[l] & kmask3) << 0)) - 32);
|
|
|
|
sums[3] += y[l+96] * ((int8_t)((q2[l] >> 4) | ((qh[l] & kmask4) >> 2)) - 32);
|
2023-06-08 18:46:22 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
sumf += dall * (sums[0] * sc[0] + sums[1] * sc[2] + sums[2] * sc[4] + sums[3] * sc[6]);
|
|
|
|
|
|
|
|
}
|
2023-07-20 17:19:45 +02:00
|
|
|
|
k-quants : support for super-block size of 64 (#2001)
* k_quants: WIP super-blocks with 64 weights
* k_quants: WIP super-blocks with 64 weights
Q6_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q4_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q2_K scalar and AVX2 works. Q2_K is way too slow (it is actually slower
than the scalar implementation)
* k_quants: WIP super-blocks with 64 weights
Q3_K scalar and AVX2 works.
* k_quants: WIP super-blocks with 64 weights
Q5_K scalar and AVX2 works, and with that all
k_quants are done on AVX2 and scalar
* k_quants: WIP super-blocks with 64 weights
Q6_K working on CUDA. Cannot make it run quite as gast as
with super-blocks with 256 weigths: 8% slower on 4080,
20% slower on the 1660 (but there we fit 1 less layer on the
GPU because pf the larger model size), so some fraction of
these 20% is due to that,
* k_quants: WIP super-blocks with 64 weights
Q4_K working on CUDA. ~10% slower on GTX-1660,
16% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q2_K working on CUDA. ~3% slower on GTX-1660,
10% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q3_K working on CUDA.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on CUDA, and with this CUDA is done.
* k_quants: WIP super-blocks with 64 weights
Q6_K working on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Q4_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q2_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q3_K working on ARM_NEON, but quite a bit slower than 256 weights.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on ARM_NEON, but quite a bit slower than 256 weights.
With that, we have full support for ARM_NEON, although
performance is not quite there.
* k_quants: WIP super-blocks with 64 weights
Slightly more efficient Q3_K and Q5_K
* k_quants: WIP super-blocks with 64 weights
Another small improvement for Q3_K and Q5_K on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Yet another speedup for Q5_K on ARM_NEON.
We are now within 10% of the QK_K = 256 version.
* k_quants: WIP super-blocks with 64 weights
* We are able to pass preprocessor macros to the Metal
compiler
* Q6_K works and is actually slightly more efficient than
the QK_K = 256 version (25.2 ms vs 25.8 ms)
* k_quants: WIP super-blocks with 64 weights
Q4_K works on Metal and is actually slightly faster
than QK_K = 256 (21.95 ms vs 24.0 ms).
* k_quants: WIP super-blocks with 64 weights
Q2_K works on Metal and is very slightly faster
than QK_K = 256 (23.8 ms vs 24.2 ms).
* k_quants: WIP super-blocks with 64 weights
Q3_K works on Metal and is slightly faster
than QK_K = 256 (26.6 ms vs 28.3 ms).
* k_quants: WIP super-blocks with 64 weights
Q5_K works on Metal and is slightly faster
than QK_K = 256 (23.7 ms vs 26.3 ms).
* k_quants: call them _K, not _k, also on Metal
* k_quants: correctly define QK_K in llama.cpp
* Fixed bug in q4_K quantization added with the 64-block addition
* Simplify via lambda
* k_quants: swicth Q3_K to 4-bit scales when QK_K = 64
Otherwise there isn't much benefit from this
quantization type. There is some very slight loss
in accuracy, but we reduce size by ~7%.
E.g., for OpenLLaMA-3B, Q3_K_S perplexity is
8.6131 with 8-bit scales and 8.6352 with 4-bit,
while file size decreases from 1.53G to 1.44G.
* k_quants: switch Q4_K to 4-bit scales when QK_K = 64
Here the loss in accuracy is greater than for Q3_K,
but the Q4_K points still move further to the left on
the perplexity vs size curve.
* k_quants: forgot to add the Metal changes in last commit
* k_quants: change Q5_K to be type 0 when QK_K = 64
Still needs AVX2 implementation
* k_quants: AVX2 implementation for new 64-weight Q5_K
* k_quants: 10% faster ARM_NEON Q5_K dot product
* k_quants: fixed issue caused by merging with master
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-26 18:43:07 +02:00
|
|
|
#else
|
2023-07-20 17:19:45 +02:00
|
|
|
const int ix = tiisg/4;
|
|
|
|
const int il = 4*(tiisg%4);
|
k-quants : support for super-block size of 64 (#2001)
* k_quants: WIP super-blocks with 64 weights
* k_quants: WIP super-blocks with 64 weights
Q6_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q4_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q2_K scalar and AVX2 works. Q2_K is way too slow (it is actually slower
than the scalar implementation)
* k_quants: WIP super-blocks with 64 weights
Q3_K scalar and AVX2 works.
* k_quants: WIP super-blocks with 64 weights
Q5_K scalar and AVX2 works, and with that all
k_quants are done on AVX2 and scalar
* k_quants: WIP super-blocks with 64 weights
Q6_K working on CUDA. Cannot make it run quite as gast as
with super-blocks with 256 weigths: 8% slower on 4080,
20% slower on the 1660 (but there we fit 1 less layer on the
GPU because pf the larger model size), so some fraction of
these 20% is due to that,
* k_quants: WIP super-blocks with 64 weights
Q4_K working on CUDA. ~10% slower on GTX-1660,
16% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q2_K working on CUDA. ~3% slower on GTX-1660,
10% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q3_K working on CUDA.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on CUDA, and with this CUDA is done.
* k_quants: WIP super-blocks with 64 weights
Q6_K working on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Q4_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q2_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q3_K working on ARM_NEON, but quite a bit slower than 256 weights.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on ARM_NEON, but quite a bit slower than 256 weights.
With that, we have full support for ARM_NEON, although
performance is not quite there.
* k_quants: WIP super-blocks with 64 weights
Slightly more efficient Q3_K and Q5_K
* k_quants: WIP super-blocks with 64 weights
Another small improvement for Q3_K and Q5_K on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Yet another speedup for Q5_K on ARM_NEON.
We are now within 10% of the QK_K = 256 version.
* k_quants: WIP super-blocks with 64 weights
* We are able to pass preprocessor macros to the Metal
compiler
* Q6_K works and is actually slightly more efficient than
the QK_K = 256 version (25.2 ms vs 25.8 ms)
* k_quants: WIP super-blocks with 64 weights
Q4_K works on Metal and is actually slightly faster
than QK_K = 256 (21.95 ms vs 24.0 ms).
* k_quants: WIP super-blocks with 64 weights
Q2_K works on Metal and is very slightly faster
than QK_K = 256 (23.8 ms vs 24.2 ms).
* k_quants: WIP super-blocks with 64 weights
Q3_K works on Metal and is slightly faster
than QK_K = 256 (26.6 ms vs 28.3 ms).
* k_quants: WIP super-blocks with 64 weights
Q5_K works on Metal and is slightly faster
than QK_K = 256 (23.7 ms vs 26.3 ms).
* k_quants: call them _K, not _k, also on Metal
* k_quants: correctly define QK_K in llama.cpp
* Fixed bug in q4_K quantization added with the 64-block addition
* Simplify via lambda
* k_quants: swicth Q3_K to 4-bit scales when QK_K = 64
Otherwise there isn't much benefit from this
quantization type. There is some very slight loss
in accuracy, but we reduce size by ~7%.
E.g., for OpenLLaMA-3B, Q3_K_S perplexity is
8.6131 with 8-bit scales and 8.6352 with 4-bit,
while file size decreases from 1.53G to 1.44G.
* k_quants: switch Q4_K to 4-bit scales when QK_K = 64
Here the loss in accuracy is greater than for Q3_K,
but the Q4_K points still move further to the left on
the perplexity vs size curve.
* k_quants: forgot to add the Metal changes in last commit
* k_quants: change Q5_K to be type 0 when QK_K = 64
Still needs AVX2 implementation
* k_quants: AVX2 implementation for new 64-weight Q5_K
* k_quants: 10% faster ARM_NEON Q5_K dot product
* k_quants: fixed issue caused by merging with master
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-26 18:43:07 +02:00
|
|
|
|
2023-07-20 17:19:45 +02:00
|
|
|
for (int i = ix; i < nb; i += 8) {
|
k-quants : support for super-block size of 64 (#2001)
* k_quants: WIP super-blocks with 64 weights
* k_quants: WIP super-blocks with 64 weights
Q6_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q4_K scalar and AVX2 works
* k_quants: WIP super-blocks with 64 weights
Q2_K scalar and AVX2 works. Q2_K is way too slow (it is actually slower
than the scalar implementation)
* k_quants: WIP super-blocks with 64 weights
Q3_K scalar and AVX2 works.
* k_quants: WIP super-blocks with 64 weights
Q5_K scalar and AVX2 works, and with that all
k_quants are done on AVX2 and scalar
* k_quants: WIP super-blocks with 64 weights
Q6_K working on CUDA. Cannot make it run quite as gast as
with super-blocks with 256 weigths: 8% slower on 4080,
20% slower on the 1660 (but there we fit 1 less layer on the
GPU because pf the larger model size), so some fraction of
these 20% is due to that,
* k_quants: WIP super-blocks with 64 weights
Q4_K working on CUDA. ~10% slower on GTX-1660,
16% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q2_K working on CUDA. ~3% slower on GTX-1660,
10% slower on 4080.
* k_quants: WIP super-blocks with 64 weights
Q3_K working on CUDA.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on CUDA, and with this CUDA is done.
* k_quants: WIP super-blocks with 64 weights
Q6_K working on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Q4_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q2_K working on ARM_NEON, but quite a bit slower than 256 weights
* k_quants: WIP super-blocks with 64 weights
Q3_K working on ARM_NEON, but quite a bit slower than 256 weights.
* k_quants: WIP super-blocks with 64 weights
Q5_K working on ARM_NEON, but quite a bit slower than 256 weights.
With that, we have full support for ARM_NEON, although
performance is not quite there.
* k_quants: WIP super-blocks with 64 weights
Slightly more efficient Q3_K and Q5_K
* k_quants: WIP super-blocks with 64 weights
Another small improvement for Q3_K and Q5_K on ARM_NEON
* k_quants: WIP super-blocks with 64 weights
Yet another speedup for Q5_K on ARM_NEON.
We are now within 10% of the QK_K = 256 version.
* k_quants: WIP super-blocks with 64 weights
* We are able to pass preprocessor macros to the Metal
compiler
* Q6_K works and is actually slightly more efficient than
the QK_K = 256 version (25.2 ms vs 25.8 ms)
* k_quants: WIP super-blocks with 64 weights
Q4_K works on Metal and is actually slightly faster
than QK_K = 256 (21.95 ms vs 24.0 ms).
* k_quants: WIP super-blocks with 64 weights
Q2_K works on Metal and is very slightly faster
than QK_K = 256 (23.8 ms vs 24.2 ms).
* k_quants: WIP super-blocks with 64 weights
Q3_K works on Metal and is slightly faster
than QK_K = 256 (26.6 ms vs 28.3 ms).
* k_quants: WIP super-blocks with 64 weights
Q5_K works on Metal and is slightly faster
than QK_K = 256 (23.7 ms vs 26.3 ms).
* k_quants: call them _K, not _k, also on Metal
* k_quants: correctly define QK_K in llama.cpp
* Fixed bug in q4_K quantization added with the 64-block addition
* Simplify via lambda
* k_quants: swicth Q3_K to 4-bit scales when QK_K = 64
Otherwise there isn't much benefit from this
quantization type. There is some very slight loss
in accuracy, but we reduce size by ~7%.
E.g., for OpenLLaMA-3B, Q3_K_S perplexity is
8.6131 with 8-bit scales and 8.6352 with 4-bit,
while file size decreases from 1.53G to 1.44G.
* k_quants: switch Q4_K to 4-bit scales when QK_K = 64
Here the loss in accuracy is greater than for Q3_K,
but the Q4_K points still move further to the left on
the perplexity vs size curve.
* k_quants: forgot to add the Metal changes in last commit
* k_quants: change Q5_K to be type 0 when QK_K = 64
Still needs AVX2 implementation
* k_quants: AVX2 implementation for new 64-weight Q5_K
* k_quants: 10% faster ARM_NEON Q5_K dot product
* k_quants: fixed issue caused by merging with master
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
2023-06-26 18:43:07 +02:00
|
|
|
device const float * y = yy + i * QK_K + il;
|
|
|
|
device const uint8_t * ql = x[i].ql + il;
|
|
|
|
device const uint8_t * qh = x[i].qh + il;
|
|
|
|
device const int8_t * s = x[i].scales;
|
|
|
|
|
|
|
|
const float d = x[i].d;
|
|
|
|
|
|
|
|
float4 sums = {0.f, 0.f, 0.f, 0.f};
|
|
|
|
for (int l = 0; l < 4; ++l) {
|
|
|
|
sums[0] += y[l+ 0] * ((int8_t)((ql[l+ 0] & 0xF) | ((qh[l] & kmask1) << 4)) - 32);
|
|
|
|
sums[1] += y[l+16] * ((int8_t)((ql[l+16] & 0xF) | ((qh[l] & kmask2) << 2)) - 32);
|
|
|
|
sums[2] += y[l+32] * ((int8_t)((ql[l+ 0] >> 4) | ((qh[l] & kmask3) >> 0)) - 32);
|
|
|
|
sums[3] += y[l+48] * ((int8_t)((ql[l+16] >> 4) | ((qh[l] & kmask4) >> 2)) - 32);
|
|
|
|
}
|
|
|
|
sumf += d * (sums[0] * s[0] + sums[1] * s[1] + sums[2] * s[2] + sums[3] * s[3]);
|
|
|
|
}
|
|
|
|
|
|
|
|
#endif
|
2023-06-08 18:46:22 +02:00
|
|
|
|
2023-07-20 17:19:45 +02:00
|
|
|
const float tot = simd_sum(sumf);
|
|
|
|
if (tiisg == 0) {
|
2023-12-07 21:26:54 +01:00
|
|
|
dst[r1*ne0 + im*ne0*ne1 + row] = tot;
|
2023-08-16 22:07:04 +02:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2023-12-13 13:04:25 +01:00
|
|
|
[[host_name("kernel_mul_mv_q6_K_f32")]]
|
|
|
|
kernel void kernel_mul_mv_q6_K_f32(
|
|
|
|
device const void * src0,
|
|
|
|
device const float * src1,
|
|
|
|
device float * dst,
|
|
|
|
constant int64_t & ne00,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant uint64_t & nb00,
|
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant uint64_t & nb02,
|
|
|
|
constant int64_t & ne10,
|
|
|
|
constant int64_t & ne11,
|
|
|
|
constant int64_t & ne12,
|
|
|
|
constant uint64_t & nb10,
|
|
|
|
constant uint64_t & nb11,
|
|
|
|
constant uint64_t & nb12,
|
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
|
|
|
constant uint & r2,
|
|
|
|
constant uint & r3,
|
2023-12-13 13:04:25 +01:00
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint tiisg[[thread_index_in_simdgroup]],
|
|
|
|
uint sgitg[[simdgroup_index_in_threadgroup]]) {
|
|
|
|
|
|
|
|
kernel_mul_mv_q6_K_f32_impl(src0, src1, dst, ne00, ne01, ne02, ne10, ne12, ne0, ne1, r2, r3, tgpig, tiisg, sgitg);
|
|
|
|
}
|
|
|
|
|
2024-01-08 16:02:32 +01:00
|
|
|
// ======================= "True" 2-bit
|
|
|
|
|
2024-01-11 20:39:39 +01:00
|
|
|
constexpr constant static uint64_t iq2xxs_grid[256] = {
|
2024-01-08 16:02:32 +01:00
|
|
|
0x0808080808080808, 0x080808080808082b, 0x0808080808081919, 0x0808080808082b08,
|
|
|
|
0x0808080808082b2b, 0x0808080808190819, 0x0808080808191908, 0x08080808082b0808,
|
|
|
|
0x08080808082b082b, 0x08080808082b2b08, 0x08080808082b2b2b, 0x0808080819080819,
|
|
|
|
0x0808080819081908, 0x0808080819190808, 0x0808080819192b08, 0x08080808192b0819,
|
|
|
|
0x08080808192b1908, 0x080808082b080808, 0x080808082b08082b, 0x080808082b082b2b,
|
|
|
|
0x080808082b2b082b, 0x0808081908080819, 0x0808081908081908, 0x0808081908190808,
|
|
|
|
0x0808081908191919, 0x0808081919080808, 0x080808192b081908, 0x080808192b192b08,
|
|
|
|
0x0808082b08080808, 0x0808082b0808082b, 0x0808082b082b082b, 0x0808082b2b08082b,
|
|
|
|
0x0808190808080819, 0x0808190808081908, 0x0808190808190808, 0x08081908082b0819,
|
|
|
|
0x08081908082b1908, 0x0808190819080808, 0x080819081908082b, 0x0808190819082b08,
|
|
|
|
0x08081908192b0808, 0x080819082b080819, 0x080819082b081908, 0x080819082b190808,
|
|
|
|
0x080819082b2b1908, 0x0808191908080808, 0x080819190808082b, 0x0808191908082b08,
|
|
|
|
0x08081919082b0808, 0x080819191908192b, 0x08081919192b2b19, 0x080819192b080808,
|
|
|
|
0x080819192b190819, 0x0808192b08082b19, 0x0808192b08190808, 0x0808192b19080808,
|
|
|
|
0x0808192b2b081908, 0x0808192b2b2b1908, 0x08082b0808080808, 0x08082b0808081919,
|
|
|
|
0x08082b0808082b08, 0x08082b0808191908, 0x08082b08082b2b08, 0x08082b0819080819,
|
|
|
|
0x08082b0819081908, 0x08082b0819190808, 0x08082b081919082b, 0x08082b082b082b08,
|
|
|
|
0x08082b1908081908, 0x08082b1919080808, 0x08082b2b0808082b, 0x08082b2b08191908,
|
|
|
|
0x0819080808080819, 0x0819080808081908, 0x0819080808190808, 0x08190808082b0819,
|
|
|
|
0x0819080819080808, 0x08190808192b0808, 0x081908082b081908, 0x081908082b190808,
|
|
|
|
0x081908082b191919, 0x0819081908080808, 0x0819081908082b08, 0x08190819082b0808,
|
|
|
|
0x0819081919190808, 0x0819081919192b2b, 0x081908192b080808, 0x0819082b082b1908,
|
|
|
|
0x0819082b19081919, 0x0819190808080808, 0x0819190808082b08, 0x08191908082b0808,
|
|
|
|
0x08191908082b1919, 0x0819190819082b19, 0x081919082b080808, 0x0819191908192b08,
|
|
|
|
0x08191919192b082b, 0x0819192b08080808, 0x0819192b0819192b, 0x08192b0808080819,
|
|
|
|
0x08192b0808081908, 0x08192b0808190808, 0x08192b0819080808, 0x08192b082b080819,
|
|
|
|
0x08192b1908080808, 0x08192b1908081919, 0x08192b192b2b0808, 0x08192b2b19190819,
|
|
|
|
0x082b080808080808, 0x082b08080808082b, 0x082b080808082b2b, 0x082b080819081908,
|
|
|
|
0x082b0808192b0819, 0x082b08082b080808, 0x082b08082b08082b, 0x082b0819082b2b19,
|
|
|
|
0x082b081919082b08, 0x082b082b08080808, 0x082b082b0808082b, 0x082b190808080819,
|
|
|
|
0x082b190808081908, 0x082b190808190808, 0x082b190819080808, 0x082b19081919192b,
|
|
|
|
0x082b191908080808, 0x082b191919080819, 0x082b1919192b1908, 0x082b192b2b190808,
|
|
|
|
0x082b2b0808082b08, 0x082b2b08082b0808, 0x082b2b082b191908, 0x082b2b2b19081908,
|
|
|
|
0x1908080808080819, 0x1908080808081908, 0x1908080808190808, 0x1908080808192b08,
|
|
|
|
0x19080808082b0819, 0x19080808082b1908, 0x1908080819080808, 0x1908080819082b08,
|
|
|
|
0x190808081919192b, 0x19080808192b0808, 0x190808082b080819, 0x190808082b081908,
|
|
|
|
0x190808082b190808, 0x1908081908080808, 0x19080819082b0808, 0x19080819192b0819,
|
|
|
|
0x190808192b080808, 0x190808192b081919, 0x1908082b08080819, 0x1908082b08190808,
|
|
|
|
0x1908082b19082b08, 0x1908082b1919192b, 0x1908082b192b2b08, 0x1908190808080808,
|
|
|
|
0x1908190808082b08, 0x19081908082b0808, 0x190819082b080808, 0x190819082b192b19,
|
|
|
|
0x190819190819082b, 0x19081919082b1908, 0x1908192b08080808, 0x19082b0808080819,
|
|
|
|
0x19082b0808081908, 0x19082b0808190808, 0x19082b0819080808, 0x19082b0819081919,
|
|
|
|
0x19082b1908080808, 0x19082b1919192b08, 0x19082b19192b0819, 0x19082b192b08082b,
|
|
|
|
0x19082b2b19081919, 0x19082b2b2b190808, 0x1919080808080808, 0x1919080808082b08,
|
|
|
|
0x1919080808190819, 0x1919080808192b19, 0x19190808082b0808, 0x191908082b080808,
|
|
|
|
0x191908082b082b08, 0x1919081908081908, 0x191908191908082b, 0x191908192b2b1908,
|
|
|
|
0x1919082b2b190819, 0x191919082b190808, 0x191919082b19082b, 0x1919191908082b2b,
|
|
|
|
0x1919192b08080819, 0x1919192b19191908, 0x19192b0808080808, 0x19192b0808190819,
|
|
|
|
0x19192b0808192b19, 0x19192b08192b1908, 0x19192b1919080808, 0x19192b2b08082b08,
|
|
|
|
0x192b080808081908, 0x192b080808190808, 0x192b080819080808, 0x192b0808192b2b08,
|
|
|
|
0x192b081908080808, 0x192b081919191919, 0x192b082b08192b08, 0x192b082b192b0808,
|
|
|
|
0x192b190808080808, 0x192b190808081919, 0x192b191908190808, 0x192b19190819082b,
|
|
|
|
0x192b19192b081908, 0x192b2b081908082b, 0x2b08080808080808, 0x2b0808080808082b,
|
|
|
|
0x2b08080808082b2b, 0x2b08080819080819, 0x2b0808082b08082b, 0x2b08081908081908,
|
|
|
|
0x2b08081908192b08, 0x2b08081919080808, 0x2b08082b08190819, 0x2b08190808080819,
|
|
|
|
0x2b08190808081908, 0x2b08190808190808, 0x2b08190808191919, 0x2b08190819080808,
|
|
|
|
0x2b081908192b0808, 0x2b08191908080808, 0x2b0819191908192b, 0x2b0819192b191908,
|
|
|
|
0x2b08192b08082b19, 0x2b08192b19080808, 0x2b08192b192b0808, 0x2b082b080808082b,
|
|
|
|
0x2b082b1908081908, 0x2b082b2b08190819, 0x2b19080808081908, 0x2b19080808190808,
|
|
|
|
0x2b190808082b1908, 0x2b19080819080808, 0x2b1908082b2b0819, 0x2b1908190819192b,
|
|
|
|
0x2b1908192b080808, 0x2b19082b19081919, 0x2b19190808080808, 0x2b191908082b082b,
|
|
|
|
0x2b19190819081908, 0x2b19191919190819, 0x2b192b082b080819, 0x2b192b19082b0808,
|
|
|
|
0x2b2b08080808082b, 0x2b2b080819190808, 0x2b2b08082b081919, 0x2b2b081908082b19,
|
|
|
|
0x2b2b082b08080808, 0x2b2b190808192b08, 0x2b2b2b0819190808, 0x2b2b2b1908081908,
|
|
|
|
};
|
|
|
|
|
2024-01-11 20:39:39 +01:00
|
|
|
constexpr constant static uint64_t iq2xs_grid[512] = {
|
|
|
|
0x0808080808080808, 0x080808080808082b, 0x0808080808081919, 0x0808080808082b08,
|
|
|
|
0x0808080808082b2b, 0x0808080808190819, 0x0808080808191908, 0x080808080819192b,
|
|
|
|
0x0808080808192b19, 0x08080808082b0808, 0x08080808082b082b, 0x08080808082b1919,
|
|
|
|
0x08080808082b2b08, 0x0808080819080819, 0x0808080819081908, 0x080808081908192b,
|
|
|
|
0x0808080819082b19, 0x0808080819190808, 0x080808081919082b, 0x0808080819191919,
|
|
|
|
0x0808080819192b08, 0x08080808192b0819, 0x08080808192b1908, 0x080808082b080808,
|
|
|
|
0x080808082b08082b, 0x080808082b081919, 0x080808082b082b08, 0x080808082b190819,
|
|
|
|
0x080808082b191908, 0x080808082b192b19, 0x080808082b2b0808, 0x0808081908080819,
|
|
|
|
0x0808081908081908, 0x080808190808192b, 0x0808081908082b19, 0x0808081908190808,
|
|
|
|
0x080808190819082b, 0x0808081908191919, 0x0808081908192b08, 0x0808081908192b2b,
|
|
|
|
0x08080819082b0819, 0x08080819082b1908, 0x0808081919080808, 0x080808191908082b,
|
|
|
|
0x0808081919081919, 0x0808081919082b08, 0x0808081919190819, 0x0808081919191908,
|
|
|
|
0x08080819192b0808, 0x08080819192b2b08, 0x080808192b080819, 0x080808192b081908,
|
|
|
|
0x080808192b190808, 0x0808082b08080808, 0x0808082b0808082b, 0x0808082b08081919,
|
|
|
|
0x0808082b08082b08, 0x0808082b08190819, 0x0808082b08191908, 0x0808082b082b0808,
|
|
|
|
0x0808082b19080819, 0x0808082b19081908, 0x0808082b19190808, 0x0808082b19191919,
|
|
|
|
0x0808082b2b080808, 0x0808082b2b082b2b, 0x0808190808080819, 0x0808190808081908,
|
|
|
|
0x080819080808192b, 0x0808190808082b19, 0x0808190808190808, 0x080819080819082b,
|
|
|
|
0x0808190808191919, 0x0808190808192b08, 0x08081908082b0819, 0x08081908082b1908,
|
|
|
|
0x0808190819080808, 0x080819081908082b, 0x0808190819081919, 0x0808190819082b08,
|
|
|
|
0x0808190819190819, 0x0808190819191908, 0x080819081919192b, 0x08081908192b0808,
|
|
|
|
0x080819082b080819, 0x080819082b081908, 0x080819082b190808, 0x0808191908080808,
|
|
|
|
0x080819190808082b, 0x0808191908081919, 0x0808191908082b08, 0x0808191908190819,
|
|
|
|
0x0808191908191908, 0x08081919082b0808, 0x0808191919080819, 0x0808191919081908,
|
|
|
|
0x0808191919190808, 0x08081919192b0819, 0x080819192b080808, 0x0808192b08080819,
|
|
|
|
0x0808192b08081908, 0x0808192b08190808, 0x0808192b082b192b, 0x0808192b19080808,
|
|
|
|
0x0808192b1908082b, 0x0808192b2b081908, 0x08082b0808080808, 0x08082b080808082b,
|
|
|
|
0x08082b0808081919, 0x08082b0808082b08, 0x08082b0808082b2b, 0x08082b0808190819,
|
|
|
|
0x08082b0808191908, 0x08082b08082b0808, 0x08082b08082b1919, 0x08082b0819080819,
|
|
|
|
0x08082b0819081908, 0x08082b0819190808, 0x08082b0819192b08, 0x08082b082b080808,
|
|
|
|
0x08082b082b2b0808, 0x08082b082b2b2b2b, 0x08082b1908080819, 0x08082b1908081908,
|
|
|
|
0x08082b1908190808, 0x08082b1919080808, 0x08082b192b080819, 0x08082b192b082b19,
|
|
|
|
0x08082b2b08080808, 0x08082b2b082b0808, 0x08082b2b082b2b08, 0x08082b2b2b19192b,
|
|
|
|
0x08082b2b2b2b0808, 0x0819080808080819, 0x0819080808081908, 0x081908080808192b,
|
|
|
|
0x0819080808082b19, 0x0819080808190808, 0x081908080819082b, 0x0819080808191919,
|
|
|
|
0x0819080808192b08, 0x08190808082b0819, 0x08190808082b1908, 0x0819080819080808,
|
|
|
|
0x081908081908082b, 0x0819080819081919, 0x0819080819082b08, 0x0819080819190819,
|
|
|
|
0x0819080819191908, 0x08190808192b0808, 0x08190808192b2b2b, 0x081908082b080819,
|
|
|
|
0x081908082b081908, 0x081908082b190808, 0x0819081908080808, 0x081908190808082b,
|
|
|
|
0x0819081908081919, 0x0819081908082b08, 0x0819081908190819, 0x0819081908191908,
|
|
|
|
0x08190819082b0808, 0x0819081919080819, 0x0819081919081908, 0x0819081919190808,
|
|
|
|
0x081908192b080808, 0x081908192b191908, 0x081908192b19192b, 0x0819082b08080819,
|
|
|
|
0x0819082b08081908, 0x0819082b0808192b, 0x0819082b08190808, 0x0819082b19080808,
|
|
|
|
0x0819082b192b0808, 0x0819190808080808, 0x081919080808082b, 0x0819190808081919,
|
|
|
|
0x0819190808082b08, 0x0819190808190819, 0x0819190808191908, 0x08191908082b0808,
|
|
|
|
0x0819190819080819, 0x0819190819081908, 0x0819190819082b19, 0x0819190819190808,
|
|
|
|
0x08191908192b1908, 0x081919082b080808, 0x0819191908080819, 0x0819191908081908,
|
|
|
|
0x0819191908190808, 0x0819191919080808, 0x0819192b08080808, 0x0819192b08191908,
|
|
|
|
0x0819192b19082b19, 0x08192b0808080819, 0x08192b0808081908, 0x08192b0808190808,
|
|
|
|
0x08192b080819082b, 0x08192b0819080808, 0x08192b0819191908, 0x08192b082b08192b,
|
|
|
|
0x08192b1908080808, 0x08192b1908081919, 0x08192b19192b192b, 0x08192b2b19190819,
|
|
|
|
0x08192b2b2b2b2b19, 0x082b080808080808, 0x082b08080808082b, 0x082b080808081919,
|
|
|
|
0x082b080808082b08, 0x082b080808082b2b, 0x082b080808190819, 0x082b080808191908,
|
|
|
|
0x082b0808082b0808, 0x082b080819080819, 0x082b080819081908, 0x082b080819190808,
|
|
|
|
0x082b08082b080808, 0x082b08082b2b0808, 0x082b081908080819, 0x082b081908081908,
|
|
|
|
0x082b081908190808, 0x082b081919080808, 0x082b081919082b08, 0x082b0819192b1919,
|
|
|
|
0x082b082b08080808, 0x082b082b082b082b, 0x082b082b2b080808, 0x082b082b2b2b2b08,
|
|
|
|
0x082b190808080819, 0x082b190808081908, 0x082b190808190808, 0x082b1908082b2b19,
|
|
|
|
0x082b190819080808, 0x082b191908080808, 0x082b191919080819, 0x082b19191919082b,
|
|
|
|
0x082b19192b192b19, 0x082b192b08080819, 0x082b192b08192b2b, 0x082b192b2b2b192b,
|
|
|
|
0x082b2b0808080808, 0x082b2b0808082b08, 0x082b2b0808082b2b, 0x082b2b08082b0808,
|
|
|
|
0x082b2b0819191919, 0x082b2b082b082b08, 0x082b2b082b2b082b, 0x082b2b19192b2b08,
|
|
|
|
0x082b2b192b190808, 0x082b2b2b08082b08, 0x082b2b2b082b0808, 0x082b2b2b2b08082b,
|
|
|
|
0x082b2b2b2b082b08, 0x082b2b2b2b082b2b, 0x1908080808080819, 0x1908080808081908,
|
|
|
|
0x190808080808192b, 0x1908080808082b19, 0x1908080808190808, 0x190808080819082b,
|
|
|
|
0x1908080808191919, 0x1908080808192b08, 0x19080808082b0819, 0x19080808082b1908,
|
|
|
|
0x1908080819080808, 0x190808081908082b, 0x1908080819081919, 0x1908080819082b08,
|
|
|
|
0x1908080819082b2b, 0x1908080819190819, 0x1908080819191908, 0x19080808192b0808,
|
|
|
|
0x19080808192b1919, 0x190808082b080819, 0x190808082b081908, 0x190808082b190808,
|
|
|
|
0x1908081908080808, 0x190808190808082b, 0x1908081908081919, 0x1908081908082b08,
|
|
|
|
0x1908081908190819, 0x1908081908191908, 0x19080819082b0808, 0x1908081919080819,
|
|
|
|
0x1908081919081908, 0x1908081919190808, 0x190808192b080808, 0x190808192b081919,
|
|
|
|
0x190808192b2b082b, 0x1908082b08080819, 0x1908082b08081908, 0x1908082b08190808,
|
|
|
|
0x1908082b0819082b, 0x1908082b082b2b19, 0x1908082b19080808, 0x1908190808080808,
|
|
|
|
0x190819080808082b, 0x1908190808081919, 0x1908190808082b08, 0x1908190808190819,
|
|
|
|
0x1908190808191908, 0x1908190808192b19, 0x19081908082b0808, 0x1908190819080819,
|
|
|
|
0x1908190819081908, 0x1908190819190808, 0x190819082b080808, 0x190819082b191908,
|
|
|
|
0x1908191908080819, 0x1908191908081908, 0x1908191908190808, 0x19081919082b1908,
|
|
|
|
0x1908191919080808, 0x190819192b192b2b, 0x1908192b08080808, 0x1908192b08082b2b,
|
|
|
|
0x1908192b19081908, 0x1908192b19190808, 0x19082b0808080819, 0x19082b0808081908,
|
|
|
|
0x19082b0808190808, 0x19082b0819080808, 0x19082b0819081919, 0x19082b0819191908,
|
|
|
|
0x19082b08192b082b, 0x19082b1908080808, 0x19082b1908190819, 0x19082b1919081908,
|
|
|
|
0x19082b1919190808, 0x19082b19192b2b19, 0x19082b2b08081908, 0x1919080808080808,
|
|
|
|
0x191908080808082b, 0x1919080808081919, 0x1919080808082b08, 0x1919080808190819,
|
|
|
|
0x1919080808191908, 0x19190808082b0808, 0x19190808082b2b08, 0x1919080819080819,
|
|
|
|
0x1919080819081908, 0x1919080819190808, 0x191908082b080808, 0x1919081908080819,
|
|
|
|
0x1919081908081908, 0x1919081908190808, 0x1919081908191919, 0x1919081919080808,
|
|
|
|
0x191908191908082b, 0x1919082b08080808, 0x1919082b19081908, 0x1919082b2b2b2b2b,
|
|
|
|
0x1919190808080819, 0x1919190808081908, 0x1919190808190808, 0x19191908082b0819,
|
|
|
|
0x1919190819080808, 0x19191908192b0808, 0x191919082b080819, 0x191919082b2b0819,
|
|
|
|
0x1919191908080808, 0x1919191908082b08, 0x191919192b080808, 0x191919192b082b08,
|
|
|
|
0x1919192b082b0819, 0x1919192b192b2b08, 0x1919192b2b2b0819, 0x19192b0808080808,
|
|
|
|
0x19192b0808191908, 0x19192b0819080819, 0x19192b0819190808, 0x19192b082b192b19,
|
|
|
|
0x19192b1908192b2b, 0x19192b1919080808, 0x19192b191908082b, 0x19192b2b2b081919,
|
|
|
|
0x192b080808080819, 0x192b080808081908, 0x192b080808190808, 0x192b080819080808,
|
|
|
|
0x192b080819191908, 0x192b0808192b082b, 0x192b08082b08192b, 0x192b08082b2b2b19,
|
|
|
|
0x192b081908080808, 0x192b082b082b1908, 0x192b082b19082b2b, 0x192b082b2b19082b,
|
|
|
|
0x192b190808080808, 0x192b19080819192b, 0x192b191908190808, 0x192b191919080808,
|
|
|
|
0x192b191919081919, 0x192b19192b2b1908, 0x192b2b0808080819, 0x192b2b08192b2b2b,
|
|
|
|
0x192b2b19082b1919, 0x192b2b2b0808192b, 0x192b2b2b19191908, 0x192b2b2b192b082b,
|
|
|
|
0x2b08080808080808, 0x2b0808080808082b, 0x2b08080808081919, 0x2b08080808082b08,
|
|
|
|
0x2b08080808190819, 0x2b08080808191908, 0x2b080808082b0808, 0x2b080808082b2b2b,
|
|
|
|
0x2b08080819080819, 0x2b08080819081908, 0x2b08080819190808, 0x2b0808082b080808,
|
|
|
|
0x2b0808082b08082b, 0x2b0808082b2b2b08, 0x2b0808082b2b2b2b, 0x2b08081908080819,
|
|
|
|
0x2b08081908081908, 0x2b0808190808192b, 0x2b08081908190808, 0x2b08081919080808,
|
|
|
|
0x2b08081919190819, 0x2b08081919192b19, 0x2b08082b08080808, 0x2b08082b082b0808,
|
|
|
|
0x2b08082b2b080808, 0x2b08082b2b08082b, 0x2b08082b2b2b0808, 0x2b08082b2b2b2b08,
|
|
|
|
0x2b08190808080819, 0x2b08190808081908, 0x2b08190808190808, 0x2b0819080819082b,
|
|
|
|
0x2b08190808191919, 0x2b08190819080808, 0x2b081908192b0808, 0x2b0819082b082b19,
|
|
|
|
0x2b08191908080808, 0x2b08191919081908, 0x2b0819192b2b1919, 0x2b08192b08192b08,
|
|
|
|
0x2b08192b192b2b2b, 0x2b082b0808080808, 0x2b082b0808082b08, 0x2b082b08082b1919,
|
|
|
|
0x2b082b0819192b2b, 0x2b082b082b080808, 0x2b082b082b08082b, 0x2b082b082b2b2b08,
|
|
|
|
0x2b082b190808192b, 0x2b082b2b082b082b, 0x2b082b2b2b080808, 0x2b082b2b2b082b08,
|
|
|
|
0x2b082b2b2b19192b, 0x2b082b2b2b2b2b08, 0x2b19080808080819, 0x2b19080808081908,
|
|
|
|
0x2b19080808190808, 0x2b19080819080808, 0x2b1908081919192b, 0x2b1908082b081908,
|
|
|
|
0x2b19081908080808, 0x2b190819082b082b, 0x2b190819192b1908, 0x2b19082b1919192b,
|
|
|
|
0x2b19082b2b082b19, 0x2b19190808080808, 0x2b19190808081919, 0x2b19190819081908,
|
|
|
|
0x2b19190819190808, 0x2b19190819192b08, 0x2b191919082b2b19, 0x2b1919192b190808,
|
|
|
|
0x2b1919192b19082b, 0x2b19192b19080819, 0x2b192b0819190819, 0x2b192b082b2b192b,
|
|
|
|
0x2b192b1919082b19, 0x2b192b2b08191919, 0x2b192b2b192b0808, 0x2b2b080808080808,
|
|
|
|
0x2b2b08080808082b, 0x2b2b080808082b08, 0x2b2b080808082b2b, 0x2b2b0808082b0808,
|
|
|
|
0x2b2b0808082b2b2b, 0x2b2b08082b2b0808, 0x2b2b081919190819, 0x2b2b081919192b19,
|
|
|
|
0x2b2b08192b2b192b, 0x2b2b082b08080808, 0x2b2b082b0808082b, 0x2b2b082b08082b08,
|
|
|
|
0x2b2b082b082b2b2b, 0x2b2b082b2b080808, 0x2b2b082b2b2b0808, 0x2b2b190819080808,
|
|
|
|
0x2b2b19082b191919, 0x2b2b192b192b1919, 0x2b2b192b2b192b08, 0x2b2b2b0808082b2b,
|
|
|
|
0x2b2b2b08082b0808, 0x2b2b2b08082b082b, 0x2b2b2b08082b2b08, 0x2b2b2b082b2b0808,
|
|
|
|
0x2b2b2b082b2b2b08, 0x2b2b2b1908081908, 0x2b2b2b192b081908, 0x2b2b2b192b08192b,
|
|
|
|
0x2b2b2b2b082b2b08, 0x2b2b2b2b082b2b2b, 0x2b2b2b2b2b190819, 0x2b2b2b2b2b2b2b2b,
|
|
|
|
};
|
|
|
|
|
2024-01-30 14:14:12 +01:00
|
|
|
constexpr constant static uint32_t iq3xxs_grid[256] = {
|
2024-01-30 18:15:28 +01:00
|
|
|
0x04040404, 0x04040414, 0x04040424, 0x04040c0c, 0x04040c1c, 0x04040c3e, 0x04041404, 0x04041414,
|
|
|
|
0x04041c0c, 0x04042414, 0x04043e1c, 0x04043e2c, 0x040c040c, 0x040c041c, 0x040c0c04, 0x040c0c14,
|
|
|
|
0x040c140c, 0x040c142c, 0x040c1c04, 0x040c1c14, 0x040c240c, 0x040c2c24, 0x040c3e04, 0x04140404,
|
|
|
|
0x04140414, 0x04140424, 0x04140c0c, 0x04141404, 0x04141414, 0x04141c0c, 0x04141c1c, 0x04141c3e,
|
|
|
|
0x04142c0c, 0x04142c3e, 0x04143e2c, 0x041c040c, 0x041c043e, 0x041c0c04, 0x041c0c14, 0x041c142c,
|
|
|
|
0x041c3e04, 0x04240c1c, 0x04241c3e, 0x04242424, 0x04242c3e, 0x04243e1c, 0x04243e2c, 0x042c040c,
|
|
|
|
0x042c043e, 0x042c1c14, 0x042c2c14, 0x04341c2c, 0x04343424, 0x043e0c04, 0x043e0c24, 0x043e0c34,
|
|
|
|
0x043e241c, 0x043e340c, 0x0c04040c, 0x0c04041c, 0x0c040c04, 0x0c040c14, 0x0c04140c, 0x0c04141c,
|
|
|
|
0x0c041c04, 0x0c041c14, 0x0c041c24, 0x0c04243e, 0x0c042c04, 0x0c0c0404, 0x0c0c0414, 0x0c0c0c0c,
|
2024-01-30 14:14:12 +01:00
|
|
|
0x0c0c1404, 0x0c0c1414, 0x0c14040c, 0x0c14041c, 0x0c140c04, 0x0c140c14, 0x0c14140c, 0x0c141c04,
|
2024-01-30 18:15:28 +01:00
|
|
|
0x0c143e14, 0x0c1c0404, 0x0c1c0414, 0x0c1c1404, 0x0c1c1c0c, 0x0c1c2434, 0x0c1c3434, 0x0c24040c,
|
|
|
|
0x0c24042c, 0x0c242c04, 0x0c2c1404, 0x0c2c1424, 0x0c2c2434, 0x0c2c3e0c, 0x0c34042c, 0x0c3e1414,
|
|
|
|
0x0c3e2404, 0x14040404, 0x14040414, 0x14040c0c, 0x14040c1c, 0x14041404, 0x14041414, 0x14041434,
|
2024-01-30 14:14:12 +01:00
|
|
|
0x14041c0c, 0x14042414, 0x140c040c, 0x140c041c, 0x140c042c, 0x140c0c04, 0x140c0c14, 0x140c140c,
|
2024-01-30 18:15:28 +01:00
|
|
|
0x140c1c04, 0x140c341c, 0x140c343e, 0x140c3e04, 0x14140404, 0x14140414, 0x14140c0c, 0x14140c3e,
|
|
|
|
0x14141404, 0x14141414, 0x14141c3e, 0x14142404, 0x14142c2c, 0x141c040c, 0x141c0c04, 0x141c0c24,
|
|
|
|
0x141c3e04, 0x141c3e24, 0x14241c2c, 0x14242c1c, 0x142c041c, 0x142c143e, 0x142c240c, 0x142c3e24,
|
|
|
|
0x143e040c, 0x143e041c, 0x143e0c34, 0x143e242c, 0x1c04040c, 0x1c040c04, 0x1c040c14, 0x1c04140c,
|
|
|
|
0x1c04141c, 0x1c042c04, 0x1c04342c, 0x1c043e14, 0x1c0c0404, 0x1c0c0414, 0x1c0c1404, 0x1c0c1c0c,
|
|
|
|
0x1c0c2424, 0x1c0c2434, 0x1c14040c, 0x1c14041c, 0x1c140c04, 0x1c14142c, 0x1c142c14, 0x1c143e14,
|
|
|
|
0x1c1c0c0c, 0x1c1c1c1c, 0x1c241c04, 0x1c24243e, 0x1c243e14, 0x1c2c0404, 0x1c2c0434, 0x1c2c1414,
|
|
|
|
0x1c2c2c2c, 0x1c340c24, 0x1c341c34, 0x1c34341c, 0x1c3e1c1c, 0x1c3e3404, 0x24040424, 0x24040c3e,
|
|
|
|
0x24041c2c, 0x24041c3e, 0x24042c1c, 0x24042c3e, 0x240c3e24, 0x24141404, 0x24141c3e, 0x24142404,
|
|
|
|
0x24143404, 0x24143434, 0x241c043e, 0x241c242c, 0x24240424, 0x24242c0c, 0x24243424, 0x242c142c,
|
|
|
|
0x242c241c, 0x242c3e04, 0x243e042c, 0x243e0c04, 0x243e0c14, 0x243e1c04, 0x2c040c14, 0x2c04240c,
|
|
|
|
0x2c043e04, 0x2c0c0404, 0x2c0c0434, 0x2c0c1434, 0x2c0c2c2c, 0x2c140c24, 0x2c141c14, 0x2c143e14,
|
|
|
|
0x2c1c0414, 0x2c1c2c1c, 0x2c240c04, 0x2c24141c, 0x2c24143e, 0x2c243e14, 0x2c2c0414, 0x2c2c1c0c,
|
|
|
|
0x2c342c04, 0x2c3e1424, 0x2c3e2414, 0x34041424, 0x34042424, 0x34042434, 0x34043424, 0x340c140c,
|
|
|
|
0x340c340c, 0x34140c3e, 0x34143424, 0x341c1c04, 0x341c1c34, 0x34242424, 0x342c042c, 0x342c2c14,
|
|
|
|
0x34341c1c, 0x343e041c, 0x343e140c, 0x3e04041c, 0x3e04042c, 0x3e04043e, 0x3e040c04, 0x3e041c14,
|
|
|
|
0x3e042c14, 0x3e0c1434, 0x3e0c2404, 0x3e140c14, 0x3e14242c, 0x3e142c14, 0x3e1c0404, 0x3e1c0c2c,
|
|
|
|
0x3e1c1c1c, 0x3e1c3404, 0x3e24140c, 0x3e24240c, 0x3e2c0404, 0x3e2c0414, 0x3e2c1424, 0x3e341c04,
|
2024-01-30 14:14:12 +01:00
|
|
|
};
|
|
|
|
|
|
|
|
|
2024-01-08 16:02:32 +01:00
|
|
|
constexpr constant static uint8_t ksigns_iq2xs[128] = {
|
|
|
|
0, 129, 130, 3, 132, 5, 6, 135, 136, 9, 10, 139, 12, 141, 142, 15,
|
|
|
|
144, 17, 18, 147, 20, 149, 150, 23, 24, 153, 154, 27, 156, 29, 30, 159,
|
|
|
|
160, 33, 34, 163, 36, 165, 166, 39, 40, 169, 170, 43, 172, 45, 46, 175,
|
|
|
|
48, 177, 178, 51, 180, 53, 54, 183, 184, 57, 58, 187, 60, 189, 190, 63,
|
|
|
|
192, 65, 66, 195, 68, 197, 198, 71, 72, 201, 202, 75, 204, 77, 78, 207,
|
|
|
|
80, 209, 210, 83, 212, 85, 86, 215, 216, 89, 90, 219, 92, 221, 222, 95,
|
|
|
|
96, 225, 226, 99, 228, 101, 102, 231, 232, 105, 106, 235, 108, 237, 238, 111,
|
|
|
|
240, 113, 114, 243, 116, 245, 246, 119, 120, 249, 250, 123, 252, 125, 126, 255,
|
|
|
|
};
|
|
|
|
|
|
|
|
constexpr constant static uint8_t kmask_iq2xs[8] = {1, 2, 4, 8, 16, 32, 64, 128};
|
|
|
|
|
|
|
|
void kernel_mul_mv_iq2_xxs_f32_impl(
|
|
|
|
device const void * src0,
|
|
|
|
device const float * src1,
|
|
|
|
device float * dst,
|
|
|
|
constant int64_t & ne00,
|
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant int64_t & ne10,
|
|
|
|
constant int64_t & ne12,
|
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
|
|
|
constant uint & r2,
|
|
|
|
constant uint & r3,
|
|
|
|
threadgroup int8_t * shared_values [[threadgroup(0)]],
|
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint tiisg[[thread_index_in_simdgroup]],
|
|
|
|
uint sgitg[[simdgroup_index_in_threadgroup]]) {
|
|
|
|
|
|
|
|
const int nb = ne00/QK_K;
|
|
|
|
const int r0 = tgpig.x;
|
|
|
|
const int r1 = tgpig.y;
|
|
|
|
const int im = tgpig.z;
|
|
|
|
|
|
|
|
const int first_row = (r0 * N_SIMDGROUP + sgitg) * N_DST;
|
|
|
|
const int ib_row = first_row * nb;
|
|
|
|
|
|
|
|
const uint i12 = im%ne12;
|
|
|
|
const uint i13 = im/ne12;
|
|
|
|
|
|
|
|
const uint offset0 = (i12/r2)*(nb*ne01) + (i13/r3)*(nb*ne01*ne02);
|
|
|
|
|
|
|
|
device const block_iq2_xxs * x = (device const block_iq2_xxs *) src0 + ib_row + offset0;
|
|
|
|
device const float * y = (device const float *) src1 + r1*ne10 + im*ne00*ne1;
|
|
|
|
|
|
|
|
float yl[32];
|
|
|
|
float sumf[N_DST]={0.f}, all_sum;
|
|
|
|
|
|
|
|
const int nb32 = nb * (QK_K / 32);
|
|
|
|
|
|
|
|
threadgroup uint64_t * values = (threadgroup uint64_t *)shared_values;
|
|
|
|
threadgroup uint8_t * shared_signs = (threadgroup uint8_t *)(values + 256);
|
|
|
|
{
|
|
|
|
int nval = 4;
|
|
|
|
int pos = (32*sgitg + tiisg)*nval;
|
2024-01-11 20:39:39 +01:00
|
|
|
for (int i = 0; i < nval; ++i) values[pos + i] = iq2xxs_grid[pos + i];
|
2024-01-08 16:02:32 +01:00
|
|
|
nval = 2;
|
|
|
|
pos = (32*sgitg + tiisg)*nval;
|
|
|
|
for (int i = 0; i < nval; ++i) shared_signs[pos+i] = ksigns_iq2xs[pos+i];
|
|
|
|
threadgroup_barrier(mem_flags::mem_threadgroup);
|
|
|
|
}
|
|
|
|
|
|
|
|
#if QK_K == 256
|
|
|
|
const int ix = tiisg;
|
|
|
|
|
|
|
|
device const float * y4 = y + 32 * ix;
|
|
|
|
|
|
|
|
for (int ib32 = ix; ib32 < nb32; ib32 += 32) {
|
|
|
|
|
|
|
|
for (int i = 0; i < 32; ++i) {
|
|
|
|
yl[i] = y4[i];
|
|
|
|
}
|
|
|
|
|
|
|
|
const int ibl = ib32 / (QK_K / 32);
|
|
|
|
const int ib = ib32 % (QK_K / 32);
|
|
|
|
|
|
|
|
device const block_iq2_xxs * xr = x + ibl;
|
|
|
|
device const uint16_t * q2 = xr->qs + 4 * ib;
|
|
|
|
device const half * dh = &xr->d;
|
|
|
|
|
|
|
|
for (int row = 0; row < N_DST; row++) {
|
|
|
|
|
|
|
|
const float db = dh[0];
|
|
|
|
device const uint8_t * aux8 = (device const uint8_t *)q2;
|
|
|
|
const uint32_t aux32 = q2[2] | (q2[3] << 16);
|
|
|
|
const float d = db * (0.5f + (aux32 >> 28));
|
|
|
|
|
|
|
|
float sum = 0;
|
|
|
|
for (int l = 0; l < 4; ++l) {
|
|
|
|
const threadgroup uint8_t * grid = (const threadgroup uint8_t *)(values + aux8[l]);
|
|
|
|
const uint8_t signs = shared_signs[(aux32 >> 7*l) & 127];
|
|
|
|
for (int j = 0; j < 8; ++j) {
|
|
|
|
sum += yl[8*l + j] * grid[j] * (signs & kmask_iq2xs[j] ? -1.f : 1.f);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
sumf[row] += d * sum;
|
|
|
|
|
|
|
|
dh += nb*sizeof(block_iq2_xxs)/2;
|
|
|
|
q2 += nb*sizeof(block_iq2_xxs)/2;
|
|
|
|
}
|
|
|
|
|
|
|
|
y4 += 32 * 32;
|
|
|
|
}
|
|
|
|
#else
|
|
|
|
// TODO
|
|
|
|
#endif
|
|
|
|
|
|
|
|
for (int row = 0; row < N_DST; ++row) {
|
|
|
|
all_sum = simd_sum(sumf[row]);
|
|
|
|
if (tiisg == 0) {
|
|
|
|
dst[r1*ne0 + im*ne0*ne1 + first_row + row] = all_sum * 0.25f;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
[[host_name("kernel_mul_mv_iq2_xxs_f32")]]
|
|
|
|
kernel void kernel_mul_mv_iq2_xxs_f32(
|
|
|
|
device const void * src0,
|
|
|
|
device const float * src1,
|
|
|
|
device float * dst,
|
|
|
|
constant int64_t & ne00,
|
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant uint64_t & nb00,
|
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant uint64_t & nb02,
|
|
|
|
constant int64_t & ne10,
|
|
|
|
constant int64_t & ne11,
|
|
|
|
constant int64_t & ne12,
|
|
|
|
constant uint64_t & nb10,
|
|
|
|
constant uint64_t & nb11,
|
|
|
|
constant uint64_t & nb12,
|
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
|
|
|
constant uint & r2,
|
|
|
|
constant uint & r3,
|
|
|
|
threadgroup int8_t * shared_values [[threadgroup(0)]],
|
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint tiisg[[thread_index_in_simdgroup]],
|
|
|
|
uint sgitg[[simdgroup_index_in_threadgroup]]) {
|
|
|
|
|
|
|
|
kernel_mul_mv_iq2_xxs_f32_impl(src0, src1, dst, ne00, ne01, ne02, ne10, ne12, ne0, ne1, r2, r3, shared_values, tgpig, tiisg, sgitg);
|
|
|
|
}
|
|
|
|
|
2024-01-11 20:39:39 +01:00
|
|
|
void kernel_mul_mv_iq2_xs_f32_impl(
|
|
|
|
device const void * src0,
|
|
|
|
device const float * src1,
|
|
|
|
device float * dst,
|
|
|
|
constant int64_t & ne00,
|
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant int64_t & ne10,
|
|
|
|
constant int64_t & ne12,
|
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
|
|
|
constant uint & r2,
|
|
|
|
constant uint & r3,
|
|
|
|
threadgroup int8_t * shared_values [[threadgroup(0)]],
|
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint tiisg[[thread_index_in_simdgroup]],
|
|
|
|
uint sgitg[[simdgroup_index_in_threadgroup]]) {
|
|
|
|
|
|
|
|
const int nb = ne00/QK_K;
|
|
|
|
const int r0 = tgpig.x;
|
|
|
|
const int r1 = tgpig.y;
|
|
|
|
const int im = tgpig.z;
|
|
|
|
|
|
|
|
const int first_row = (r0 * N_SIMDGROUP + sgitg) * N_DST;
|
|
|
|
const int ib_row = first_row * nb;
|
|
|
|
|
|
|
|
const uint i12 = im%ne12;
|
|
|
|
const uint i13 = im/ne12;
|
|
|
|
|
|
|
|
const uint offset0 = (i12/r2)*(nb*ne01) + (i13/r3)*(nb*ne01*ne02);
|
|
|
|
|
|
|
|
device const block_iq2_xs * x = (device const block_iq2_xs *) src0 + ib_row + offset0;
|
|
|
|
device const float * y = (device const float *) src1 + r1*ne10 + im*ne00*ne1;
|
|
|
|
|
|
|
|
float yl[32];
|
|
|
|
float sumf[N_DST]={0.f}, all_sum;
|
|
|
|
|
|
|
|
const int nb32 = nb * (QK_K / 32);
|
|
|
|
|
|
|
|
threadgroup uint64_t * values = (threadgroup uint64_t *)shared_values;
|
|
|
|
threadgroup uint8_t * shared_signs = (threadgroup uint8_t *)(values + 512);
|
|
|
|
{
|
|
|
|
int nval = 8;
|
|
|
|
int pos = (32*sgitg + tiisg)*nval;
|
|
|
|
for (int i = 0; i < nval; ++i) values[pos + i] = iq2xs_grid[pos + i];
|
|
|
|
nval = 2;
|
|
|
|
pos = (32*sgitg + tiisg)*nval;
|
|
|
|
for (int i = 0; i < nval; ++i) shared_signs[pos+i] = ksigns_iq2xs[pos+i];
|
|
|
|
threadgroup_barrier(mem_flags::mem_threadgroup);
|
|
|
|
}
|
|
|
|
|
|
|
|
#if QK_K == 256
|
|
|
|
const int ix = tiisg;
|
|
|
|
|
|
|
|
device const float * y4 = y + 32 * ix;
|
|
|
|
|
|
|
|
for (int ib32 = ix; ib32 < nb32; ib32 += 32) {
|
|
|
|
|
|
|
|
for (int i = 0; i < 32; ++i) {
|
|
|
|
yl[i] = y4[i];
|
|
|
|
}
|
|
|
|
|
|
|
|
const int ibl = ib32 / (QK_K / 32);
|
|
|
|
const int ib = ib32 % (QK_K / 32);
|
|
|
|
|
|
|
|
device const block_iq2_xs * xr = x + ibl;
|
|
|
|
device const uint16_t * q2 = xr->qs + 4 * ib;
|
|
|
|
device const uint8_t * sc = xr->scales + ib;
|
|
|
|
device const half * dh = &xr->d;
|
|
|
|
|
|
|
|
for (int row = 0; row < N_DST; row++) {
|
|
|
|
|
|
|
|
const float db = dh[0];
|
|
|
|
const uint8_t ls1 = sc[0] & 0xf;
|
|
|
|
const uint8_t ls2 = sc[0] >> 4;
|
|
|
|
const float d1 = db * (0.5f + ls1);
|
|
|
|
const float d2 = db * (0.5f + ls2);
|
|
|
|
|
|
|
|
float sum1 = 0, sum2 = 0;
|
|
|
|
for (int l = 0; l < 2; ++l) {
|
|
|
|
const threadgroup uint8_t * grid = (const threadgroup uint8_t *)(values + (q2[l] & 511));
|
|
|
|
const uint8_t signs = shared_signs[(q2[l] >> 9)];
|
|
|
|
for (int j = 0; j < 8; ++j) {
|
|
|
|
sum1 += yl[8*l + j] * grid[j] * (signs & kmask_iq2xs[j] ? -1.f : 1.f);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
for (int l = 2; l < 4; ++l) {
|
|
|
|
const threadgroup uint8_t * grid = (const threadgroup uint8_t *)(values + (q2[l] & 511));
|
|
|
|
const uint8_t signs = shared_signs[(q2[l] >> 9)];
|
|
|
|
for (int j = 0; j < 8; ++j) {
|
|
|
|
sum2 += yl[8*l + j] * grid[j] * (signs & kmask_iq2xs[j] ? -1.f : 1.f);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
sumf[row] += d1 * sum1 + d2 * sum2;
|
|
|
|
|
|
|
|
dh += nb*sizeof(block_iq2_xs)/2;
|
|
|
|
q2 += nb*sizeof(block_iq2_xs)/2;
|
|
|
|
sc += nb*sizeof(block_iq2_xs);
|
|
|
|
}
|
|
|
|
|
|
|
|
y4 += 32 * 32;
|
|
|
|
}
|
|
|
|
#else
|
|
|
|
// TODO
|
|
|
|
#endif
|
|
|
|
|
|
|
|
for (int row = 0; row < N_DST; ++row) {
|
|
|
|
all_sum = simd_sum(sumf[row]);
|
|
|
|
if (tiisg == 0) {
|
|
|
|
dst[r1*ne0 + im*ne0*ne1 + first_row + row] = all_sum * 0.25f;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
[[host_name("kernel_mul_mv_iq2_xs_f32")]]
|
|
|
|
kernel void kernel_mul_mv_iq2_xs_f32(
|
|
|
|
device const void * src0,
|
|
|
|
device const float * src1,
|
|
|
|
device float * dst,
|
|
|
|
constant int64_t & ne00,
|
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant uint64_t & nb00,
|
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant uint64_t & nb02,
|
|
|
|
constant int64_t & ne10,
|
|
|
|
constant int64_t & ne11,
|
|
|
|
constant int64_t & ne12,
|
|
|
|
constant uint64_t & nb10,
|
|
|
|
constant uint64_t & nb11,
|
|
|
|
constant uint64_t & nb12,
|
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
|
|
|
constant uint & r2,
|
|
|
|
constant uint & r3,
|
|
|
|
threadgroup int8_t * shared_values [[threadgroup(0)]],
|
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint tiisg[[thread_index_in_simdgroup]],
|
|
|
|
uint sgitg[[simdgroup_index_in_threadgroup]]) {
|
|
|
|
|
|
|
|
kernel_mul_mv_iq2_xs_f32_impl(src0, src1, dst, ne00, ne01, ne02, ne10, ne12, ne0, ne1, r2, r3, shared_values, tgpig, tiisg, sgitg);
|
|
|
|
}
|
|
|
|
|
2024-01-30 14:14:12 +01:00
|
|
|
void kernel_mul_mv_iq3_xxs_f32_impl(
|
|
|
|
device const void * src0,
|
|
|
|
device const float * src1,
|
|
|
|
device float * dst,
|
|
|
|
constant int64_t & ne00,
|
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant int64_t & ne10,
|
|
|
|
constant int64_t & ne12,
|
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
|
|
|
constant uint & r2,
|
|
|
|
constant uint & r3,
|
|
|
|
threadgroup int8_t * shared_values [[threadgroup(0)]],
|
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint tiisg[[thread_index_in_simdgroup]],
|
|
|
|
uint sgitg[[simdgroup_index_in_threadgroup]]) {
|
|
|
|
|
|
|
|
const int nb = ne00/QK_K;
|
|
|
|
const int r0 = tgpig.x;
|
|
|
|
const int r1 = tgpig.y;
|
|
|
|
const int im = tgpig.z;
|
|
|
|
|
|
|
|
const int first_row = (r0 * N_SIMDGROUP + sgitg) * N_DST;
|
|
|
|
const int ib_row = first_row * nb;
|
|
|
|
|
|
|
|
const uint i12 = im%ne12;
|
|
|
|
const uint i13 = im/ne12;
|
|
|
|
|
|
|
|
const uint offset0 = (i12/r2)*(nb*ne01) + (i13/r3)*(nb*ne01*ne02);
|
|
|
|
|
|
|
|
device const block_iq3_xxs * x = (device const block_iq3_xxs *) src0 + ib_row + offset0;
|
|
|
|
device const float * y = (device const float *) src1 + r1*ne10 + im*ne00*ne1;
|
|
|
|
|
|
|
|
float yl[32];
|
|
|
|
float sumf[N_DST]={0.f}, all_sum;
|
|
|
|
|
|
|
|
const int nb32 = nb * (QK_K / 32);
|
|
|
|
|
|
|
|
threadgroup uint32_t * values = (threadgroup uint32_t *)shared_values;
|
|
|
|
threadgroup uint8_t * shared_signs = (threadgroup uint8_t *)(values + 256);
|
|
|
|
{
|
|
|
|
int nval = 4;
|
|
|
|
int pos = (32*sgitg + tiisg)*nval;
|
|
|
|
for (int i = 0; i < nval; ++i) values[pos + i] = iq3xxs_grid[pos + i];
|
|
|
|
nval = 2;
|
|
|
|
pos = (32*sgitg + tiisg)*nval;
|
|
|
|
for (int i = 0; i < nval; ++i) shared_signs[pos+i] = ksigns_iq2xs[pos+i];
|
|
|
|
threadgroup_barrier(mem_flags::mem_threadgroup);
|
|
|
|
}
|
|
|
|
|
|
|
|
#if QK_K == 256
|
|
|
|
const int ix = tiisg;
|
|
|
|
|
|
|
|
device const float * y4 = y + 32 * ix;
|
|
|
|
|
|
|
|
for (int ib32 = ix; ib32 < nb32; ib32 += 32) {
|
|
|
|
|
|
|
|
for (int i = 0; i < 32; ++i) {
|
|
|
|
yl[i] = y4[i];
|
|
|
|
}
|
|
|
|
|
|
|
|
const int ibl = ib32 / (QK_K / 32);
|
|
|
|
const int ib = ib32 % (QK_K / 32);
|
|
|
|
|
|
|
|
device const block_iq3_xxs * xr = x + ibl;
|
|
|
|
device const uint8_t * q3 = xr->qs + 8 * ib;
|
|
|
|
device const uint16_t * gas = (device const uint16_t *)(xr->qs + QK_K/4) + 2 * ib;
|
|
|
|
device const half * dh = &xr->d;
|
|
|
|
|
|
|
|
for (int row = 0; row < N_DST; row++) {
|
|
|
|
|
|
|
|
const float db = dh[0];
|
|
|
|
const uint32_t aux32 = gas[0] | (gas[1] << 16);
|
|
|
|
const float d = db * (0.5f + (aux32 >> 28));
|
|
|
|
|
|
|
|
float2 sum = {0};
|
|
|
|
for (int l = 0; l < 4; ++l) {
|
|
|
|
const threadgroup uint8_t * grid1 = (const threadgroup uint8_t *)(values + q3[2*l+0]);
|
|
|
|
const threadgroup uint8_t * grid2 = (const threadgroup uint8_t *)(values + q3[2*l+1]);
|
|
|
|
const uint8_t signs = shared_signs[(aux32 >> 7*l) & 127];
|
|
|
|
for (int j = 0; j < 4; ++j) {
|
|
|
|
sum[0] += yl[8*l + j + 0] * grid1[j] * (signs & kmask_iq2xs[j+0] ? -1.f : 1.f);
|
|
|
|
sum[1] += yl[8*l + j + 4] * grid2[j] * (signs & kmask_iq2xs[j+4] ? -1.f : 1.f);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
sumf[row] += d * (sum[0] + sum[1]);
|
|
|
|
|
|
|
|
dh += nb*sizeof(block_iq3_xxs)/2;
|
|
|
|
q3 += nb*sizeof(block_iq3_xxs);
|
|
|
|
gas += nb*sizeof(block_iq3_xxs)/2;
|
|
|
|
}
|
|
|
|
|
|
|
|
y4 += 32 * 32;
|
|
|
|
}
|
|
|
|
#else
|
|
|
|
// TODO
|
|
|
|
#endif
|
|
|
|
|
|
|
|
for (int row = 0; row < N_DST; ++row) {
|
|
|
|
all_sum = simd_sum(sumf[row]);
|
|
|
|
if (tiisg == 0) {
|
|
|
|
dst[r1*ne0 + im*ne0*ne1 + first_row + row] = all_sum * 0.5f;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
[[host_name("kernel_mul_mv_iq3_xxs_f32")]]
|
|
|
|
kernel void kernel_mul_mv_iq3_xxs_f32(
|
|
|
|
device const void * src0,
|
|
|
|
device const float * src1,
|
|
|
|
device float * dst,
|
|
|
|
constant int64_t & ne00,
|
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant uint64_t & nb00,
|
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant uint64_t & nb02,
|
|
|
|
constant int64_t & ne10,
|
|
|
|
constant int64_t & ne11,
|
|
|
|
constant int64_t & ne12,
|
|
|
|
constant uint64_t & nb10,
|
|
|
|
constant uint64_t & nb11,
|
|
|
|
constant uint64_t & nb12,
|
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
|
|
|
constant uint & r2,
|
|
|
|
constant uint & r3,
|
|
|
|
threadgroup int8_t * shared_values [[threadgroup(0)]],
|
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint tiisg[[thread_index_in_simdgroup]],
|
|
|
|
uint sgitg[[simdgroup_index_in_threadgroup]]) {
|
|
|
|
|
|
|
|
kernel_mul_mv_iq3_xxs_f32_impl(src0, src1, dst, ne00, ne01, ne02, ne10, ne12, ne0, ne1, r2, r3, shared_values, tgpig, tiisg, sgitg);
|
|
|
|
}
|
|
|
|
|
|
|
|
|
2023-08-16 22:07:04 +02:00
|
|
|
//============================= templates and their specializations =============================
|
|
|
|
|
2023-09-15 10:09:24 +02:00
|
|
|
// NOTE: this is not dequantizing - we are simply fitting the template
|
|
|
|
template <typename type4x4>
|
|
|
|
void dequantize_f32(device const float4x4 * src, short il, thread type4x4 & reg) {
|
|
|
|
float4x4 temp = *(((device float4x4 *)src));
|
|
|
|
for (int i = 0; i < 16; i++){
|
|
|
|
reg[i/4][i%4] = temp[i/4][i%4];
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2023-08-16 22:07:04 +02:00
|
|
|
template <typename type4x4>
|
|
|
|
void dequantize_f16(device const half4x4 * src, short il, thread type4x4 & reg) {
|
|
|
|
half4x4 temp = *(((device half4x4 *)src));
|
|
|
|
for (int i = 0; i < 16; i++){
|
|
|
|
reg[i/4][i%4] = temp[i/4][i%4];
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
template <typename type4x4>
|
|
|
|
void dequantize_q4_0(device const block_q4_0 *xb, short il, thread type4x4 & reg) {
|
|
|
|
device const uint16_t * qs = ((device const uint16_t *)xb + 1);
|
2023-09-11 09:30:11 +02:00
|
|
|
const float d1 = il ? (xb->d / 16.h) : xb->d;
|
|
|
|
const float d2 = d1 / 256.f;
|
|
|
|
const float md = -8.h * xb->d;
|
2023-08-16 22:07:04 +02:00
|
|
|
const ushort mask0 = il ? 0x00F0 : 0x000F;
|
2023-09-11 09:30:11 +02:00
|
|
|
const ushort mask1 = mask0 << 8;
|
2023-08-16 22:07:04 +02:00
|
|
|
|
|
|
|
for (int i=0;i<8;i++) {
|
2023-09-11 09:30:11 +02:00
|
|
|
reg[i/2][2*(i%2)+0] = d1 * (qs[i] & mask0) + md;
|
|
|
|
reg[i/2][2*(i%2)+1] = d2 * (qs[i] & mask1) + md;
|
2023-08-16 22:07:04 +02:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
template <typename type4x4>
|
|
|
|
void dequantize_q4_1(device const block_q4_1 *xb, short il, thread type4x4 & reg) {
|
|
|
|
device const uint16_t * qs = ((device const uint16_t *)xb + 2);
|
2023-09-11 09:30:11 +02:00
|
|
|
const float d1 = il ? (xb->d / 16.h) : xb->d;
|
|
|
|
const float d2 = d1 / 256.f;
|
|
|
|
const float m = xb->m;
|
2023-08-16 22:07:04 +02:00
|
|
|
const ushort mask0 = il ? 0x00F0 : 0x000F;
|
2023-09-11 09:30:11 +02:00
|
|
|
const ushort mask1 = mask0 << 8;
|
2023-08-16 22:07:04 +02:00
|
|
|
|
|
|
|
for (int i=0;i<8;i++) {
|
2023-09-11 09:30:11 +02:00
|
|
|
reg[i/2][2*(i%2)+0] = ((qs[i] & mask0) * d1) + m;
|
|
|
|
reg[i/2][2*(i%2)+1] = ((qs[i] & mask1) * d2) + m;
|
2023-08-16 22:07:04 +02:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2023-10-18 14:21:48 +02:00
|
|
|
template <typename type4x4>
|
|
|
|
void dequantize_q5_0(device const block_q5_0 *xb, short il, thread type4x4 & reg) {
|
|
|
|
device const uint16_t * qs = ((device const uint16_t *)xb + 3);
|
|
|
|
const float d = xb->d;
|
|
|
|
const float md = -16.h * xb->d;
|
|
|
|
const ushort mask = il ? 0x00F0 : 0x000F;
|
|
|
|
|
|
|
|
const uint32_t qh = *((device const uint32_t *)xb->qh);
|
|
|
|
|
|
|
|
const int x_mv = il ? 4 : 0;
|
|
|
|
|
|
|
|
const int gh_mv = il ? 12 : 0;
|
|
|
|
const int gh_bk = il ? 0 : 4;
|
|
|
|
|
|
|
|
for (int i = 0; i < 8; i++) {
|
|
|
|
// extract the 5-th bits for x0 and x1
|
|
|
|
const uint8_t xh_0 = ((qh >> (gh_mv + 2*i )) << gh_bk) & 0x10;
|
|
|
|
const uint8_t xh_1 = ((qh >> (gh_mv + 2*i+1)) << gh_bk) & 0x10;
|
|
|
|
|
|
|
|
// combine the 4-bits from qs with the 5th bit
|
|
|
|
const int32_t x0 = ((((qs[i] ) & mask) >> x_mv) | xh_0);
|
|
|
|
const int32_t x1 = ((((qs[i] >> 8) & mask) >> x_mv) | xh_1);
|
|
|
|
|
|
|
|
reg[i/2][2*(i%2)+0] = d * x0 + md;
|
|
|
|
reg[i/2][2*(i%2)+1] = d * x1 + md;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
template <typename type4x4>
|
|
|
|
void dequantize_q5_1(device const block_q5_1 *xb, short il, thread type4x4 & reg) {
|
|
|
|
device const uint16_t * qs = ((device const uint16_t *)xb + 4);
|
|
|
|
const float d = xb->d;
|
|
|
|
const float m = xb->m;
|
|
|
|
const ushort mask = il ? 0x00F0 : 0x000F;
|
|
|
|
|
|
|
|
const uint32_t qh = *((device const uint32_t *)xb->qh);
|
|
|
|
|
|
|
|
const int x_mv = il ? 4 : 0;
|
|
|
|
|
|
|
|
const int gh_mv = il ? 12 : 0;
|
|
|
|
const int gh_bk = il ? 0 : 4;
|
|
|
|
|
|
|
|
for (int i = 0; i < 8; i++) {
|
|
|
|
// extract the 5-th bits for x0 and x1
|
|
|
|
const uint8_t xh_0 = ((qh >> (gh_mv + 2*i )) << gh_bk) & 0x10;
|
|
|
|
const uint8_t xh_1 = ((qh >> (gh_mv + 2*i+1)) << gh_bk) & 0x10;
|
|
|
|
|
|
|
|
// combine the 4-bits from qs with the 5th bit
|
|
|
|
const int32_t x0 = ((((qs[i] ) & mask) >> x_mv) | xh_0);
|
|
|
|
const int32_t x1 = ((((qs[i] >> 8) & mask) >> x_mv) | xh_1);
|
|
|
|
|
|
|
|
reg[i/2][2*(i%2)+0] = d * x0 + m;
|
|
|
|
reg[i/2][2*(i%2)+1] = d * x1 + m;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2023-08-24 15:19:57 +02:00
|
|
|
template <typename type4x4>
|
|
|
|
void dequantize_q8_0(device const block_q8_0 *xb, short il, thread type4x4 & reg) {
|
|
|
|
device const int8_t * qs = ((device const int8_t *)xb->qs);
|
|
|
|
const half d = xb->d;
|
|
|
|
|
2024-01-02 09:57:44 +01:00
|
|
|
for (int i = 0; i < 16; i++) {
|
2023-08-24 15:19:57 +02:00
|
|
|
reg[i/4][i%4] = (qs[i + 16*il] * d);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2023-08-16 22:07:04 +02:00
|
|
|
template <typename type4x4>
|
|
|
|
void dequantize_q2_K(device const block_q2_K *xb, short il, thread type4x4 & reg) {
|
2023-12-13 20:54:54 +01:00
|
|
|
const float d = xb->d;
|
|
|
|
const float min = xb->dmin;
|
2023-08-16 22:07:04 +02:00
|
|
|
device const uint8_t * q = (device const uint8_t *)xb->qs;
|
2023-12-13 20:54:54 +01:00
|
|
|
float dl, ml;
|
2023-08-16 22:07:04 +02:00
|
|
|
uint8_t sc = xb->scales[il];
|
|
|
|
|
|
|
|
#if QK_K == 256
|
|
|
|
q = q + 32*(il/8) + 16*(il&1);
|
|
|
|
il = (il/2)%4;
|
|
|
|
#endif
|
|
|
|
half coef = il>1 ? (il>2 ? 1/64.h : 1/16.h) : (il>0 ? 1/4.h : 1.h);
|
|
|
|
uchar mask = il>1 ? (il>2 ? 192 : 48) : (il>0 ? 12 : 3);
|
|
|
|
dl = d * (sc & 0xF) * coef, ml = min * (sc >> 4);
|
|
|
|
for (int i = 0; i < 16; ++i) {
|
|
|
|
reg[i/4][i%4] = dl * (q[i] & mask) - ml;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
template <typename type4x4>
|
|
|
|
void dequantize_q3_K(device const block_q3_K *xb, short il, thread type4x4 & reg) {
|
2023-09-11 09:30:11 +02:00
|
|
|
const half d_all = xb->d;
|
2023-08-16 22:07:04 +02:00
|
|
|
device const uint8_t * q = (device const uint8_t *)xb->qs;
|
|
|
|
device const uint8_t * h = (device const uint8_t *)xb->hmask;
|
|
|
|
device const int8_t * scales = (device const int8_t *)xb->scales;
|
|
|
|
|
|
|
|
#if QK_K == 256
|
|
|
|
q = q + 32 * (il/8) + 16 * (il&1);
|
|
|
|
h = h + 16 * (il&1);
|
|
|
|
uint8_t m = 1 << (il/2);
|
|
|
|
uint16_t kmask1 = (il/4)>1 ? ((il/4)>2 ? 192 : 48) : \
|
|
|
|
((il/4)>0 ? 12 : 3);
|
|
|
|
uint16_t kmask2 = il/8 ? 0xF0 : 0x0F;
|
|
|
|
uint16_t scale_2 = scales[il%8], scale_1 = scales[8 + il%4];
|
2023-09-11 09:30:11 +02:00
|
|
|
int16_t dl_int = (il/4)&1 ? (scale_2&kmask2) | ((scale_1&kmask1) << 2)
|
|
|
|
: (scale_2&kmask2) | ((scale_1&kmask1) << 4);
|
2024-01-09 18:37:08 +01:00
|
|
|
float dl = il<8 ? d_all * (dl_int - 32.f) : d_all * (dl_int / 16.f - 32.f);
|
|
|
|
const float ml = 4.f * dl;
|
2023-08-16 22:07:04 +02:00
|
|
|
|
2023-09-11 09:30:11 +02:00
|
|
|
il = (il/2) & 3;
|
|
|
|
const half coef = il>1 ? (il>2 ? 1/64.h : 1/16.h) : (il>0 ? 1/4.h : 1.h);
|
|
|
|
const uint8_t mask = il>1 ? (il>2 ? 192 : 48) : (il>0 ? 12 : 3);
|
|
|
|
dl *= coef;
|
2023-08-16 22:07:04 +02:00
|
|
|
|
|
|
|
for (int i = 0; i < 16; ++i) {
|
2023-09-11 09:30:11 +02:00
|
|
|
reg[i/4][i%4] = dl * (q[i] & mask) - (h[i] & m ? 0 : ml);
|
2023-08-16 22:07:04 +02:00
|
|
|
}
|
|
|
|
#else
|
|
|
|
float kcoef = il&1 ? 1.f/16.f : 1.f;
|
|
|
|
uint16_t kmask = il&1 ? 0xF0 : 0x0F;
|
|
|
|
float dl = d_all * ((scales[il/2] & kmask) * kcoef - 8);
|
|
|
|
float coef = il>1 ? (il>2 ? 1/64.h : 1/16.h) : (il>0 ? 1/4.h : 1.h);
|
|
|
|
uint8_t mask = il>1 ? (il>2 ? 192 : 48) : (il>0 ? 12 : 3);
|
|
|
|
uint8_t m = 1<<(il*2);
|
|
|
|
for (int i = 0; i < 16; ++i) {
|
|
|
|
reg[i/4][i%4] = coef * dl * ((q[i] & mask) - ((h[i%8] & (m * (1 + i/8))) ? 0 : 4.f/coef));
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
}
|
|
|
|
|
2023-09-11 09:30:11 +02:00
|
|
|
static inline uchar2 get_scale_min_k4_just2(int j, int k, device const uchar * q) {
|
|
|
|
return j < 4 ? uchar2{uchar(q[j+0+k] & 63), uchar(q[j+4+k] & 63)}
|
|
|
|
: uchar2{uchar((q[j+4+k] & 0xF) | ((q[j-4+k] & 0xc0) >> 2)), uchar((q[j+4+k] >> 4) | ((q[j-0+k] & 0xc0) >> 2))};
|
|
|
|
}
|
|
|
|
|
2023-08-16 22:07:04 +02:00
|
|
|
template <typename type4x4>
|
|
|
|
void dequantize_q4_K(device const block_q4_K *xb, short il, thread type4x4 & reg) {
|
2023-09-11 09:30:11 +02:00
|
|
|
device const uchar * q = xb->qs;
|
2023-08-16 22:07:04 +02:00
|
|
|
|
|
|
|
#if QK_K == 256
|
|
|
|
short is = (il/4) * 2;
|
|
|
|
q = q + (il/4) * 32 + 16 * (il&1);
|
2023-09-11 09:30:11 +02:00
|
|
|
il = il & 3;
|
|
|
|
const uchar2 sc = get_scale_min_k4_just2(is, il/2, xb->scales);
|
2023-12-13 20:54:54 +01:00
|
|
|
const float d = il < 2 ? xb->d : xb->d / 16.h;
|
|
|
|
const float min = xb->dmin;
|
|
|
|
const float dl = d * sc[0];
|
|
|
|
const float ml = min * sc[1];
|
2023-08-16 22:07:04 +02:00
|
|
|
#else
|
|
|
|
q = q + 16 * (il&1);
|
|
|
|
device const uint8_t * s = xb->scales;
|
|
|
|
device const half2 * dh = (device const half2 *)xb->d;
|
|
|
|
const float2 d = (float2)dh[0];
|
|
|
|
const float dl = il<2 ? d[0] * (s[0]&0xF) : d[0] * (s[1]&0xF)/16.h;
|
2023-09-11 09:30:11 +02:00
|
|
|
const float ml = il<2 ? d[1] * (s[0]>>4) : d[1] * (s[1]>>4);
|
2023-08-16 22:07:04 +02:00
|
|
|
#endif
|
|
|
|
const ushort mask = il<2 ? 0x0F : 0xF0;
|
|
|
|
for (int i = 0; i < 16; ++i) {
|
|
|
|
reg[i/4][i%4] = dl * (q[i] & mask) - ml;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
template <typename type4x4>
|
|
|
|
void dequantize_q5_K(device const block_q5_K *xb, short il, thread type4x4 & reg) {
|
|
|
|
device const uint8_t * q = xb->qs;
|
|
|
|
device const uint8_t * qh = xb->qh;
|
|
|
|
|
|
|
|
#if QK_K == 256
|
|
|
|
short is = (il/4) * 2;
|
|
|
|
q = q + 32 * (il/4) + 16 * (il&1);
|
|
|
|
qh = qh + 16 * (il&1);
|
|
|
|
uint8_t ul = 1 << (il/2);
|
2023-09-11 09:30:11 +02:00
|
|
|
il = il & 3;
|
|
|
|
const uchar2 sc = get_scale_min_k4_just2(is, il/2, xb->scales);
|
2024-01-09 18:37:08 +01:00
|
|
|
const float d = il < 2 ? xb->d : xb->d / 16.f;
|
2023-12-13 20:54:54 +01:00
|
|
|
const float min = xb->dmin;
|
|
|
|
const float dl = d * sc[0];
|
|
|
|
const float ml = min * sc[1];
|
2023-08-16 22:07:04 +02:00
|
|
|
|
2023-12-13 20:54:54 +01:00
|
|
|
const ushort mask = il<2 ? 0x0F : 0xF0;
|
|
|
|
const float qh_val = il<2 ? 16.f : 256.f;
|
2023-08-16 22:07:04 +02:00
|
|
|
for (int i = 0; i < 16; ++i) {
|
|
|
|
reg[i/4][i%4] = dl * ((q[i] & mask) + (qh[i] & ul ? qh_val : 0)) - ml;
|
|
|
|
}
|
|
|
|
#else
|
|
|
|
q = q + 16 * (il&1);
|
|
|
|
device const int8_t * s = xb->scales;
|
|
|
|
const float dl = xb->d * s[il];
|
|
|
|
uint8_t m = 1<<(il*2);
|
|
|
|
const float coef = il<2 ? 1.f : 1.f/16.f;
|
|
|
|
const ushort mask = il<2 ? 0x0F : 0xF0;
|
|
|
|
for (int i = 0; i < 16; ++i) {
|
|
|
|
reg[i/4][i%4] = coef * dl * ((q[i] & mask) - (qh[i%8] & (m*(1+i/8)) ? 0.f : 16.f/coef));
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
}
|
|
|
|
|
|
|
|
template <typename type4x4>
|
|
|
|
void dequantize_q6_K(device const block_q6_K *xb, short il, thread type4x4 & reg) {
|
2023-09-11 09:30:11 +02:00
|
|
|
const half d_all = xb->d;
|
2023-08-16 22:07:04 +02:00
|
|
|
device const uint8_t * ql = (device const uint8_t *)xb->ql;
|
|
|
|
device const uint8_t * qh = (device const uint8_t *)xb->qh;
|
|
|
|
device const int8_t * scales = (device const int8_t *)xb->scales;
|
|
|
|
|
|
|
|
#if QK_K == 256
|
|
|
|
ql = ql + 64*(il/8) + 32*((il/2)&1) + 16*(il&1);
|
|
|
|
qh = qh + 32*(il/8) + 16*(il&1);
|
2024-01-09 18:37:08 +01:00
|
|
|
float sc = scales[(il%2) + 2 * ((il/2))];
|
2023-09-11 09:30:11 +02:00
|
|
|
il = (il/2) & 3;
|
2023-08-16 22:07:04 +02:00
|
|
|
#else
|
|
|
|
ql = ql + 16 * (il&1);
|
2024-01-09 18:37:08 +01:00
|
|
|
float sc = scales[il];
|
2023-08-16 22:07:04 +02:00
|
|
|
#endif
|
2023-09-11 09:30:11 +02:00
|
|
|
const uint16_t kmask1 = il>1 ? (il>2 ? 192 : 48) : (il>0 ? 12 : 3);
|
|
|
|
const uint16_t kmask2 = il>1 ? 0xF0 : 0x0F;
|
2024-01-09 18:37:08 +01:00
|
|
|
const float coef = il>1 ? 1.f/16.f : 1.f;
|
|
|
|
const float ml = d_all * sc * 32.f;
|
|
|
|
const float dl = d_all * sc * coef;
|
2023-08-16 22:07:04 +02:00
|
|
|
for (int i = 0; i < 16; ++i) {
|
2023-09-11 09:30:11 +02:00
|
|
|
const half q = il&1 ? ((ql[i] & kmask2) | ((qh[i] & kmask1) << 2))
|
|
|
|
: ((ql[i] & kmask2) | ((qh[i] & kmask1) << 4));
|
|
|
|
reg[i/4][i%4] = dl * q - ml;
|
2023-08-16 22:07:04 +02:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2024-01-08 16:02:32 +01:00
|
|
|
template <typename type4x4>
|
|
|
|
void dequantize_iq2_xxs(device const block_iq2_xxs * xb, short il, thread type4x4 & reg) {
|
|
|
|
// il is 0...15 for QK_K = 256 => index of block of 32 is il/2
|
|
|
|
const float d = xb->d;
|
|
|
|
const int ib32 = il/2;
|
|
|
|
il = il%2;
|
|
|
|
// il = 0 or 1. il = 0 processes the first 16 quants in a block of 32, il = 1 the second 16
|
|
|
|
// each block of 32 needs 2 uint32_t's for the quants & scale, so 4 uint16_t's.
|
|
|
|
device const uint16_t * q2 = xb->qs + 4*ib32;
|
|
|
|
const uint32_t aux32_g = q2[0] | (q2[1] << 16);
|
|
|
|
const uint32_t aux32_s = q2[2] | (q2[3] << 16);
|
|
|
|
thread const uint8_t * aux8 = (thread const uint8_t *)&aux32_g;
|
|
|
|
const float dl = d * (0.5f + (aux32_s >> 28)) * 0.25f;
|
2024-01-11 20:39:39 +01:00
|
|
|
constant uint8_t * grid = (constant uint8_t *)(iq2xxs_grid + aux8[2*il+0]);
|
2024-01-08 16:02:32 +01:00
|
|
|
uint8_t signs = ksigns_iq2xs[(aux32_s >> 14*il) & 127];
|
|
|
|
for (int i = 0; i < 8; ++i) {
|
|
|
|
reg[i/4][i%4] = dl * grid[i] * (signs & kmask_iq2xs[i] ? -1.f : 1.f);
|
|
|
|
}
|
2024-01-11 20:39:39 +01:00
|
|
|
grid = (constant uint8_t *)(iq2xxs_grid + aux8[2*il+1]);
|
2024-01-08 16:02:32 +01:00
|
|
|
signs = ksigns_iq2xs[(aux32_s >> (14*il+7)) & 127];
|
|
|
|
for (int i = 0; i < 8; ++i) {
|
|
|
|
reg[2+i/4][i%4] = dl * grid[i] * (signs & kmask_iq2xs[i] ? -1.f : 1.f);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2024-01-11 20:39:39 +01:00
|
|
|
template <typename type4x4>
|
|
|
|
void dequantize_iq2_xs(device const block_iq2_xs * xb, short il, thread type4x4 & reg) {
|
|
|
|
// il is 0...15 for QK_K = 256 => index of block of 32 is il/2
|
|
|
|
const float d = xb->d;
|
|
|
|
const int ib32 = il/2;
|
|
|
|
il = il%2;
|
|
|
|
// il = 0 or 1. il = 0 processes the first 16 quants in a block of 32, il = 1 the second 16
|
|
|
|
device const uint16_t * q2 = xb->qs + 4*ib32;
|
|
|
|
const float dl = d * (0.5f + ((xb->scales[ib32] >> 4*il) & 0xf)) * 0.25f;
|
|
|
|
constant uint8_t * grid = (constant uint8_t *)(iq2xs_grid + (q2[2*il+0] & 511));
|
|
|
|
uint8_t signs = ksigns_iq2xs[q2[2*il+0] >> 9];
|
|
|
|
for (int i = 0; i < 8; ++i) {
|
|
|
|
reg[i/4][i%4] = dl * grid[i] * (signs & kmask_iq2xs[i] ? -1.f : 1.f);
|
|
|
|
}
|
|
|
|
grid = (constant uint8_t *)(iq2xs_grid + (q2[2*il+1] & 511));
|
|
|
|
signs = ksigns_iq2xs[q2[2*il+1] >> 9];
|
|
|
|
for (int i = 0; i < 8; ++i) {
|
|
|
|
reg[2+i/4][i%4] = dl * grid[i] * (signs & kmask_iq2xs[i] ? -1.f : 1.f);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2024-01-30 14:14:12 +01:00
|
|
|
template <typename type4x4>
|
|
|
|
void dequantize_iq3_xxs(device const block_iq3_xxs * xb, short il, thread type4x4 & reg) {
|
|
|
|
// il is 0...15 for QK_K = 256 => index of block of 32 is il/2
|
|
|
|
const float d = xb->d;
|
|
|
|
const int ib32 = il/2;
|
|
|
|
il = il%2;
|
|
|
|
// il = 0 or 1. il = 0 processes the first 16 quants in a block of 32, il = 1 the second 16
|
|
|
|
device const uint8_t * q3 = xb->qs + 8*ib32;
|
|
|
|
device const uint16_t * gas = (device const uint16_t *)(xb->qs + QK_K/4) + 2*ib32;
|
|
|
|
const uint32_t aux32 = gas[0] | (gas[1] << 16);
|
|
|
|
const float dl = d * (0.5f + (aux32 >> 28)) * 0.5f;
|
|
|
|
constant uint8_t * grid1 = (constant uint8_t *)(iq3xxs_grid + q3[4*il+0]);
|
|
|
|
constant uint8_t * grid2 = (constant uint8_t *)(iq3xxs_grid + q3[4*il+1]);
|
|
|
|
uint8_t signs = ksigns_iq2xs[(aux32 >> 14*il) & 127];
|
|
|
|
for (int i = 0; i < 4; ++i) {
|
|
|
|
reg[0][i] = dl * grid1[i] * (signs & kmask_iq2xs[i+0] ? -1.f : 1.f);
|
|
|
|
reg[1][i] = dl * grid2[i] * (signs & kmask_iq2xs[i+4] ? -1.f : 1.f);
|
|
|
|
}
|
|
|
|
grid1 = (constant uint8_t *)(iq3xxs_grid + q3[4*il+2]);
|
|
|
|
grid2 = (constant uint8_t *)(iq3xxs_grid + q3[4*il+3]);
|
|
|
|
signs = ksigns_iq2xs[(aux32 >> (14*il+7)) & 127];
|
|
|
|
for (int i = 0; i < 4; ++i) {
|
|
|
|
reg[2][i] = dl * grid1[i] * (signs & kmask_iq2xs[i+0] ? -1.f : 1.f);
|
|
|
|
reg[3][i] = dl * grid2[i] * (signs & kmask_iq2xs[i+4] ? -1.f : 1.f);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2023-08-16 22:07:04 +02:00
|
|
|
template<typename block_q, short nl, void (*dequantize_func)(device const block_q *, short, thread float4x4 &)>
|
|
|
|
kernel void kernel_get_rows(
|
|
|
|
device const void * src0,
|
2023-12-13 13:04:25 +01:00
|
|
|
device const char * src1,
|
2023-08-16 22:07:04 +02:00
|
|
|
device float * dst,
|
|
|
|
constant int64_t & ne00,
|
|
|
|
constant uint64_t & nb01,
|
2023-12-13 13:04:25 +01:00
|
|
|
constant uint64_t & nb02,
|
|
|
|
constant int64_t & ne10,
|
|
|
|
constant uint64_t & nb10,
|
|
|
|
constant uint64_t & nb11,
|
2023-08-16 22:07:04 +02:00
|
|
|
constant uint64_t & nb1,
|
2023-12-13 13:04:25 +01:00
|
|
|
constant uint64_t & nb2,
|
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
2023-08-16 22:07:04 +02:00
|
|
|
uint tiitg[[thread_index_in_threadgroup]],
|
2023-12-13 13:04:25 +01:00
|
|
|
uint3 tptg [[threads_per_threadgroup]]) {
|
|
|
|
//const int64_t i = tgpig;
|
|
|
|
//const int64_t r = ((device int32_t *) src1)[i];
|
2023-08-16 22:07:04 +02:00
|
|
|
|
2023-12-13 13:04:25 +01:00
|
|
|
const int64_t i10 = tgpig.x;
|
|
|
|
const int64_t i11 = tgpig.y;
|
|
|
|
|
|
|
|
const int64_t r = ((device int32_t *) ((device char *) src1 + i11*nb11 + i10*nb10))[0];
|
|
|
|
|
|
|
|
const int64_t i02 = i11;
|
|
|
|
|
|
|
|
for (int64_t ind = tiitg; ind < ne00/16; ind += tptg.x) {
|
2023-08-16 22:07:04 +02:00
|
|
|
float4x4 temp;
|
|
|
|
dequantize_func(
|
2023-12-13 13:04:25 +01:00
|
|
|
((device const block_q *) ((device char *) src0 + r*nb01 + i02*nb02)) + ind/nl, ind%nl, temp);
|
|
|
|
*(((device float4x4 *) ((device char *) dst + i11*nb2 + i10*nb1)) + ind) = temp;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
kernel void kernel_get_rows_f32(
|
|
|
|
device const void * src0,
|
|
|
|
device const char * src1,
|
|
|
|
device float * dst,
|
|
|
|
constant int64_t & ne00,
|
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant uint64_t & nb02,
|
|
|
|
constant int64_t & ne10,
|
|
|
|
constant uint64_t & nb10,
|
|
|
|
constant uint64_t & nb11,
|
|
|
|
constant uint64_t & nb1,
|
|
|
|
constant uint64_t & nb2,
|
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint tiitg[[thread_index_in_threadgroup]],
|
|
|
|
uint3 tptg [[threads_per_threadgroup]]) {
|
|
|
|
const int64_t i10 = tgpig.x;
|
|
|
|
const int64_t i11 = tgpig.y;
|
|
|
|
|
|
|
|
const int64_t r = ((device int32_t *) ((device char *) src1 + i11*nb11 + i10*nb10))[0];
|
|
|
|
|
|
|
|
const int64_t i02 = i11;
|
|
|
|
|
|
|
|
for (int ind = tiitg; ind < ne00; ind += tptg.x) {
|
|
|
|
((device float *) ((device char *) dst + i11*nb2 + i10*nb1))[ind] =
|
|
|
|
((device float *) ((device char *) src0 + r*nb01 + i02*nb02))[ind];
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
kernel void kernel_get_rows_f16(
|
|
|
|
device const void * src0,
|
|
|
|
device const char * src1,
|
|
|
|
device float * dst,
|
|
|
|
constant int64_t & ne00,
|
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant uint64_t & nb02,
|
|
|
|
constant int64_t & ne10,
|
|
|
|
constant uint64_t & nb10,
|
|
|
|
constant uint64_t & nb11,
|
|
|
|
constant uint64_t & nb1,
|
|
|
|
constant uint64_t & nb2,
|
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint tiitg[[thread_index_in_threadgroup]],
|
|
|
|
uint3 tptg [[threads_per_threadgroup]]) {
|
|
|
|
const int64_t i10 = tgpig.x;
|
|
|
|
const int64_t i11 = tgpig.y;
|
|
|
|
|
|
|
|
const int64_t r = ((device int32_t *) ((device char *) src1 + i11*nb11 + i10*nb10))[0];
|
|
|
|
|
|
|
|
const int64_t i02 = i11;
|
|
|
|
|
|
|
|
for (int ind = tiitg; ind < ne00; ind += tptg.x) {
|
|
|
|
((device float *) ((device char *) dst + i11*nb2 + i10*nb1))[ind] =
|
|
|
|
((device half *) ((device char *) src0 + r*nb01 + i02*nb02))[ind];
|
2023-08-16 22:07:04 +02:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2024-01-03 10:35:46 +01:00
|
|
|
kernel void kernel_get_rows_i32(
|
|
|
|
device const void * src0,
|
|
|
|
device const char * src1,
|
|
|
|
device int32_t * dst,
|
|
|
|
constant int64_t & ne00,
|
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant uint64_t & nb02,
|
|
|
|
constant int64_t & ne10,
|
|
|
|
constant uint64_t & nb10,
|
|
|
|
constant uint64_t & nb11,
|
|
|
|
constant uint64_t & nb1,
|
|
|
|
constant uint64_t & nb2,
|
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint tiitg[[thread_index_in_threadgroup]],
|
|
|
|
uint3 tptg [[threads_per_threadgroup]]) {
|
|
|
|
const int64_t i10 = tgpig.x;
|
|
|
|
const int64_t i11 = tgpig.y;
|
|
|
|
|
|
|
|
const int64_t r = ((device int32_t *) ((device char *) src1 + i11*nb11 + i10*nb10))[0];
|
|
|
|
|
|
|
|
const int64_t i02 = i11;
|
|
|
|
|
|
|
|
for (int ind = tiitg; ind < ne00; ind += tptg.x) {
|
|
|
|
((device int32_t *) ((device char *) dst + i11*nb2 + i10*nb1))[ind] =
|
|
|
|
((device int32_t *) ((device char *) src0 + r*nb01 + i02*nb02))[ind];
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
|
2023-08-16 22:07:04 +02:00
|
|
|
#define BLOCK_SIZE_M 64 // 8 simdgroup matrices from matrix A
|
2023-10-08 09:01:53 +02:00
|
|
|
#define BLOCK_SIZE_N 32 // 4 simdgroup matrices from matrix B
|
2023-08-16 22:07:04 +02:00
|
|
|
#define BLOCK_SIZE_K 32
|
|
|
|
#define THREAD_MAT_M 4 // each thread take 4 simdgroup matrices from matrix A
|
|
|
|
#define THREAD_MAT_N 2 // each thread take 2 simdgroup matrices from matrix B
|
|
|
|
#define THREAD_PER_BLOCK 128
|
|
|
|
#define THREAD_PER_ROW 2 // 2 thread for each row in matrix A to load numbers
|
|
|
|
#define THREAD_PER_COL 4 // 4 thread for each row in matrix B to load numbers
|
|
|
|
#define SG_MAT_SIZE 64 // simdgroup matrix is of shape 8x8
|
|
|
|
#define SG_MAT_ROW 8
|
|
|
|
|
|
|
|
// each block_q contains 16*nl weights
|
|
|
|
template<typename block_q, short nl, void (*dequantize_func)(device const block_q *, short, thread half4x4 &)>
|
2023-12-07 21:26:54 +01:00
|
|
|
void kernel_mul_mm_impl(device const uchar * src0,
|
|
|
|
device const uchar * src1,
|
|
|
|
device float * dst,
|
|
|
|
constant int64_t & ne00,
|
|
|
|
constant int64_t & ne02,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant uint64_t & nb02,
|
2023-12-07 21:26:54 +01:00
|
|
|
constant int64_t & ne12,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant uint64_t & nb10,
|
|
|
|
constant uint64_t & nb11,
|
|
|
|
constant uint64_t & nb12,
|
2023-12-07 21:26:54 +01:00
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
|
|
|
constant uint & r2,
|
|
|
|
constant uint & r3,
|
|
|
|
threadgroup uchar * shared_memory [[threadgroup(0)]],
|
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint tiitg[[thread_index_in_threadgroup]],
|
|
|
|
uint sgitg[[simdgroup_index_in_threadgroup]]) {
|
2023-09-15 10:09:24 +02:00
|
|
|
|
|
|
|
threadgroup half * sa = (threadgroup half *)(shared_memory);
|
2023-08-16 22:07:04 +02:00
|
|
|
threadgroup float * sb = (threadgroup float *)(shared_memory + 4096);
|
|
|
|
|
|
|
|
const uint r0 = tgpig.y;
|
|
|
|
const uint r1 = tgpig.x;
|
|
|
|
const uint im = tgpig.z;
|
2023-10-08 09:01:53 +02:00
|
|
|
|
2023-08-16 22:07:04 +02:00
|
|
|
// if this block is of 64x32 shape or smaller
|
|
|
|
short n_rows = (ne0 - r0 * BLOCK_SIZE_M < BLOCK_SIZE_M) ? (ne0 - r0 * BLOCK_SIZE_M) : BLOCK_SIZE_M;
|
|
|
|
short n_cols = (ne1 - r1 * BLOCK_SIZE_N < BLOCK_SIZE_N) ? (ne1 - r1 * BLOCK_SIZE_N) : BLOCK_SIZE_N;
|
2023-10-08 09:01:53 +02:00
|
|
|
|
2023-08-16 22:07:04 +02:00
|
|
|
// a thread shouldn't load data outside of the matrix
|
|
|
|
short thread_row = ((short)tiitg/THREAD_PER_ROW) < n_rows ? ((short)tiitg/THREAD_PER_ROW) : n_rows - 1;
|
|
|
|
short thread_col = ((short)tiitg/THREAD_PER_COL) < n_cols ? ((short)tiitg/THREAD_PER_COL) : n_cols - 1;
|
|
|
|
|
2023-09-15 10:09:24 +02:00
|
|
|
simdgroup_half8x8 ma[4];
|
2023-08-16 22:07:04 +02:00
|
|
|
simdgroup_float8x8 mb[2];
|
|
|
|
simdgroup_float8x8 c_res[8];
|
|
|
|
for (int i = 0; i < 8; i++){
|
|
|
|
c_res[i] = make_filled_simdgroup_matrix<float, 8>(0.f);
|
|
|
|
}
|
|
|
|
|
|
|
|
short il = (tiitg % THREAD_PER_ROW);
|
2023-09-15 10:09:24 +02:00
|
|
|
|
2023-12-07 21:26:54 +01:00
|
|
|
const uint i12 = im%ne12;
|
|
|
|
const uint i13 = im/ne12;
|
|
|
|
|
|
|
|
uint offset0 = (i12/r2)*nb02 + (i13/r3)*(nb02*ne02);
|
2023-09-15 10:09:24 +02:00
|
|
|
ushort offset1 = il/nl;
|
|
|
|
|
|
|
|
device const block_q * x = (device const block_q *)(src0 + (r0 * BLOCK_SIZE_M + thread_row) * nb01 + offset0) + offset1;
|
|
|
|
device const float * y = (device const float *)(src1
|
|
|
|
+ nb12 * im
|
|
|
|
+ nb11 * (r1 * BLOCK_SIZE_N + thread_col)
|
|
|
|
+ nb10 * (BLOCK_SIZE_K / THREAD_PER_COL * (tiitg % THREAD_PER_COL)));
|
2023-08-16 22:07:04 +02:00
|
|
|
|
|
|
|
for (int loop_k = 0; loop_k < ne00; loop_k += BLOCK_SIZE_K) {
|
2023-10-08 09:01:53 +02:00
|
|
|
// load data and store to threadgroup memory
|
2023-08-16 22:07:04 +02:00
|
|
|
half4x4 temp_a;
|
|
|
|
dequantize_func(x, il, temp_a);
|
2023-08-22 08:18:40 +02:00
|
|
|
threadgroup_barrier(mem_flags::mem_threadgroup);
|
2023-10-08 09:01:53 +02:00
|
|
|
|
2023-08-16 22:07:04 +02:00
|
|
|
#pragma unroll(16)
|
|
|
|
for (int i = 0; i < 16; i++) {
|
|
|
|
*(sa + SG_MAT_SIZE * ((tiitg / THREAD_PER_ROW / 8) \
|
2023-10-08 09:01:53 +02:00
|
|
|
+ (tiitg % THREAD_PER_ROW) * 16 + (i / 8) * 8) \
|
|
|
|
+ (tiitg / THREAD_PER_ROW) % 8 + (i & 7) * 8) = temp_a[i/4][i%4];
|
2023-08-16 22:07:04 +02:00
|
|
|
}
|
2023-10-08 09:01:53 +02:00
|
|
|
|
|
|
|
*(threadgroup float2x4 *)(sb + (tiitg % THREAD_PER_COL) * 8 * 32 + 8 * (tiitg / THREAD_PER_COL)) = *((device float2x4 *)y);
|
|
|
|
|
2023-08-16 22:07:04 +02:00
|
|
|
il = (il + 2 < nl) ? il + 2 : il % 2;
|
|
|
|
x = (il < 2) ? x + (2+nl-1)/nl : x;
|
|
|
|
y += BLOCK_SIZE_K;
|
|
|
|
|
|
|
|
threadgroup_barrier(mem_flags::mem_threadgroup);
|
2023-10-08 09:01:53 +02:00
|
|
|
|
|
|
|
// load matrices from threadgroup memory and conduct outer products
|
2023-08-16 22:07:04 +02:00
|
|
|
threadgroup half * lsma = (sa + THREAD_MAT_M * SG_MAT_SIZE * (sgitg % 2));
|
|
|
|
threadgroup float * lsmb = (sb + THREAD_MAT_N * SG_MAT_SIZE * (sgitg / 2));
|
2023-10-08 09:01:53 +02:00
|
|
|
|
2023-08-16 22:07:04 +02:00
|
|
|
#pragma unroll(4)
|
|
|
|
for (int ik = 0; ik < BLOCK_SIZE_K / 8; ik++) {
|
|
|
|
#pragma unroll(4)
|
|
|
|
for (int i = 0; i < 4; i++) {
|
|
|
|
simdgroup_load(ma[i],lsma + SG_MAT_SIZE * i);
|
|
|
|
}
|
|
|
|
simdgroup_barrier(mem_flags::mem_none);
|
|
|
|
#pragma unroll(2)
|
|
|
|
for (int i = 0; i < 2; i++) {
|
|
|
|
simdgroup_load(mb[i],lsmb + SG_MAT_SIZE * i);
|
|
|
|
}
|
|
|
|
|
|
|
|
lsma += BLOCK_SIZE_M / SG_MAT_ROW * SG_MAT_SIZE;
|
|
|
|
lsmb += BLOCK_SIZE_N / SG_MAT_ROW * SG_MAT_SIZE;
|
2023-10-08 09:01:53 +02:00
|
|
|
|
2023-08-16 22:07:04 +02:00
|
|
|
#pragma unroll(8)
|
|
|
|
for (int i = 0; i < 8; i++){
|
|
|
|
simdgroup_multiply_accumulate(c_res[i], mb[i/4], ma[i%4], c_res[i]);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
if ((r0 + 1) * BLOCK_SIZE_M <= ne0 && (r1 + 1) * BLOCK_SIZE_N <= ne1) {
|
2023-10-08 09:01:53 +02:00
|
|
|
device float * C = dst + (BLOCK_SIZE_M * r0 + 32 * (sgitg & 1)) \
|
|
|
|
+ (BLOCK_SIZE_N * r1 + 16 * (sgitg >> 1)) * ne0 + im*ne1*ne0;
|
2023-08-16 22:07:04 +02:00
|
|
|
for (int i = 0; i < 8; i++) {
|
|
|
|
simdgroup_store(c_res[i], C + 8 * (i%4) + 8 * ne0 * (i/4), ne0);
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
// block is smaller than 64x32, we should avoid writing data outside of the matrix
|
2023-08-22 08:18:40 +02:00
|
|
|
threadgroup_barrier(mem_flags::mem_threadgroup);
|
2023-10-08 09:01:53 +02:00
|
|
|
threadgroup float * temp_str = ((threadgroup float *)shared_memory) \
|
2023-08-16 22:07:04 +02:00
|
|
|
+ 32 * (sgitg&1) + (16 * (sgitg>>1)) * BLOCK_SIZE_M;
|
|
|
|
for (int i = 0; i < 8; i++) {
|
|
|
|
simdgroup_store(c_res[i], temp_str + 8 * (i%4) + 8 * BLOCK_SIZE_M * (i/4), BLOCK_SIZE_M);
|
|
|
|
}
|
|
|
|
|
2023-08-22 08:18:40 +02:00
|
|
|
threadgroup_barrier(mem_flags::mem_threadgroup);
|
2023-10-08 09:01:53 +02:00
|
|
|
|
|
|
|
device float * C = dst + (BLOCK_SIZE_M * r0) + (BLOCK_SIZE_N * r1) * ne0 + im*ne1*ne0;
|
|
|
|
if (sgitg == 0) {
|
2023-08-16 22:07:04 +02:00
|
|
|
for (int i = 0; i < n_rows; i++) {
|
2023-10-08 09:01:53 +02:00
|
|
|
for (int j = tiitg; j < n_cols; j += BLOCK_SIZE_N) {
|
2023-08-16 22:07:04 +02:00
|
|
|
*(C + i + j * ne0) = *(temp_str + i + j * BLOCK_SIZE_M);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
2023-06-08 18:46:22 +02:00
|
|
|
}
|
|
|
|
}
|
2023-08-16 22:07:04 +02:00
|
|
|
|
2024-01-02 20:07:47 +01:00
|
|
|
// same as kernel_mul_mm_impl, but src1 and dst are accessed via indices stored in src1ids
|
|
|
|
template<typename block_q, short nl, void (*dequantize_func)(device const block_q *, short, thread half4x4 &)>
|
|
|
|
void kernel_mul_mm_id_impl(
|
|
|
|
device const uchar * src0,
|
|
|
|
device const uchar * src1,
|
|
|
|
thread short * src1ids,
|
|
|
|
device float * dst,
|
|
|
|
constant int64_t & ne00,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant uint64_t & nb02,
|
|
|
|
constant int64_t & ne12,
|
|
|
|
constant uint64_t & nb10,
|
|
|
|
constant uint64_t & nb11,
|
|
|
|
constant uint64_t & nb12,
|
|
|
|
constant int64_t & ne0,
|
|
|
|
int64_t ne1,
|
|
|
|
constant uint & r2,
|
|
|
|
constant uint & r3,
|
|
|
|
threadgroup uchar * shared_memory,
|
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint tiitg[[thread_index_in_threadgroup]],
|
|
|
|
uint sgitg[[simdgroup_index_in_threadgroup]]) {
|
|
|
|
|
|
|
|
threadgroup half * sa = (threadgroup half *)(shared_memory);
|
|
|
|
threadgroup float * sb = (threadgroup float *)(shared_memory + 4096);
|
|
|
|
|
|
|
|
const uint r0 = tgpig.y;
|
|
|
|
const uint r1 = tgpig.x;
|
|
|
|
const uint im = tgpig.z;
|
|
|
|
|
|
|
|
if (r1 * BLOCK_SIZE_N >= ne1) return;
|
|
|
|
|
|
|
|
// if this block is of 64x32 shape or smaller
|
|
|
|
short n_rows = (ne0 - r0 * BLOCK_SIZE_M < BLOCK_SIZE_M) ? (ne0 - r0 * BLOCK_SIZE_M) : BLOCK_SIZE_M;
|
|
|
|
short n_cols = (ne1 - r1 * BLOCK_SIZE_N < BLOCK_SIZE_N) ? (ne1 - r1 * BLOCK_SIZE_N) : BLOCK_SIZE_N;
|
|
|
|
|
|
|
|
// a thread shouldn't load data outside of the matrix
|
|
|
|
short thread_row = ((short)tiitg/THREAD_PER_ROW) < n_rows ? ((short)tiitg/THREAD_PER_ROW) : n_rows - 1;
|
|
|
|
short thread_col = ((short)tiitg/THREAD_PER_COL) < n_cols ? ((short)tiitg/THREAD_PER_COL) : n_cols - 1;
|
|
|
|
|
|
|
|
simdgroup_half8x8 ma[4];
|
|
|
|
simdgroup_float8x8 mb[2];
|
|
|
|
simdgroup_float8x8 c_res[8];
|
|
|
|
for (int i = 0; i < 8; i++){
|
|
|
|
c_res[i] = make_filled_simdgroup_matrix<float, 8>(0.f);
|
|
|
|
}
|
|
|
|
|
|
|
|
short il = (tiitg % THREAD_PER_ROW);
|
|
|
|
|
|
|
|
const uint i12 = im%ne12;
|
|
|
|
const uint i13 = im/ne12;
|
|
|
|
|
|
|
|
uint offset0 = (i12/r2)*nb02 + (i13/r3)*(nb02*ne02);
|
|
|
|
ushort offset1 = il/nl;
|
|
|
|
|
|
|
|
device const block_q * x = (device const block_q *)(src0 + (r0 * BLOCK_SIZE_M + thread_row) * nb01 + offset0) + offset1;
|
|
|
|
device const float * y = (device const float *)(src1
|
|
|
|
+ nb12 * im
|
|
|
|
+ nb11 * src1ids[r1 * BLOCK_SIZE_N + thread_col]
|
|
|
|
+ nb10 * (BLOCK_SIZE_K / THREAD_PER_COL * (tiitg % THREAD_PER_COL)));
|
|
|
|
|
|
|
|
for (int loop_k = 0; loop_k < ne00; loop_k += BLOCK_SIZE_K) {
|
|
|
|
// load data and store to threadgroup memory
|
|
|
|
half4x4 temp_a;
|
|
|
|
dequantize_func(x, il, temp_a);
|
|
|
|
threadgroup_barrier(mem_flags::mem_threadgroup);
|
|
|
|
|
|
|
|
for (int i = 0; i < 16; i++) {
|
|
|
|
*(sa + SG_MAT_SIZE * ((tiitg / THREAD_PER_ROW / 8) \
|
|
|
|
+ (tiitg % THREAD_PER_ROW) * 16 + (i / 8) * 8) \
|
|
|
|
+ (tiitg / THREAD_PER_ROW) % 8 + (i & 7) * 8) = temp_a[i/4][i%4];
|
|
|
|
}
|
|
|
|
|
|
|
|
*(threadgroup float2x4 *)(sb + (tiitg % THREAD_PER_COL) * 8 * 32 + 8 * (tiitg / THREAD_PER_COL)) = *((device float2x4 *)y);
|
|
|
|
|
|
|
|
il = (il + 2 < nl) ? il + 2 : il % 2;
|
|
|
|
x = (il < 2) ? x + (2+nl-1)/nl : x;
|
|
|
|
y += BLOCK_SIZE_K;
|
|
|
|
|
|
|
|
threadgroup_barrier(mem_flags::mem_threadgroup);
|
|
|
|
|
|
|
|
// load matrices from threadgroup memory and conduct outer products
|
|
|
|
threadgroup half * lsma = (sa + THREAD_MAT_M * SG_MAT_SIZE * (sgitg % 2));
|
|
|
|
threadgroup float * lsmb = (sb + THREAD_MAT_N * SG_MAT_SIZE * (sgitg / 2));
|
|
|
|
|
|
|
|
for (int ik = 0; ik < BLOCK_SIZE_K / 8; ik++) {
|
|
|
|
for (int i = 0; i < 4; i++) {
|
|
|
|
simdgroup_load(ma[i],lsma + SG_MAT_SIZE * i);
|
|
|
|
}
|
|
|
|
simdgroup_barrier(mem_flags::mem_none);
|
|
|
|
for (int i = 0; i < 2; i++) {
|
|
|
|
simdgroup_load(mb[i],lsmb + SG_MAT_SIZE * i);
|
|
|
|
}
|
|
|
|
|
|
|
|
lsma += BLOCK_SIZE_M / SG_MAT_ROW * SG_MAT_SIZE;
|
|
|
|
lsmb += BLOCK_SIZE_N / SG_MAT_ROW * SG_MAT_SIZE;
|
|
|
|
|
|
|
|
for (int i = 0; i < 8; i++){
|
|
|
|
simdgroup_multiply_accumulate(c_res[i], mb[i/4], ma[i%4], c_res[i]);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
{
|
|
|
|
threadgroup_barrier(mem_flags::mem_threadgroup);
|
|
|
|
threadgroup float * temp_str = ((threadgroup float *)shared_memory) \
|
|
|
|
+ 32 * (sgitg&1) + (16 * (sgitg>>1)) * BLOCK_SIZE_M;
|
|
|
|
for (int i = 0; i < 8; i++) {
|
|
|
|
simdgroup_store(c_res[i], temp_str + 8 * (i%4) + 8 * BLOCK_SIZE_M * (i/4), BLOCK_SIZE_M);
|
|
|
|
}
|
|
|
|
|
|
|
|
threadgroup_barrier(mem_flags::mem_threadgroup);
|
|
|
|
|
|
|
|
device float * C = dst + (BLOCK_SIZE_M * r0) + im*ne1*ne0;
|
|
|
|
if (sgitg == 0) {
|
|
|
|
for (int i = 0; i < n_rows; i++) {
|
|
|
|
for (int j = tiitg; j < n_cols; j += BLOCK_SIZE_N) {
|
|
|
|
*(C + i + src1ids[j + r1*BLOCK_SIZE_N] * ne0) = *(temp_str + i + j * BLOCK_SIZE_M);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2023-12-07 21:26:54 +01:00
|
|
|
template<typename block_q, short nl, void (*dequantize_func)(device const block_q *, short, thread half4x4 &)>
|
|
|
|
kernel void kernel_mul_mm(device const uchar * src0,
|
|
|
|
device const uchar * src1,
|
|
|
|
device float * dst,
|
|
|
|
constant int64_t & ne00,
|
|
|
|
constant int64_t & ne02,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant uint64_t & nb02,
|
2023-12-07 21:26:54 +01:00
|
|
|
constant int64_t & ne12,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant uint64_t & nb10,
|
|
|
|
constant uint64_t & nb11,
|
|
|
|
constant uint64_t & nb12,
|
2023-12-07 21:26:54 +01:00
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
|
|
|
constant uint & r2,
|
|
|
|
constant uint & r3,
|
|
|
|
threadgroup uchar * shared_memory [[threadgroup(0)]],
|
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint tiitg[[thread_index_in_threadgroup]],
|
|
|
|
uint sgitg[[simdgroup_index_in_threadgroup]]) {
|
|
|
|
kernel_mul_mm_impl<block_q, nl, dequantize_func>(
|
|
|
|
src0,
|
|
|
|
src1,
|
|
|
|
dst,
|
|
|
|
ne00,
|
|
|
|
ne02,
|
|
|
|
nb01,
|
|
|
|
nb02,
|
|
|
|
ne12,
|
|
|
|
nb10,
|
|
|
|
nb11,
|
|
|
|
nb12,
|
|
|
|
ne0,
|
|
|
|
ne1,
|
|
|
|
r2,
|
|
|
|
r3,
|
|
|
|
shared_memory,
|
|
|
|
tgpig,
|
|
|
|
tiitg,
|
|
|
|
sgitg);
|
|
|
|
}
|
|
|
|
|
|
|
|
template<typename block_q, short nl, void (*dequantize_func)(device const block_q *, short, thread half4x4 &)>
|
|
|
|
kernel void kernel_mul_mm_id(
|
2023-12-13 13:04:25 +01:00
|
|
|
device const uchar * ids,
|
2023-12-07 21:26:54 +01:00
|
|
|
device const uchar * src1,
|
2024-01-02 20:07:47 +01:00
|
|
|
device float * dst,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant uint64_t & nbi1,
|
2023-12-07 21:26:54 +01:00
|
|
|
constant int64_t & ne00,
|
|
|
|
constant int64_t & ne02,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant uint64_t & nb02,
|
2023-12-07 21:26:54 +01:00
|
|
|
constant int64_t & ne12,
|
2023-12-13 13:04:25 +01:00
|
|
|
constant int64_t & ne13,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant uint64_t & nb10,
|
|
|
|
constant uint64_t & nb11,
|
|
|
|
constant uint64_t & nb12,
|
2023-12-07 21:26:54 +01:00
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant uint64_t & nb1,
|
2023-12-07 21:26:54 +01:00
|
|
|
constant uint & r2,
|
|
|
|
constant uint & r3,
|
|
|
|
constant int & idx,
|
|
|
|
device const uchar * src00,
|
|
|
|
device const uchar * src01,
|
|
|
|
device const uchar * src02,
|
|
|
|
device const uchar * src03,
|
|
|
|
device const uchar * src04,
|
|
|
|
device const uchar * src05,
|
|
|
|
device const uchar * src06,
|
|
|
|
device const uchar * src07,
|
|
|
|
threadgroup uchar * shared_memory [[threadgroup(0)]],
|
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint tiitg[[thread_index_in_threadgroup]],
|
|
|
|
uint sgitg[[simdgroup_index_in_threadgroup]]) {
|
2024-01-02 20:07:47 +01:00
|
|
|
device const uchar * src0s[8] = {src00, src01, src02, src03, src04, src05, src06, src07};
|
2023-12-07 21:26:54 +01:00
|
|
|
|
2024-01-02 20:07:47 +01:00
|
|
|
// expert id
|
|
|
|
const int32_t id = tgpig.z/(ne12*ne13);
|
2023-12-13 13:04:25 +01:00
|
|
|
|
|
|
|
tgpig.z = tgpig.z%(ne12*ne13);
|
|
|
|
|
2024-01-02 20:07:47 +01:00
|
|
|
// row indices of src1 for expert id
|
|
|
|
int64_t _ne1 = 0;
|
|
|
|
short src1ids[512];
|
2023-12-13 13:04:25 +01:00
|
|
|
|
2024-01-02 20:07:47 +01:00
|
|
|
for (int64_t i1 = 0; i1 < ne1; i1++) {
|
|
|
|
if (((device int32_t *) (ids + i1*nbi1))[idx] == id) {
|
|
|
|
src1ids[_ne1++] = i1;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
kernel_mul_mm_id_impl<block_q, nl, dequantize_func>(
|
|
|
|
src0s[id],
|
|
|
|
src1,
|
|
|
|
src1ids,
|
|
|
|
dst,
|
2023-12-07 21:26:54 +01:00
|
|
|
ne00,
|
|
|
|
ne02,
|
|
|
|
nb01,
|
|
|
|
nb02,
|
|
|
|
ne12,
|
|
|
|
nb10,
|
|
|
|
nb11,
|
|
|
|
nb12,
|
|
|
|
ne0,
|
2024-01-02 20:07:47 +01:00
|
|
|
_ne1,
|
2023-12-07 21:26:54 +01:00
|
|
|
r2,
|
|
|
|
r3,
|
|
|
|
shared_memory,
|
|
|
|
tgpig,
|
|
|
|
tiitg,
|
|
|
|
sgitg);
|
|
|
|
}
|
|
|
|
|
2023-08-16 22:07:04 +02:00
|
|
|
#if QK_K == 256
|
|
|
|
#define QK_NL 16
|
|
|
|
#else
|
|
|
|
#define QK_NL 4
|
|
|
|
#endif
|
|
|
|
|
2023-12-13 13:04:25 +01:00
|
|
|
//
|
|
|
|
// get rows
|
|
|
|
//
|
|
|
|
|
2023-12-07 21:26:54 +01:00
|
|
|
typedef void (get_rows_t)(
|
|
|
|
device const void * src0,
|
2023-12-13 13:04:25 +01:00
|
|
|
device const char * src1,
|
2023-12-07 21:26:54 +01:00
|
|
|
device float * dst,
|
|
|
|
constant int64_t & ne00,
|
|
|
|
constant uint64_t & nb01,
|
2023-12-13 13:04:25 +01:00
|
|
|
constant uint64_t & nb02,
|
|
|
|
constant int64_t & ne10,
|
|
|
|
constant uint64_t & nb10,
|
|
|
|
constant uint64_t & nb11,
|
2023-12-07 21:26:54 +01:00
|
|
|
constant uint64_t & nb1,
|
2023-12-13 13:04:25 +01:00
|
|
|
constant uint64_t & nb2,
|
|
|
|
uint3, uint, uint3);
|
2023-08-16 22:07:04 +02:00
|
|
|
|
2023-12-13 13:04:25 +01:00
|
|
|
//template [[host_name("kernel_get_rows_f32")]] kernel get_rows_t kernel_get_rows<float4x4, 1, dequantize_f32>;
|
|
|
|
//template [[host_name("kernel_get_rows_f16")]] kernel get_rows_t kernel_get_rows<half4x4, 1, dequantize_f16>;
|
2023-08-16 22:07:04 +02:00
|
|
|
template [[host_name("kernel_get_rows_q4_0")]] kernel get_rows_t kernel_get_rows<block_q4_0, 2, dequantize_q4_0>;
|
|
|
|
template [[host_name("kernel_get_rows_q4_1")]] kernel get_rows_t kernel_get_rows<block_q4_1, 2, dequantize_q4_1>;
|
2023-10-18 14:21:48 +02:00
|
|
|
template [[host_name("kernel_get_rows_q5_0")]] kernel get_rows_t kernel_get_rows<block_q5_0, 2, dequantize_q5_0>;
|
|
|
|
template [[host_name("kernel_get_rows_q5_1")]] kernel get_rows_t kernel_get_rows<block_q5_1, 2, dequantize_q5_1>;
|
2023-08-24 15:19:57 +02:00
|
|
|
template [[host_name("kernel_get_rows_q8_0")]] kernel get_rows_t kernel_get_rows<block_q8_0, 2, dequantize_q8_0>;
|
2023-08-16 22:07:04 +02:00
|
|
|
template [[host_name("kernel_get_rows_q2_K")]] kernel get_rows_t kernel_get_rows<block_q2_K, QK_NL, dequantize_q2_K>;
|
|
|
|
template [[host_name("kernel_get_rows_q3_K")]] kernel get_rows_t kernel_get_rows<block_q3_K, QK_NL, dequantize_q3_K>;
|
|
|
|
template [[host_name("kernel_get_rows_q4_K")]] kernel get_rows_t kernel_get_rows<block_q4_K, QK_NL, dequantize_q4_K>;
|
|
|
|
template [[host_name("kernel_get_rows_q5_K")]] kernel get_rows_t kernel_get_rows<block_q5_K, QK_NL, dequantize_q5_K>;
|
|
|
|
template [[host_name("kernel_get_rows_q6_K")]] kernel get_rows_t kernel_get_rows<block_q6_K, QK_NL, dequantize_q6_K>;
|
2024-01-08 16:02:32 +01:00
|
|
|
template [[host_name("kernel_get_rows_iq2_xxs")]] kernel get_rows_t kernel_get_rows<block_iq2_xxs, QK_NL, dequantize_iq2_xxs>;
|
2024-01-11 20:39:39 +01:00
|
|
|
template [[host_name("kernel_get_rows_iq2_xs")]] kernel get_rows_t kernel_get_rows<block_iq2_xs, QK_NL, dequantize_iq2_xs>;
|
2024-01-30 14:14:12 +01:00
|
|
|
template [[host_name("kernel_get_rows_iq3_xxs")]] kernel get_rows_t kernel_get_rows<block_iq3_xxs, QK_NL, dequantize_iq3_xxs>;
|
2023-08-16 22:07:04 +02:00
|
|
|
|
2023-12-13 13:04:25 +01:00
|
|
|
//
|
|
|
|
// matrix-matrix multiplication
|
|
|
|
//
|
|
|
|
|
2023-09-15 10:09:24 +02:00
|
|
|
typedef void (mat_mm_t)(
|
|
|
|
device const uchar * src0,
|
|
|
|
device const uchar * src1,
|
|
|
|
device float * dst,
|
|
|
|
constant int64_t & ne00,
|
|
|
|
constant int64_t & ne02,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant uint64_t & nb02,
|
2023-09-15 10:09:24 +02:00
|
|
|
constant int64_t & ne12,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant uint64_t & nb10,
|
|
|
|
constant uint64_t & nb11,
|
|
|
|
constant uint64_t & nb12,
|
2023-09-15 10:09:24 +02:00
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
2023-12-07 21:26:54 +01:00
|
|
|
constant uint & r2,
|
|
|
|
constant uint & r3,
|
|
|
|
threadgroup uchar *,
|
|
|
|
uint3, uint, uint);
|
2023-09-15 10:09:24 +02:00
|
|
|
|
2023-09-15 18:06:03 +02:00
|
|
|
template [[host_name("kernel_mul_mm_f32_f32")]] kernel mat_mm_t kernel_mul_mm<float4x4, 1, dequantize_f32>;
|
2023-09-15 10:09:24 +02:00
|
|
|
template [[host_name("kernel_mul_mm_f16_f32")]] kernel mat_mm_t kernel_mul_mm<half4x4, 1, dequantize_f16>;
|
|
|
|
template [[host_name("kernel_mul_mm_q4_0_f32")]] kernel mat_mm_t kernel_mul_mm<block_q4_0, 2, dequantize_q4_0>;
|
|
|
|
template [[host_name("kernel_mul_mm_q4_1_f32")]] kernel mat_mm_t kernel_mul_mm<block_q4_1, 2, dequantize_q4_1>;
|
2023-10-18 14:21:48 +02:00
|
|
|
template [[host_name("kernel_mul_mm_q5_0_f32")]] kernel mat_mm_t kernel_mul_mm<block_q5_0, 2, dequantize_q5_0>;
|
|
|
|
template [[host_name("kernel_mul_mm_q5_1_f32")]] kernel mat_mm_t kernel_mul_mm<block_q5_1, 2, dequantize_q5_1>;
|
2023-09-15 10:09:24 +02:00
|
|
|
template [[host_name("kernel_mul_mm_q8_0_f32")]] kernel mat_mm_t kernel_mul_mm<block_q8_0, 2, dequantize_q8_0>;
|
2023-08-16 22:07:04 +02:00
|
|
|
template [[host_name("kernel_mul_mm_q2_K_f32")]] kernel mat_mm_t kernel_mul_mm<block_q2_K, QK_NL, dequantize_q2_K>;
|
|
|
|
template [[host_name("kernel_mul_mm_q3_K_f32")]] kernel mat_mm_t kernel_mul_mm<block_q3_K, QK_NL, dequantize_q3_K>;
|
|
|
|
template [[host_name("kernel_mul_mm_q4_K_f32")]] kernel mat_mm_t kernel_mul_mm<block_q4_K, QK_NL, dequantize_q4_K>;
|
|
|
|
template [[host_name("kernel_mul_mm_q5_K_f32")]] kernel mat_mm_t kernel_mul_mm<block_q5_K, QK_NL, dequantize_q5_K>;
|
|
|
|
template [[host_name("kernel_mul_mm_q6_K_f32")]] kernel mat_mm_t kernel_mul_mm<block_q6_K, QK_NL, dequantize_q6_K>;
|
2024-01-08 16:02:32 +01:00
|
|
|
template [[host_name("kernel_mul_mm_iq2_xxs_f32")]] kernel mat_mm_t kernel_mul_mm<block_iq2_xxs, QK_NL, dequantize_iq2_xxs>;
|
2024-01-11 20:39:39 +01:00
|
|
|
template [[host_name("kernel_mul_mm_iq2_xs_f32")]] kernel mat_mm_t kernel_mul_mm<block_iq2_xs, QK_NL, dequantize_iq2_xs>;
|
2024-01-30 14:14:12 +01:00
|
|
|
template [[host_name("kernel_mul_mm_iq3_xxs_f32")]] kernel mat_mm_t kernel_mul_mm<block_iq3_xxs, QK_NL, dequantize_iq3_xxs>;
|
2023-12-07 21:26:54 +01:00
|
|
|
|
2023-12-13 13:04:25 +01:00
|
|
|
//
|
|
|
|
// indirect matrix-matrix multiplication
|
|
|
|
//
|
|
|
|
|
2023-12-07 21:26:54 +01:00
|
|
|
typedef void (mat_mm_id_t)(
|
2023-12-13 13:04:25 +01:00
|
|
|
device const uchar * ids,
|
2023-12-07 21:26:54 +01:00
|
|
|
device const uchar * src1,
|
2024-01-02 20:07:47 +01:00
|
|
|
device float * dst,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant uint64_t & nbi1,
|
2023-12-07 21:26:54 +01:00
|
|
|
constant int64_t & ne00,
|
|
|
|
constant int64_t & ne02,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant uint64_t & nb02,
|
2023-12-07 21:26:54 +01:00
|
|
|
constant int64_t & ne12,
|
2023-12-13 13:04:25 +01:00
|
|
|
constant int64_t & ne13,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant uint64_t & nb10,
|
|
|
|
constant uint64_t & nb11,
|
|
|
|
constant uint64_t & nb12,
|
2023-12-07 21:26:54 +01:00
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant uint64_t & nb1,
|
2023-12-07 21:26:54 +01:00
|
|
|
constant uint & r2,
|
|
|
|
constant uint & r3,
|
|
|
|
constant int & idx,
|
|
|
|
device const uchar * src00,
|
|
|
|
device const uchar * src01,
|
|
|
|
device const uchar * src02,
|
|
|
|
device const uchar * src03,
|
|
|
|
device const uchar * src04,
|
|
|
|
device const uchar * src05,
|
|
|
|
device const uchar * src06,
|
|
|
|
device const uchar * src07,
|
|
|
|
threadgroup uchar *,
|
|
|
|
uint3, uint, uint);
|
|
|
|
|
|
|
|
template [[host_name("kernel_mul_mm_id_f32_f32")]] kernel mat_mm_id_t kernel_mul_mm_id<float4x4, 1, dequantize_f32>;
|
|
|
|
template [[host_name("kernel_mul_mm_id_f16_f32")]] kernel mat_mm_id_t kernel_mul_mm_id<half4x4, 1, dequantize_f16>;
|
|
|
|
template [[host_name("kernel_mul_mm_id_q4_0_f32")]] kernel mat_mm_id_t kernel_mul_mm_id<block_q4_0, 2, dequantize_q4_0>;
|
|
|
|
template [[host_name("kernel_mul_mm_id_q4_1_f32")]] kernel mat_mm_id_t kernel_mul_mm_id<block_q4_1, 2, dequantize_q4_1>;
|
|
|
|
template [[host_name("kernel_mul_mm_id_q5_0_f32")]] kernel mat_mm_id_t kernel_mul_mm_id<block_q5_0, 2, dequantize_q5_0>;
|
|
|
|
template [[host_name("kernel_mul_mm_id_q5_1_f32")]] kernel mat_mm_id_t kernel_mul_mm_id<block_q5_1, 2, dequantize_q5_1>;
|
|
|
|
template [[host_name("kernel_mul_mm_id_q8_0_f32")]] kernel mat_mm_id_t kernel_mul_mm_id<block_q8_0, 2, dequantize_q8_0>;
|
|
|
|
template [[host_name("kernel_mul_mm_id_q2_K_f32")]] kernel mat_mm_id_t kernel_mul_mm_id<block_q2_K, QK_NL, dequantize_q2_K>;
|
|
|
|
template [[host_name("kernel_mul_mm_id_q3_K_f32")]] kernel mat_mm_id_t kernel_mul_mm_id<block_q3_K, QK_NL, dequantize_q3_K>;
|
|
|
|
template [[host_name("kernel_mul_mm_id_q4_K_f32")]] kernel mat_mm_id_t kernel_mul_mm_id<block_q4_K, QK_NL, dequantize_q4_K>;
|
|
|
|
template [[host_name("kernel_mul_mm_id_q5_K_f32")]] kernel mat_mm_id_t kernel_mul_mm_id<block_q5_K, QK_NL, dequantize_q5_K>;
|
|
|
|
template [[host_name("kernel_mul_mm_id_q6_K_f32")]] kernel mat_mm_id_t kernel_mul_mm_id<block_q6_K, QK_NL, dequantize_q6_K>;
|
2024-01-08 16:02:32 +01:00
|
|
|
template [[host_name("kernel_mul_mm_id_iq2_xxs_f32")]] kernel mat_mm_id_t kernel_mul_mm_id<block_iq2_xxs, QK_NL, dequantize_iq2_xxs>;
|
2024-01-11 20:39:39 +01:00
|
|
|
template [[host_name("kernel_mul_mm_id_iq2_xs_f32")]] kernel mat_mm_id_t kernel_mul_mm_id<block_iq2_xs, QK_NL, dequantize_iq2_xs>;
|
2024-01-30 14:14:12 +01:00
|
|
|
template [[host_name("kernel_mul_mm_id_iq3_xxs_f32")]] kernel mat_mm_id_t kernel_mul_mm_id<block_iq3_xxs, QK_NL, dequantize_iq3_xxs>;
|
2023-12-13 13:04:25 +01:00
|
|
|
|
|
|
|
//
|
|
|
|
// matrix-vector multiplication
|
|
|
|
//
|
|
|
|
|
|
|
|
[[host_name("kernel_mul_mv_id_f32_f32")]]
|
|
|
|
kernel void kernel_mul_mv_id_f32_f32(
|
|
|
|
device const char * ids,
|
|
|
|
device const char * src1,
|
2024-01-02 20:07:47 +01:00
|
|
|
device float * dst,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant uint64_t & nbi1,
|
2023-12-13 13:04:25 +01:00
|
|
|
constant int64_t & ne00,
|
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant uint64_t & nb00,
|
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant uint64_t & nb02,
|
|
|
|
constant int64_t & ne10,
|
|
|
|
constant int64_t & ne11,
|
|
|
|
constant int64_t & ne12,
|
|
|
|
constant int64_t & ne13,
|
|
|
|
constant uint64_t & nb10,
|
|
|
|
constant uint64_t & nb11,
|
|
|
|
constant uint64_t & nb12,
|
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant uint64_t & nb1,
|
2023-12-13 13:04:25 +01:00
|
|
|
constant uint & r2,
|
|
|
|
constant uint & r3,
|
|
|
|
constant int & idx,
|
|
|
|
device const char * src00,
|
|
|
|
device const char * src01,
|
|
|
|
device const char * src02,
|
|
|
|
device const char * src03,
|
|
|
|
device const char * src04,
|
|
|
|
device const char * src05,
|
|
|
|
device const char * src06,
|
|
|
|
device const char * src07,
|
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint tiitg[[thread_index_in_threadgroup]],
|
|
|
|
uint tiisg[[thread_index_in_simdgroup]],
|
|
|
|
uint sgitg[[simdgroup_index_in_threadgroup]]) {
|
|
|
|
device const char * src0[8] = {src00, src01, src02, src03, src04, src05, src06, src07};
|
|
|
|
|
|
|
|
const int64_t bid = tgpig.z/(ne12*ne13);
|
|
|
|
|
|
|
|
tgpig.z = tgpig.z%(ne12*ne13);
|
|
|
|
|
|
|
|
const int32_t id = ((device int32_t *) (ids + bid*nbi1))[idx];
|
|
|
|
|
|
|
|
kernel_mul_mv_f32_f32_impl(
|
|
|
|
src0[id],
|
|
|
|
src1 + bid*nb11,
|
2024-01-02 20:07:47 +01:00
|
|
|
dst + bid*ne0,
|
2023-12-13 13:04:25 +01:00
|
|
|
ne00,
|
|
|
|
ne01,
|
|
|
|
ne02,
|
|
|
|
nb00,
|
|
|
|
nb01,
|
|
|
|
nb02,
|
|
|
|
ne10,
|
|
|
|
ne11,
|
|
|
|
ne12,
|
|
|
|
nb10,
|
|
|
|
nb11,
|
|
|
|
nb12,
|
|
|
|
ne0,
|
|
|
|
ne1,
|
|
|
|
r2,
|
|
|
|
r3,
|
|
|
|
tgpig,
|
|
|
|
tiisg);
|
|
|
|
}
|
|
|
|
|
|
|
|
[[host_name("kernel_mul_mv_id_f16_f32")]]
|
|
|
|
kernel void kernel_mul_mv_id_f16_f32(
|
|
|
|
device const char * ids,
|
|
|
|
device const char * src1,
|
2024-01-02 20:07:47 +01:00
|
|
|
device float * dst,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant uint64_t & nbi1,
|
2023-12-13 13:04:25 +01:00
|
|
|
constant int64_t & ne00,
|
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant uint64_t & nb00,
|
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant uint64_t & nb02,
|
|
|
|
constant int64_t & ne10,
|
|
|
|
constant int64_t & ne11,
|
|
|
|
constant int64_t & ne12,
|
|
|
|
constant int64_t & ne13,
|
|
|
|
constant uint64_t & nb10,
|
|
|
|
constant uint64_t & nb11,
|
|
|
|
constant uint64_t & nb12,
|
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant uint64_t & nb1,
|
2023-12-13 13:04:25 +01:00
|
|
|
constant uint & r2,
|
|
|
|
constant uint & r3,
|
|
|
|
constant int & idx,
|
|
|
|
device const char * src00,
|
|
|
|
device const char * src01,
|
|
|
|
device const char * src02,
|
|
|
|
device const char * src03,
|
|
|
|
device const char * src04,
|
|
|
|
device const char * src05,
|
|
|
|
device const char * src06,
|
|
|
|
device const char * src07,
|
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint tiitg[[thread_index_in_threadgroup]],
|
|
|
|
uint tiisg[[thread_index_in_simdgroup]],
|
|
|
|
uint sgitg[[simdgroup_index_in_threadgroup]]) {
|
|
|
|
device const char * src0[8] = {src00, src01, src02, src03, src04, src05, src06, src07};
|
|
|
|
|
|
|
|
const int64_t bid = tgpig.z/(ne12*ne13);
|
|
|
|
|
|
|
|
tgpig.z = tgpig.z%(ne12*ne13);
|
|
|
|
|
|
|
|
const int32_t id = ((device int32_t *) (ids + bid*nbi1))[idx];
|
|
|
|
|
|
|
|
kernel_mul_mv_f16_f32_impl(
|
|
|
|
src0[id],
|
|
|
|
src1 + bid*nb11,
|
2024-01-02 20:07:47 +01:00
|
|
|
dst + bid*ne0,
|
2023-12-13 13:04:25 +01:00
|
|
|
ne00,
|
|
|
|
ne01,
|
|
|
|
ne02,
|
|
|
|
nb00,
|
|
|
|
nb01,
|
|
|
|
nb02,
|
|
|
|
ne10,
|
|
|
|
ne11,
|
|
|
|
ne12,
|
|
|
|
nb10,
|
|
|
|
nb11,
|
|
|
|
nb12,
|
|
|
|
ne0,
|
|
|
|
ne1,
|
|
|
|
r2,
|
|
|
|
r3,
|
|
|
|
tgpig,
|
|
|
|
tiisg);
|
|
|
|
}
|
|
|
|
|
|
|
|
[[host_name("kernel_mul_mv_id_q8_0_f32")]]
|
|
|
|
kernel void kernel_mul_mv_id_q8_0_f32(
|
|
|
|
device const char * ids,
|
|
|
|
device const char * src1,
|
2024-01-02 20:07:47 +01:00
|
|
|
device float * dst,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant uint64_t & nbi1,
|
2023-12-13 13:04:25 +01:00
|
|
|
constant int64_t & ne00,
|
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant uint64_t & nb00,
|
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant uint64_t & nb02,
|
|
|
|
constant int64_t & ne10,
|
|
|
|
constant int64_t & ne11,
|
|
|
|
constant int64_t & ne12,
|
|
|
|
constant int64_t & ne13,
|
|
|
|
constant uint64_t & nb10,
|
|
|
|
constant uint64_t & nb11,
|
|
|
|
constant uint64_t & nb12,
|
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant uint64_t & nb1,
|
2023-12-13 13:04:25 +01:00
|
|
|
constant uint & r2,
|
|
|
|
constant uint & r3,
|
|
|
|
constant int & idx,
|
|
|
|
device const char * src00,
|
|
|
|
device const char * src01,
|
|
|
|
device const char * src02,
|
|
|
|
device const char * src03,
|
|
|
|
device const char * src04,
|
|
|
|
device const char * src05,
|
|
|
|
device const char * src06,
|
|
|
|
device const char * src07,
|
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint tiitg[[thread_index_in_threadgroup]],
|
|
|
|
uint tiisg[[thread_index_in_simdgroup]],
|
|
|
|
uint sgitg[[simdgroup_index_in_threadgroup]]) {
|
|
|
|
device const char * src0[8] = {src00, src01, src02, src03, src04, src05, src06, src07};
|
|
|
|
|
|
|
|
const int64_t bid = tgpig.z/(ne12*ne13);
|
|
|
|
|
|
|
|
tgpig.z = tgpig.z%(ne12*ne13);
|
|
|
|
|
|
|
|
const int32_t id = ((device int32_t *) (ids + bid*nbi1))[idx];
|
|
|
|
|
|
|
|
kernel_mul_mv_q8_0_f32_impl(
|
|
|
|
src0[id],
|
|
|
|
(device const float *) (src1 + bid*nb11),
|
2024-01-02 20:07:47 +01:00
|
|
|
dst + bid*ne0,
|
2023-12-13 13:04:25 +01:00
|
|
|
ne00,
|
|
|
|
ne01,
|
|
|
|
ne02,
|
|
|
|
ne10,
|
|
|
|
ne12,
|
|
|
|
ne0,
|
|
|
|
ne1,
|
|
|
|
r2,
|
|
|
|
r3,
|
|
|
|
tgpig,
|
|
|
|
tiisg,
|
|
|
|
sgitg);
|
|
|
|
}
|
|
|
|
|
|
|
|
[[host_name("kernel_mul_mv_id_q4_0_f32")]]
|
|
|
|
kernel void kernel_mul_mv_id_q4_0_f32(
|
|
|
|
device const char * ids,
|
|
|
|
device const char * src1,
|
2024-01-02 20:07:47 +01:00
|
|
|
device float * dst,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant uint64_t & nbi1,
|
2023-12-13 13:04:25 +01:00
|
|
|
constant int64_t & ne00,
|
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant uint64_t & nb00,
|
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant uint64_t & nb02,
|
|
|
|
constant int64_t & ne10,
|
|
|
|
constant int64_t & ne11,
|
|
|
|
constant int64_t & ne12,
|
|
|
|
constant int64_t & ne13,
|
|
|
|
constant uint64_t & nb10,
|
|
|
|
constant uint64_t & nb11,
|
|
|
|
constant uint64_t & nb12,
|
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant uint64_t & nb1,
|
2023-12-13 13:04:25 +01:00
|
|
|
constant uint & r2,
|
|
|
|
constant uint & r3,
|
|
|
|
constant int & idx,
|
|
|
|
device const char * src00,
|
|
|
|
device const char * src01,
|
|
|
|
device const char * src02,
|
|
|
|
device const char * src03,
|
|
|
|
device const char * src04,
|
|
|
|
device const char * src05,
|
|
|
|
device const char * src06,
|
|
|
|
device const char * src07,
|
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint tiitg[[thread_index_in_threadgroup]],
|
|
|
|
uint tiisg[[thread_index_in_simdgroup]],
|
|
|
|
uint sgitg[[simdgroup_index_in_threadgroup]]) {
|
|
|
|
device const char * src0[8] = {src00, src01, src02, src03, src04, src05, src06, src07};
|
|
|
|
|
|
|
|
const int64_t bid = tgpig.z/(ne12*ne13);
|
|
|
|
|
|
|
|
tgpig.z = tgpig.z%(ne12*ne13);
|
|
|
|
|
|
|
|
const int32_t id = ((device int32_t *) (ids + bid*nbi1))[idx];
|
|
|
|
|
|
|
|
mul_vec_q_n_f32_impl<block_q4_0, N_DST, N_SIMDGROUP, N_SIMDWIDTH>(
|
|
|
|
src0[id],
|
|
|
|
(device const float *) (src1 + bid*nb11),
|
2024-01-02 20:07:47 +01:00
|
|
|
dst + bid*ne0,
|
2023-12-13 13:04:25 +01:00
|
|
|
ne00,
|
|
|
|
ne01,
|
|
|
|
ne02,
|
|
|
|
ne10,
|
|
|
|
ne12,
|
|
|
|
ne0,
|
|
|
|
ne1,
|
|
|
|
r2,
|
|
|
|
r3,
|
|
|
|
tgpig,
|
|
|
|
tiisg,
|
|
|
|
sgitg);
|
|
|
|
}
|
|
|
|
|
|
|
|
[[host_name("kernel_mul_mv_id_q4_1_f32")]]
|
|
|
|
kernel void kernel_mul_mv_id_q4_1_f32(
|
|
|
|
device const char * ids,
|
|
|
|
device const char * src1,
|
2024-01-02 20:07:47 +01:00
|
|
|
device float * dst,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant uint64_t & nbi1,
|
2023-12-13 13:04:25 +01:00
|
|
|
constant int64_t & ne00,
|
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant uint64_t & nb00,
|
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant uint64_t & nb02,
|
|
|
|
constant int64_t & ne10,
|
|
|
|
constant int64_t & ne11,
|
|
|
|
constant int64_t & ne12,
|
|
|
|
constant int64_t & ne13,
|
|
|
|
constant uint64_t & nb10,
|
|
|
|
constant uint64_t & nb11,
|
|
|
|
constant uint64_t & nb12,
|
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant uint64_t & nb1,
|
2023-12-13 13:04:25 +01:00
|
|
|
constant uint & r2,
|
|
|
|
constant uint & r3,
|
|
|
|
constant int & idx,
|
|
|
|
device const char * src00,
|
|
|
|
device const char * src01,
|
|
|
|
device const char * src02,
|
|
|
|
device const char * src03,
|
|
|
|
device const char * src04,
|
|
|
|
device const char * src05,
|
|
|
|
device const char * src06,
|
|
|
|
device const char * src07,
|
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint tiitg[[thread_index_in_threadgroup]],
|
|
|
|
uint tiisg[[thread_index_in_simdgroup]],
|
|
|
|
uint sgitg[[simdgroup_index_in_threadgroup]]) {
|
|
|
|
device const char * src0[8] = {src00, src01, src02, src03, src04, src05, src06, src07};
|
|
|
|
|
|
|
|
const int64_t bid = tgpig.z/(ne12*ne13);
|
|
|
|
|
|
|
|
tgpig.z = tgpig.z%(ne12*ne13);
|
|
|
|
|
|
|
|
const int32_t id = ((device int32_t *) (ids + bid*nbi1))[idx];
|
|
|
|
|
|
|
|
mul_vec_q_n_f32_impl<block_q4_1, N_DST, N_SIMDGROUP, N_SIMDWIDTH>(
|
|
|
|
src0[id],
|
|
|
|
(device const float *) (src1 + bid*nb11),
|
2024-01-02 20:07:47 +01:00
|
|
|
dst + bid*ne0,
|
2023-12-13 13:04:25 +01:00
|
|
|
ne00,
|
|
|
|
ne01,
|
|
|
|
ne02,
|
|
|
|
ne10,
|
|
|
|
ne12,
|
|
|
|
ne0,
|
|
|
|
ne1,
|
|
|
|
r2,
|
|
|
|
r3,
|
|
|
|
tgpig,
|
|
|
|
tiisg,
|
|
|
|
sgitg);
|
|
|
|
}
|
|
|
|
|
|
|
|
[[host_name("kernel_mul_mv_id_q5_0_f32")]]
|
|
|
|
kernel void kernel_mul_mv_id_q5_0_f32(
|
|
|
|
device const char * ids,
|
|
|
|
device const char * src1,
|
2024-01-02 20:07:47 +01:00
|
|
|
device float * dst,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant uint64_t & nbi1,
|
2023-12-13 13:04:25 +01:00
|
|
|
constant int64_t & ne00,
|
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant uint64_t & nb00,
|
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant uint64_t & nb02,
|
|
|
|
constant int64_t & ne10,
|
|
|
|
constant int64_t & ne11,
|
|
|
|
constant int64_t & ne12,
|
|
|
|
constant int64_t & ne13,
|
|
|
|
constant uint64_t & nb10,
|
|
|
|
constant uint64_t & nb11,
|
|
|
|
constant uint64_t & nb12,
|
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant uint64_t & nb1,
|
2023-12-13 13:04:25 +01:00
|
|
|
constant uint & r2,
|
|
|
|
constant uint & r3,
|
|
|
|
constant int & idx,
|
|
|
|
device const char * src00,
|
|
|
|
device const char * src01,
|
|
|
|
device const char * src02,
|
|
|
|
device const char * src03,
|
|
|
|
device const char * src04,
|
|
|
|
device const char * src05,
|
|
|
|
device const char * src06,
|
|
|
|
device const char * src07,
|
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint tiitg[[thread_index_in_threadgroup]],
|
|
|
|
uint tiisg[[thread_index_in_simdgroup]],
|
|
|
|
uint sgitg[[simdgroup_index_in_threadgroup]]) {
|
|
|
|
device const char * src0[8] = {src00, src01, src02, src03, src04, src05, src06, src07};
|
|
|
|
|
|
|
|
const int64_t bid = tgpig.z/(ne12*ne13);
|
|
|
|
|
|
|
|
tgpig.z = tgpig.z%(ne12*ne13);
|
|
|
|
|
|
|
|
const int32_t id = ((device int32_t *) (ids + bid*nbi1))[idx];
|
|
|
|
|
|
|
|
mul_vec_q_n_f32_impl<block_q5_0, N_DST, N_SIMDGROUP, N_SIMDWIDTH>(
|
|
|
|
src0[id],
|
|
|
|
(device const float *) (src1 + bid*nb11),
|
2024-01-02 20:07:47 +01:00
|
|
|
dst + bid*ne0,
|
2023-12-13 13:04:25 +01:00
|
|
|
ne00,
|
|
|
|
ne01,
|
|
|
|
ne02,
|
|
|
|
ne10,
|
|
|
|
ne12,
|
|
|
|
ne0,
|
|
|
|
ne1,
|
|
|
|
r2,
|
|
|
|
r3,
|
|
|
|
tgpig,
|
|
|
|
tiisg,
|
|
|
|
sgitg);
|
|
|
|
}
|
|
|
|
|
|
|
|
[[host_name("kernel_mul_mv_id_q5_1_f32")]]
|
|
|
|
kernel void kernel_mul_mv_id_q5_1_f32(
|
|
|
|
device const char * ids,
|
|
|
|
device const char * src1,
|
2024-01-02 20:07:47 +01:00
|
|
|
device float * dst,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant uint64_t & nbi1,
|
2023-12-13 13:04:25 +01:00
|
|
|
constant int64_t & ne00,
|
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant uint64_t & nb00,
|
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant uint64_t & nb02,
|
|
|
|
constant int64_t & ne10,
|
|
|
|
constant int64_t & ne11,
|
|
|
|
constant int64_t & ne12,
|
|
|
|
constant int64_t & ne13,
|
|
|
|
constant uint64_t & nb10,
|
|
|
|
constant uint64_t & nb11,
|
|
|
|
constant uint64_t & nb12,
|
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant uint64_t & nb1,
|
2023-12-13 13:04:25 +01:00
|
|
|
constant uint & r2,
|
|
|
|
constant uint & r3,
|
|
|
|
constant int & idx,
|
|
|
|
device const char * src00,
|
|
|
|
device const char * src01,
|
|
|
|
device const char * src02,
|
|
|
|
device const char * src03,
|
|
|
|
device const char * src04,
|
|
|
|
device const char * src05,
|
|
|
|
device const char * src06,
|
|
|
|
device const char * src07,
|
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint tiitg[[thread_index_in_threadgroup]],
|
|
|
|
uint tiisg[[thread_index_in_simdgroup]],
|
|
|
|
uint sgitg[[simdgroup_index_in_threadgroup]]) {
|
|
|
|
device const char * src0[8] = {src00, src01, src02, src03, src04, src05, src06, src07};
|
|
|
|
|
|
|
|
const int64_t bid = tgpig.z/(ne12*ne13);
|
|
|
|
|
|
|
|
tgpig.z = tgpig.z%(ne12*ne13);
|
|
|
|
|
|
|
|
const int32_t id = ((device int32_t *) (ids + bid*nbi1))[idx];
|
|
|
|
|
|
|
|
mul_vec_q_n_f32_impl<block_q5_1, N_DST, N_SIMDGROUP, N_SIMDWIDTH>(
|
|
|
|
src0[id],
|
|
|
|
(device const float *) (src1 + bid*nb11),
|
2024-01-02 20:07:47 +01:00
|
|
|
dst + bid*ne0,
|
2023-12-13 13:04:25 +01:00
|
|
|
ne00,
|
|
|
|
ne01,
|
|
|
|
ne02,
|
|
|
|
ne10,
|
|
|
|
ne12,
|
|
|
|
ne0,
|
|
|
|
ne1,
|
|
|
|
r2,
|
|
|
|
r3,
|
|
|
|
tgpig,
|
|
|
|
tiisg,
|
|
|
|
sgitg);
|
|
|
|
}
|
|
|
|
|
|
|
|
[[host_name("kernel_mul_mv_id_q2_K_f32")]]
|
|
|
|
kernel void kernel_mul_mv_id_q2_K_f32(
|
|
|
|
device const char * ids,
|
|
|
|
device const char * src1,
|
2024-01-02 20:07:47 +01:00
|
|
|
device float * dst,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant uint64_t & nbi1,
|
2023-12-13 13:04:25 +01:00
|
|
|
constant int64_t & ne00,
|
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant uint64_t & nb00,
|
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant uint64_t & nb02,
|
|
|
|
constant int64_t & ne10,
|
|
|
|
constant int64_t & ne11,
|
|
|
|
constant int64_t & ne12,
|
|
|
|
constant int64_t & ne13,
|
|
|
|
constant uint64_t & nb10,
|
|
|
|
constant uint64_t & nb11,
|
|
|
|
constant uint64_t & nb12,
|
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant uint64_t & nb1,
|
2023-12-13 13:04:25 +01:00
|
|
|
constant uint & r2,
|
|
|
|
constant uint & r3,
|
|
|
|
constant int & idx,
|
|
|
|
device const char * src00,
|
|
|
|
device const char * src01,
|
|
|
|
device const char * src02,
|
|
|
|
device const char * src03,
|
|
|
|
device const char * src04,
|
|
|
|
device const char * src05,
|
|
|
|
device const char * src06,
|
|
|
|
device const char * src07,
|
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint tiitg[[thread_index_in_threadgroup]],
|
|
|
|
uint tiisg[[thread_index_in_simdgroup]],
|
|
|
|
uint sgitg[[simdgroup_index_in_threadgroup]]) {
|
|
|
|
device const char * src0[8] = {src00, src01, src02, src03, src04, src05, src06, src07};
|
|
|
|
|
|
|
|
const int64_t bid = tgpig.z/(ne12*ne13);
|
|
|
|
|
|
|
|
tgpig.z = tgpig.z%(ne12*ne13);
|
|
|
|
|
|
|
|
const int32_t id = ((device int32_t *) (ids + bid*nbi1))[idx];
|
|
|
|
|
|
|
|
kernel_mul_mv_q2_K_f32_impl(
|
|
|
|
src0[id],
|
|
|
|
(device const float *) (src1 + bid*nb11),
|
2024-01-02 20:07:47 +01:00
|
|
|
dst + bid*ne0,
|
2023-12-13 13:04:25 +01:00
|
|
|
ne00,
|
|
|
|
ne01,
|
|
|
|
ne02,
|
|
|
|
ne10,
|
|
|
|
ne12,
|
|
|
|
ne0,
|
|
|
|
ne1,
|
|
|
|
r2,
|
|
|
|
r3,
|
|
|
|
tgpig,
|
|
|
|
tiisg,
|
|
|
|
sgitg);
|
|
|
|
}
|
|
|
|
|
|
|
|
[[host_name("kernel_mul_mv_id_q3_K_f32")]]
|
|
|
|
kernel void kernel_mul_mv_id_q3_K_f32(
|
|
|
|
device const char * ids,
|
|
|
|
device const char * src1,
|
2024-01-02 20:07:47 +01:00
|
|
|
device float * dst,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant uint64_t & nbi1,
|
2023-12-13 13:04:25 +01:00
|
|
|
constant int64_t & ne00,
|
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant uint64_t & nb00,
|
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant uint64_t & nb02,
|
|
|
|
constant int64_t & ne10,
|
|
|
|
constant int64_t & ne11,
|
|
|
|
constant int64_t & ne12,
|
|
|
|
constant int64_t & ne13,
|
|
|
|
constant uint64_t & nb10,
|
|
|
|
constant uint64_t & nb11,
|
|
|
|
constant uint64_t & nb12,
|
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant uint64_t & nb1,
|
2023-12-13 13:04:25 +01:00
|
|
|
constant uint & r2,
|
|
|
|
constant uint & r3,
|
|
|
|
constant int & idx,
|
|
|
|
device const char * src00,
|
|
|
|
device const char * src01,
|
|
|
|
device const char * src02,
|
|
|
|
device const char * src03,
|
|
|
|
device const char * src04,
|
|
|
|
device const char * src05,
|
|
|
|
device const char * src06,
|
|
|
|
device const char * src07,
|
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint tiitg[[thread_index_in_threadgroup]],
|
|
|
|
uint tiisg[[thread_index_in_simdgroup]],
|
|
|
|
uint sgitg[[simdgroup_index_in_threadgroup]]) {
|
|
|
|
device const char * src0[8] = {src00, src01, src02, src03, src04, src05, src06, src07};
|
|
|
|
|
|
|
|
const int64_t bid = tgpig.z/(ne12*ne13);
|
|
|
|
|
|
|
|
tgpig.z = tgpig.z%(ne12*ne13);
|
|
|
|
|
|
|
|
const int32_t id = ((device int32_t *) (ids + bid*nbi1))[idx];
|
|
|
|
|
|
|
|
kernel_mul_mv_q3_K_f32_impl(
|
|
|
|
src0[id],
|
|
|
|
(device const float *) (src1 + bid*nb11),
|
2024-01-02 20:07:47 +01:00
|
|
|
dst + bid*ne0,
|
2023-12-13 13:04:25 +01:00
|
|
|
ne00,
|
|
|
|
ne01,
|
|
|
|
ne02,
|
|
|
|
ne10,
|
|
|
|
ne12,
|
|
|
|
ne0,
|
|
|
|
ne1,
|
|
|
|
r2,
|
|
|
|
r3,
|
|
|
|
tgpig,
|
|
|
|
tiisg,
|
|
|
|
sgitg);
|
|
|
|
}
|
|
|
|
|
|
|
|
[[host_name("kernel_mul_mv_id_q4_K_f32")]]
|
|
|
|
kernel void kernel_mul_mv_id_q4_K_f32(
|
|
|
|
device const char * ids,
|
|
|
|
device const char * src1,
|
2024-01-02 20:07:47 +01:00
|
|
|
device float * dst,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant uint64_t & nbi1,
|
2023-12-13 13:04:25 +01:00
|
|
|
constant int64_t & ne00,
|
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant uint64_t & nb00,
|
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant uint64_t & nb02,
|
|
|
|
constant int64_t & ne10,
|
|
|
|
constant int64_t & ne11,
|
|
|
|
constant int64_t & ne12,
|
|
|
|
constant int64_t & ne13,
|
|
|
|
constant uint64_t & nb10,
|
|
|
|
constant uint64_t & nb11,
|
|
|
|
constant uint64_t & nb12,
|
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant uint64_t & nb1,
|
2023-12-13 13:04:25 +01:00
|
|
|
constant uint & r2,
|
|
|
|
constant uint & r3,
|
|
|
|
constant int & idx,
|
|
|
|
device const char * src00,
|
|
|
|
device const char * src01,
|
|
|
|
device const char * src02,
|
|
|
|
device const char * src03,
|
|
|
|
device const char * src04,
|
|
|
|
device const char * src05,
|
|
|
|
device const char * src06,
|
|
|
|
device const char * src07,
|
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint tiitg[[thread_index_in_threadgroup]],
|
|
|
|
uint tiisg[[thread_index_in_simdgroup]],
|
|
|
|
uint sgitg[[simdgroup_index_in_threadgroup]]) {
|
|
|
|
device const char * src0[8] = {src00, src01, src02, src03, src04, src05, src06, src07};
|
|
|
|
|
|
|
|
const int64_t bid = tgpig.z/(ne12*ne13);
|
|
|
|
|
|
|
|
tgpig.z = tgpig.z%(ne12*ne13);
|
|
|
|
|
|
|
|
const int32_t id = ((device int32_t *) (ids + bid*nbi1))[idx];
|
|
|
|
|
|
|
|
kernel_mul_mv_q4_K_f32_impl(
|
|
|
|
src0[id],
|
|
|
|
(device const float *) (src1 + bid*nb11),
|
2024-01-02 20:07:47 +01:00
|
|
|
dst + bid*ne0,
|
2023-12-13 13:04:25 +01:00
|
|
|
ne00,
|
|
|
|
ne01,
|
|
|
|
ne02,
|
|
|
|
ne10,
|
|
|
|
ne12,
|
|
|
|
ne0,
|
|
|
|
ne1,
|
|
|
|
r2,
|
|
|
|
r3,
|
|
|
|
tgpig,
|
|
|
|
tiisg,
|
|
|
|
sgitg);
|
|
|
|
}
|
|
|
|
|
|
|
|
[[host_name("kernel_mul_mv_id_q5_K_f32")]]
|
|
|
|
kernel void kernel_mul_mv_id_q5_K_f32(
|
|
|
|
device const char * ids,
|
|
|
|
device const char * src1,
|
2024-01-02 20:07:47 +01:00
|
|
|
device float * dst,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant uint64_t & nbi1,
|
2023-12-13 13:04:25 +01:00
|
|
|
constant int64_t & ne00,
|
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant uint64_t & nb00,
|
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant uint64_t & nb02,
|
|
|
|
constant int64_t & ne10,
|
|
|
|
constant int64_t & ne11,
|
|
|
|
constant int64_t & ne12,
|
|
|
|
constant int64_t & ne13,
|
|
|
|
constant uint64_t & nb10,
|
|
|
|
constant uint64_t & nb11,
|
|
|
|
constant uint64_t & nb12,
|
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant uint64_t & nb1,
|
2023-12-13 13:04:25 +01:00
|
|
|
constant uint & r2,
|
|
|
|
constant uint & r3,
|
|
|
|
constant int & idx,
|
|
|
|
device const char * src00,
|
|
|
|
device const char * src01,
|
|
|
|
device const char * src02,
|
|
|
|
device const char * src03,
|
|
|
|
device const char * src04,
|
|
|
|
device const char * src05,
|
|
|
|
device const char * src06,
|
|
|
|
device const char * src07,
|
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint tiitg[[thread_index_in_threadgroup]],
|
|
|
|
uint tiisg[[thread_index_in_simdgroup]],
|
|
|
|
uint sgitg[[simdgroup_index_in_threadgroup]]) {
|
|
|
|
device const char * src0[8] = {src00, src01, src02, src03, src04, src05, src06, src07};
|
|
|
|
|
|
|
|
const int64_t bid = tgpig.z/(ne12*ne13);
|
|
|
|
|
|
|
|
tgpig.z = tgpig.z%(ne12*ne13);
|
|
|
|
|
|
|
|
const int32_t id = ((device int32_t *) (ids + bid*nbi1))[idx];
|
|
|
|
|
|
|
|
kernel_mul_mv_q5_K_f32_impl(
|
|
|
|
src0[id],
|
|
|
|
(device const float *) (src1 + bid*nb11),
|
2024-01-02 20:07:47 +01:00
|
|
|
dst + bid*ne0,
|
2023-12-13 13:04:25 +01:00
|
|
|
ne00,
|
|
|
|
ne01,
|
|
|
|
ne02,
|
|
|
|
ne10,
|
|
|
|
ne12,
|
|
|
|
ne0,
|
|
|
|
ne1,
|
|
|
|
r2,
|
|
|
|
r3,
|
|
|
|
tgpig,
|
|
|
|
tiisg,
|
|
|
|
sgitg);
|
|
|
|
}
|
|
|
|
|
|
|
|
[[host_name("kernel_mul_mv_id_q6_K_f32")]]
|
|
|
|
kernel void kernel_mul_mv_id_q6_K_f32(
|
|
|
|
device const char * ids,
|
|
|
|
device const char * src1,
|
2024-01-02 20:07:47 +01:00
|
|
|
device float * dst,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant uint64_t & nbi1,
|
2023-12-13 13:04:25 +01:00
|
|
|
constant int64_t & ne00,
|
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant uint64_t & nb00,
|
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant uint64_t & nb02,
|
|
|
|
constant int64_t & ne10,
|
|
|
|
constant int64_t & ne11,
|
|
|
|
constant int64_t & ne12,
|
|
|
|
constant int64_t & ne13,
|
|
|
|
constant uint64_t & nb10,
|
|
|
|
constant uint64_t & nb11,
|
|
|
|
constant uint64_t & nb12,
|
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
2024-01-02 09:57:44 +01:00
|
|
|
constant uint64_t & nb1,
|
2023-12-13 13:04:25 +01:00
|
|
|
constant uint & r2,
|
|
|
|
constant uint & r3,
|
|
|
|
constant int & idx,
|
|
|
|
device const char * src00,
|
|
|
|
device const char * src01,
|
|
|
|
device const char * src02,
|
|
|
|
device const char * src03,
|
|
|
|
device const char * src04,
|
|
|
|
device const char * src05,
|
|
|
|
device const char * src06,
|
|
|
|
device const char * src07,
|
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint tiitg[[thread_index_in_threadgroup]],
|
|
|
|
uint tiisg[[thread_index_in_simdgroup]],
|
|
|
|
uint sgitg[[simdgroup_index_in_threadgroup]]) {
|
|
|
|
device const char * src0[8] = {src00, src01, src02, src03, src04, src05, src06, src07};
|
|
|
|
|
|
|
|
const int64_t bid = tgpig.z/(ne12*ne13);
|
|
|
|
|
|
|
|
tgpig.z = tgpig.z%(ne12*ne13);
|
|
|
|
|
|
|
|
const int32_t id = ((device int32_t *) (ids + bid*nbi1))[idx];
|
|
|
|
|
|
|
|
kernel_mul_mv_q6_K_f32_impl(
|
|
|
|
src0[id],
|
|
|
|
(device const float *) (src1 + bid*nb11),
|
2024-01-02 20:07:47 +01:00
|
|
|
dst + bid*ne0,
|
2023-12-13 13:04:25 +01:00
|
|
|
ne00,
|
|
|
|
ne01,
|
|
|
|
ne02,
|
|
|
|
ne10,
|
|
|
|
ne12,
|
|
|
|
ne0,
|
|
|
|
ne1,
|
|
|
|
r2,
|
|
|
|
r3,
|
|
|
|
tgpig,
|
|
|
|
tiisg,
|
|
|
|
sgitg);
|
|
|
|
}
|
2024-01-08 16:02:32 +01:00
|
|
|
|
|
|
|
[[host_name("kernel_mul_mv_id_iq2_xxs_f32")]]
|
|
|
|
kernel void kernel_mul_mv_id_iq2_xxs_f32(
|
|
|
|
device const char * ids,
|
|
|
|
device const char * src1,
|
|
|
|
device float * dst,
|
|
|
|
constant uint64_t & nbi1,
|
|
|
|
constant int64_t & ne00,
|
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant uint64_t & nb00,
|
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant uint64_t & nb02,
|
|
|
|
constant int64_t & ne10,
|
|
|
|
constant int64_t & ne11,
|
|
|
|
constant int64_t & ne12,
|
|
|
|
constant int64_t & ne13,
|
|
|
|
constant uint64_t & nb10,
|
|
|
|
constant uint64_t & nb11,
|
|
|
|
constant uint64_t & nb12,
|
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
|
|
|
constant uint64_t & nb1,
|
|
|
|
constant uint & r2,
|
|
|
|
constant uint & r3,
|
|
|
|
constant int & idx,
|
|
|
|
device const char * src00,
|
|
|
|
device const char * src01,
|
|
|
|
device const char * src02,
|
|
|
|
device const char * src03,
|
|
|
|
device const char * src04,
|
|
|
|
device const char * src05,
|
|
|
|
device const char * src06,
|
|
|
|
device const char * src07,
|
|
|
|
threadgroup int8_t * shared_values [[threadgroup(0)]],
|
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint tiitg[[thread_index_in_threadgroup]],
|
|
|
|
uint tiisg[[thread_index_in_simdgroup]],
|
|
|
|
uint sgitg[[simdgroup_index_in_threadgroup]]) {
|
|
|
|
device const char * src0[8] = {src00, src01, src02, src03, src04, src05, src06, src07};
|
|
|
|
|
|
|
|
const int64_t bid = tgpig.z/(ne12*ne13);
|
|
|
|
|
|
|
|
tgpig.z = tgpig.z%(ne12*ne13);
|
|
|
|
|
|
|
|
const int32_t id = ((device int32_t *) (ids + bid*nbi1))[idx];
|
|
|
|
|
|
|
|
kernel_mul_mv_iq2_xxs_f32_impl(
|
|
|
|
src0[id],
|
|
|
|
(device const float *) (src1 + bid*nb11),
|
|
|
|
dst + bid*ne0,
|
|
|
|
ne00,
|
|
|
|
ne01,
|
|
|
|
ne02,
|
|
|
|
ne10,
|
|
|
|
ne12,
|
|
|
|
ne0,
|
|
|
|
ne1,
|
|
|
|
r2,
|
|
|
|
r3,
|
|
|
|
shared_values,
|
|
|
|
tgpig,
|
|
|
|
tiisg,
|
|
|
|
sgitg);
|
|
|
|
}
|
2024-01-11 20:39:39 +01:00
|
|
|
|
|
|
|
[[host_name("kernel_mul_mv_id_iq2_xs_f32")]]
|
|
|
|
kernel void kernel_mul_mv_id_iq2_xs_f32(
|
|
|
|
device const char * ids,
|
|
|
|
device const char * src1,
|
|
|
|
device float * dst,
|
|
|
|
constant uint64_t & nbi1,
|
|
|
|
constant int64_t & ne00,
|
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant uint64_t & nb00,
|
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant uint64_t & nb02,
|
|
|
|
constant int64_t & ne10,
|
|
|
|
constant int64_t & ne11,
|
|
|
|
constant int64_t & ne12,
|
|
|
|
constant int64_t & ne13,
|
|
|
|
constant uint64_t & nb10,
|
|
|
|
constant uint64_t & nb11,
|
|
|
|
constant uint64_t & nb12,
|
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
|
|
|
constant uint64_t & nb1,
|
|
|
|
constant uint & r2,
|
|
|
|
constant uint & r3,
|
|
|
|
constant int & idx,
|
|
|
|
device const char * src00,
|
|
|
|
device const char * src01,
|
|
|
|
device const char * src02,
|
|
|
|
device const char * src03,
|
|
|
|
device const char * src04,
|
|
|
|
device const char * src05,
|
|
|
|
device const char * src06,
|
|
|
|
device const char * src07,
|
|
|
|
threadgroup int8_t * shared_values [[threadgroup(0)]],
|
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint tiitg[[thread_index_in_threadgroup]],
|
|
|
|
uint tiisg[[thread_index_in_simdgroup]],
|
|
|
|
uint sgitg[[simdgroup_index_in_threadgroup]]) {
|
|
|
|
device const char * src0[8] = {src00, src01, src02, src03, src04, src05, src06, src07};
|
|
|
|
|
|
|
|
const int64_t bid = tgpig.z/(ne12*ne13);
|
|
|
|
|
|
|
|
tgpig.z = tgpig.z%(ne12*ne13);
|
|
|
|
|
|
|
|
const int32_t id = ((device int32_t *) (ids + bid*nbi1))[idx];
|
|
|
|
|
|
|
|
kernel_mul_mv_iq2_xs_f32_impl(
|
|
|
|
src0[id],
|
|
|
|
(device const float *) (src1 + bid*nb11),
|
|
|
|
dst + bid*ne0,
|
|
|
|
ne00,
|
|
|
|
ne01,
|
|
|
|
ne02,
|
|
|
|
ne10,
|
|
|
|
ne12,
|
|
|
|
ne0,
|
|
|
|
ne1,
|
|
|
|
r2,
|
|
|
|
r3,
|
|
|
|
shared_values,
|
|
|
|
tgpig,
|
|
|
|
tiisg,
|
|
|
|
sgitg);
|
|
|
|
}
|
2024-01-30 14:14:12 +01:00
|
|
|
|
|
|
|
[[host_name("kernel_mul_mv_id_iq3_xxs_f32")]]
|
|
|
|
kernel void kernel_mul_mv_id_iq3_xxs_f32(
|
|
|
|
device const char * ids,
|
|
|
|
device const char * src1,
|
|
|
|
device float * dst,
|
|
|
|
constant uint64_t & nbi1,
|
|
|
|
constant int64_t & ne00,
|
|
|
|
constant int64_t & ne01,
|
|
|
|
constant int64_t & ne02,
|
|
|
|
constant uint64_t & nb00,
|
|
|
|
constant uint64_t & nb01,
|
|
|
|
constant uint64_t & nb02,
|
|
|
|
constant int64_t & ne10,
|
|
|
|
constant int64_t & ne11,
|
|
|
|
constant int64_t & ne12,
|
|
|
|
constant int64_t & ne13,
|
|
|
|
constant uint64_t & nb10,
|
|
|
|
constant uint64_t & nb11,
|
|
|
|
constant uint64_t & nb12,
|
|
|
|
constant int64_t & ne0,
|
|
|
|
constant int64_t & ne1,
|
|
|
|
constant uint64_t & nb1,
|
|
|
|
constant uint & r2,
|
|
|
|
constant uint & r3,
|
|
|
|
constant int & idx,
|
|
|
|
device const char * src00,
|
|
|
|
device const char * src01,
|
|
|
|
device const char * src02,
|
|
|
|
device const char * src03,
|
|
|
|
device const char * src04,
|
|
|
|
device const char * src05,
|
|
|
|
device const char * src06,
|
|
|
|
device const char * src07,
|
|
|
|
threadgroup int8_t * shared_values [[threadgroup(0)]],
|
|
|
|
uint3 tgpig[[threadgroup_position_in_grid]],
|
|
|
|
uint tiitg[[thread_index_in_threadgroup]],
|
|
|
|
uint tiisg[[thread_index_in_simdgroup]],
|
|
|
|
uint sgitg[[simdgroup_index_in_threadgroup]]) {
|
|
|
|
device const char * src0[8] = {src00, src01, src02, src03, src04, src05, src06, src07};
|
|
|
|
|
|
|
|
const int64_t bid = tgpig.z/(ne12*ne13);
|
|
|
|
|
|
|
|
tgpig.z = tgpig.z%(ne12*ne13);
|
|
|
|
|
|
|
|
const int32_t id = ((device int32_t *) (ids + bid*nbi1))[idx];
|
|
|
|
|
|
|
|
kernel_mul_mv_iq3_xxs_f32_impl(
|
|
|
|
src0[id],
|
|
|
|
(device const float *) (src1 + bid*nb11),
|
|
|
|
dst + bid*ne0,
|
|
|
|
ne00,
|
|
|
|
ne01,
|
|
|
|
ne02,
|
|
|
|
ne10,
|
|
|
|
ne12,
|
|
|
|
ne0,
|
|
|
|
ne1,
|
|
|
|
r2,
|
|
|
|
r3,
|
|
|
|
shared_values,
|
|
|
|
tgpig,
|
|
|
|
tiisg,
|
|
|
|
sgitg);
|
|
|
|
}
|