Mirrors/llama.cpp
1
0
Fork 0
mirror of https://github.com/ggerganov/llama.cpp.git synced 2024-12-26 06:10:29 +01:00
Code Issues Packages Projects Releases Wiki Activity

1.5 bit: we can do even better (#5999)

Browse Source 
* iq1_s: we can do even better

Spent one of the 4 scale bits on a signs of a 0.125 shift.
I.e., quants are now -1 + delta, delta, 1 + delta, where delta
is +/- 0.125.

CUDA works, same performance as before.
PPL(LLaMA-v2-7B) is now 11.85!

* iq1_s: make scalar and AVX2 work with the new version

* iq1_s: make Neon work with new version.

~10% drop in performance, so will need some more work.

* iq1_s: make Metal work with new version

* iq1_s: very slightly faster dequantize on Metal

* iq1_s: fix dequantize on the CPU

---------

Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
This commit is contained in:
Kawrakow 2024-03-11 16:53:15 +01:00 committed by GitHub
parent 05b06210c9
commit 44ca159faf
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
4 changed files with 83 additions and 55 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

1
ggml-common.h
View File

@ -645,6 +645,7 @@ GGML_TABLE_BEGIN(uint32_t, iq3s_grid, 512)
GGML_TABLE_END() GGML_TABLE_END()
#define NGRID_IQ1S 2048 #define NGRID_IQ1S 2048
#define IQ1S_DELTA 0.125f
#if defined(GGML_COMMON_IMPL_C) #if defined(GGML_COMMON_IMPL_C)
GGML_TABLE_BEGIN(uint64_t, iq1s_grid, NGRID_IQ1S) GGML_TABLE_BEGIN(uint64_t, iq1s_grid, NGRID_IQ1S)
0xffffffffffffffff, 0xffffffffffffff01, 0xffffffffffff0000, 0xffffffffffff01ff, 0xffffffffffffffff, 0xffffffffffffff01, 0xffffffffffff0000, 0xffffffffffff01ff,

34
ggml-cuda.cu
View File

@ -1722,22 +1722,15 @@ static __global__ void dequantize_block_iq1_s(const void * __restrict__ vx, dst_
const int il = tid/8; // 0...3 const int il = tid/8; // 0...3
const int ib = tid%8; // 0...7 const int ib = tid%8; // 0...7
dst_t * y = yy + i*QK_K + 32*ib + 8*il; dst_t * y = yy + i*QK_K + 32*ib + 8*il;
const float d = (float)x[i].d * (2*((x[i].qh[ib] >> 12) & 0xf) + 1); const float delta = x[i].qh[ib] & 0x8000 ? -1 - IQ1S_DELTA : -1 + IQ1S_DELTA;
#if __CUDA_ARCH__ >= MIN_CC_DP4A // lowest compute capability for integer intrinsics const float d = (float)x[i].d * (2*((x[i].qh[ib] >> 12) & 7) + 1);
int grid32[2]; const int8_t * q = (const int8_t *)grid32; uint32_t grid32[2]; const int8_t * q = (const int8_t *)grid32;
grid32[0] = *((const int *)(iq1s_grid_gpu + (x[i].qs[4*ib+il] | (((x[i].qh[ib] >> 3*il) & 7) << 8)))); grid32[0] = iq1s_grid_gpu[x[i].qs[4*ib+il] | (((x[i].qh[ib] >> 3*il) & 7) << 8)];
grid32[1] = __vsub4((grid32[0] >> 4) & 0x0f0f0f0f, 0x01010101); grid32[1] = (grid32[0] >> 4) & 0x0f0f0f0f;
grid32[0] = __vsub4(grid32[0] & 0x0f0f0f0f, 0x01010101); grid32[0] &= 0x0f0f0f0f;
for (int j = 0; j < 8; ++j) { for (int j = 0; j < 8; ++j) {
y[j] = d * q[j]; y[j] = d * (q[j] + delta);
} }
#else
const uint8_t * grid = (const uint8_t *)(iq1s_grid_gpu + (x[i].qs[4*ib+il] | (((x[i].qh[ib] >> 3*il) & 7) << 8)));
for (int j = 0; j < 4; ++j) {
y[j+0] = d * ((grid[j] & 0xf) - 1);
y[j+4] = d * ((grid[j] >> 4) - 1);
}
#endif
#else #else
assert(false); assert(false);
#endif #endif
@ -4560,8 +4553,8 @@ static __device__ __forceinline__ float vec_dot_iq1_s_q8_1(
const int * q8 = (const int *)bq8_1[ib32].qs; const int * q8 = (const int *)bq8_1[ib32].qs;
for (int l = 0; l < 4; ++l) { for (int l = 0; l < 4; ++l) {
const int * grid = (const int *)(iq1s_grid_gpu + (bq1->qs[4*ib32+l] | (((bq1->qh[ib32] >> 3*l) & 7) << 8))); const int * grid = (const int *)(iq1s_grid_gpu + (bq1->qs[4*ib32+l] | (((bq1->qh[ib32] >> 3*l) & 7) << 8)));
int grid0 = __vsub4(grid[0] & 0x0f0f0f0f, 0x01010101); int grid0 = grid[0] & 0x0f0f0f0f;
int grid1 = __vsub4((grid[0] >> 4) & 0x0f0f0f0f, 0x01010101); int grid1 = (grid[0] >> 4) & 0x0f0f0f0f;
sumi = __dp4a(q8[2*l+1], grid1, __dp4a(q8[2*l+0], grid0, sumi)); sumi = __dp4a(q8[2*l+1], grid1, __dp4a(q8[2*l+0], grid0, sumi));
} }
#else #else
@ -4569,13 +4562,16 @@ static __device__ __forceinline__ float vec_dot_iq1_s_q8_1(
for (int l = 0; l < 4; ++l) { for (int l = 0; l < 4; ++l) {
const uint8_t * grid = (const uint8_t *)(iq1s_grid_gpu + (bq1->qs[4*ib32+l] | (((bq1->qh[ib32] >> 3*l) & 7) << 8))); const uint8_t * grid = (const uint8_t *)(iq1s_grid_gpu + (bq1->qs[4*ib32+l] | (((bq1->qh[ib32] >> 3*l) & 7) << 8)));
for (int j = 0; j < 4; ++j) { for (int j = 0; j < 4; ++j) {
sumi += q8[j] * ((grid[j] & 0xf) - 1) + q8[j+4] * ((grid[j] >> 4) - 1); sumi += q8[j] * (grid[j] & 0xf) + q8[j+4] * (grid[j] >> 4);
} }
q8 += 8; q8 += 8;
} }
#endif #endif
const float d = (float)bq1->d * __low2float(bq8_1[ib32].ds); const float delta = bq1->qh[ib32] & 0x8000 ? -1-IQ1S_DELTA : -1+IQ1S_DELTA;
return d * sumi * (2*(bq1->qh[ib32] >> 12) + 1); const float d1q = (float)bq1->d * (2*((bq1->qh[ib32] >> 12) & 7) + 1);
const float d = d1q * __low2float (bq8_1[ib32].ds);
const float m = d1q * __high2float(bq8_1[ib32].ds);
return d * sumi + m * delta;
#else #else
assert(false); assert(false);
return 0.f; return 0.f;

18
ggml-metal.metal
View File

@ -4377,7 +4377,7 @@ void kernel_mul_mv_iq1_s_f32_impl(
+ yl[j+16] * (grid3[j] & 0xf) + yl[j+20] * (grid3[j] >> 4) + yl[j+16] * (grid3[j] & 0xf) + yl[j+20] * (grid3[j] >> 4)
+ yl[j+24] * (grid4[j] & 0xf) + yl[j+28] * (grid4[j] >> 4); + yl[j+24] * (grid4[j] & 0xf) + yl[j+28] * (grid4[j] >> 4);
} }
sumf[row] += (float)dh[0] * (sum - sumy) * (2*(qh[0] >> 12) + 1); sumf[row] += (float)dh[0] * (sum + sumy * (qh[0] & 0x8000 ? -1 - IQ1S_DELTA : -1 + IQ1S_DELTA)) * (2*((qh[0] >> 12) & 7) + 1);
dh += nb*sizeof(block_iq1_s)/2; dh += nb*sizeof(block_iq1_s)/2;
qs += nb*sizeof(block_iq1_s); qs += nb*sizeof(block_iq1_s);
@ -5076,14 +5076,16 @@ void dequantize_iq1_s(device const block_iq1_s * xb, short il, thread type4x4 &
const float d = xb->d; const float d = xb->d;
device const uint8_t * qs = xb->qs + 4*ib32 + 2*il; device const uint8_t * qs = xb->qs + 4*ib32 + 2*il;
device const uint16_t * qh = xb->qh; device const uint16_t * qh = xb->qh;
const float dl = d * (2*(qh[ib32] >> 12) + 1); const float dl = d * (2*((qh[ib32] >> 12) & 7) + 1);
constant uint8_t * grid1 = (constant uint8_t *)(iq1s_grid_gpu + (qs[0] | (((qh[ib32] >> (6*il+0)) & 7) << 8))); const float ml = dl * (qh[ib32] & 0x8000 ? -1 - IQ1S_DELTA : -1 + IQ1S_DELTA);
constant uint8_t * grid2 = (constant uint8_t *)(iq1s_grid_gpu + (qs[1] | (((qh[ib32] >> (6*il+3)) & 7) << 8))); const uint16_t h = qh[ib32] >> 6*il;
constant uint8_t * grid1 = (constant uint8_t *)(iq1s_grid_gpu + (qs[0] | ((h << 8) & 0x700)));
constant uint8_t * grid2 = (constant uint8_t *)(iq1s_grid_gpu + (qs[1] | ((h << 5) & 0x700)));
for (int i = 0; i < 4; ++i) { for (int i = 0; i < 4; ++i) {
reg[0][i] = dl * (grid1[i] & 0xf) - dl; reg[0][i] = dl * (grid1[i] & 0xf) + ml;
reg[1][i] = dl * (grid1[i] >> 4) - dl; reg[1][i] = dl * (grid1[i] >> 4) + ml;
reg[2][i] = dl * (grid2[i] & 0xf) - dl; reg[2][i] = dl * (grid2[i] & 0xf) + ml;
reg[3][i] = dl * (grid2[i] >> 4) - dl; reg[3][i] = dl * (grid2[i] >> 4) + ml;
} }
} }

81
ggml-quants.c
View File

@ -3456,11 +3456,12 @@ void dequantize_row_iq1_s(const block_iq1_s * restrict x, float * restrict y, in
const uint16_t * qh = x[i].qh; const uint16_t * qh = x[i].qh;
for (int ib = 0; ib < QK_K/32; ++ib) { for (int ib = 0; ib < QK_K/32; ++ib) {
const float dl = d * (2*(qh[ib] >> 12) + 1); const float dl = d * (2*((qh[ib] >> 12) & 7) + 1);
const float delta = qh[ib] & 0x8000 ? -IQ1S_DELTA : IQ1S_DELTA;
for (int l = 0; l < 4; ++l) { for (int l = 0; l < 4; ++l) {
const int8_t * grid = (const int8_t *)(iq1s_grid + (qs[l] | (((qh[ib] >> 3*l) & 7) << 8))); const int8_t * grid = (const int8_t *)(iq1s_grid + (qs[l] | (((qh[ib] >> 3*l) & 7) << 8)));
for (int j = 0; j < 8; ++j) { for (int j = 0; j < 8; ++j) {
y[j] = dl * grid[j]; y[j] = dl * (grid[j] + delta);
} }
y += 8; y += 8;
} }
@ -9582,7 +9583,7 @@ void ggml_vec_dot_iq1_s_q8_K (int n, float * restrict s, size_t bs, const void
const uint8_t * qs = x[i].qs; const uint8_t * qs = x[i].qs;
const uint16_t * qh = x[i].qh; const uint16_t * qh = x[i].qh;
int sumi1 = 0, sumi2 = 0; int sumi1 = 0, sumi2 = 0, sumi3 = 0;
for (int ib = 0; ib < QK_K/32; ib += 2) { for (int ib = 0; ib < QK_K/32; ib += 2) {
@ -9601,12 +9602,16 @@ void ggml_vec_dot_iq1_s_q8_K (int n, float * restrict s, size_t bs, const void
const int32x4_t p1 = ggml_vdotq_s32(ggml_vdotq_s32(vdupq_n_s32(0), q1b.val[0], q8b.val[0]), q1b.val[1], q8b.val[1]); const int32x4_t p1 = ggml_vdotq_s32(ggml_vdotq_s32(vdupq_n_s32(0), q1b.val[0], q8b.val[0]), q1b.val[1], q8b.val[1]);
const int32x4_t p2 = ggml_vdotq_s32(ggml_vdotq_s32(vdupq_n_s32(0), q1b.val[2], q8b.val[2]), q1b.val[3], q8b.val[3]); const int32x4_t p2 = ggml_vdotq_s32(ggml_vdotq_s32(vdupq_n_s32(0), q1b.val[2], q8b.val[2]), q1b.val[3], q8b.val[3]);
sumi1 += vaddvq_s32(p1) * (2*(qh[ib+0] >> 12) + 1); const int ls1 = 2*((qh[ib+0] >> 12) & 7) + 1;
sumi2 += vaddvq_s32(p2) * (2*(qh[ib+1] >> 12) + 1); const int ls2 = 2*((qh[ib+1] >> 12) & 7) + 1;
sumi1 += vaddvq_s32(p1) * ls1;
sumi2 += vaddvq_s32(p2) * ls2;
sumi3 += (y[i].bsums[2*ib+0] + y[i].bsums[2*ib+1]) * ls1 * (qh[ib+0] & 0x8000 ? -1 : 1)
+ (y[i].bsums[2*ib+2] + y[i].bsums[2*ib+3]) * ls2 * (qh[ib+1] & 0x8000 ? -1 : 1);
} }
sumf += y[i].d * GGML_FP16_TO_FP32(x[i].d) * (sumi1 + sumi2); sumf += y[i].d * GGML_FP16_TO_FP32(x[i].d) * (sumi1 + sumi2 + IQ1S_DELTA * sumi3);
} }
*s = sumf; *s = sumf;
@ -9614,6 +9619,7 @@ void ggml_vec_dot_iq1_s_q8_K (int n, float * restrict s, size_t bs, const void
#elif defined __AVX2__ #elif defined __AVX2__
__m256 accum = _mm256_setzero_ps(); __m256 accum = _mm256_setzero_ps();
float accum1 = 0;
for (int i = 0; i < nb; ++i) { for (int i = 0; i < nb; ++i) {
const int8_t * q8 = y[i].qs; const int8_t * q8 = y[i].qs;
@ -9621,6 +9627,7 @@ void ggml_vec_dot_iq1_s_q8_K (int n, float * restrict s, size_t bs, const void
const uint16_t * qh = x[i].qh; const uint16_t * qh = x[i].qh;
__m256i sumi = _mm256_setzero_si256(); __m256i sumi = _mm256_setzero_si256();
int sumi1 = 0;
for (int ib = 0; ib < QK_K/32; ib += 2) { for (int ib = 0; ib < QK_K/32; ib += 2) {
const __m256i q1b_1 = _mm256_set_epi64x(iq1s_grid[qs[3] | ((qh[ib+0] >> 1) & 0x700)], iq1s_grid[qs[2] | ((qh[ib+0] << 2) & 0x700)], const __m256i q1b_1 = _mm256_set_epi64x(iq1s_grid[qs[3] | ((qh[ib+0] >> 1) & 0x700)], iq1s_grid[qs[2] | ((qh[ib+0] << 2) & 0x700)],
iq1s_grid[qs[1] | ((qh[ib+0] << 5) & 0x700)], iq1s_grid[qs[0] | ((qh[ib+0] << 8) & 0x700)]); iq1s_grid[qs[1] | ((qh[ib+0] << 5) & 0x700)], iq1s_grid[qs[0] | ((qh[ib+0] << 8) & 0x700)]);
@ -9632,17 +9639,23 @@ void ggml_vec_dot_iq1_s_q8_K (int n, float * restrict s, size_t bs, const void
const __m256i dot1 = mul_add_epi8(q1b_1, q8b_1); const __m256i dot1 = mul_add_epi8(q1b_1, q8b_1);
const __m256i dot2 = mul_add_epi8(q1b_2, q8b_2); const __m256i dot2 = mul_add_epi8(q1b_2, q8b_2);
const __m256i p1 = _mm256_madd_epi16(dot1, _mm256_set1_epi16(2*(qh[ib+0] >> 12) + 1)); const int16_t ls1 = 2*((qh[ib+0] >> 12) & 7) + 1;
const __m256i p2 = _mm256_madd_epi16(dot2, _mm256_set1_epi16(2*(qh[ib+1] >> 12) + 1)); const int16_t ls2 = 2*((qh[ib+1] >> 12) & 7) + 1;
const __m256i p1 = _mm256_madd_epi16(dot1, _mm256_set1_epi16(ls1));
const __m256i p2 = _mm256_madd_epi16(dot2, _mm256_set1_epi16(ls2));
sumi = _mm256_add_epi32(sumi, _mm256_add_epi32(p1, p2)); sumi = _mm256_add_epi32(sumi, _mm256_add_epi32(p1, p2));
sumi1 += (y[i].bsums[2*ib+0] + y[i].bsums[2*ib+1]) * (qh[ib+0] & 0x8000 ? -1 : 1) * ls1
+ (y[i].bsums[2*ib+2] + y[i].bsums[2*ib+3]) * (qh[ib+1] & 0x8000 ? -1 : 1) * ls2;
} }
accum = _mm256_fmadd_ps(_mm256_set1_ps(y[i].d * GGML_FP16_TO_FP32(x[i].d)), _mm256_cvtepi32_ps(sumi), accum); const float d = y[i].d * GGML_FP16_TO_FP32(x[i].d);
accum = _mm256_fmadd_ps(_mm256_set1_ps(d), _mm256_cvtepi32_ps(sumi), accum);
accum1 += d * sumi1;
} }
*s = hsum_float_8(accum); *s = hsum_float_8(accum) + IQ1S_DELTA * accum1;
#else #else
@ -9653,9 +9666,10 @@ void ggml_vec_dot_iq1_s_q8_K (int n, float * restrict s, size_t bs, const void
const uint8_t * qs = x[i].qs; const uint8_t * qs = x[i].qs;
const uint16_t * qh = x[i].qh; const uint16_t * qh = x[i].qh;
int sumi = 0; int sumi = 0, sumi1 = 0;
for (int ib = 0; ib < QK_K/32; ++ib) { for (int ib = 0; ib < QK_K/32; ++ib) {
const int ls = 2*(qh[ib] >> 12) + 1; const int ls = 2*((qh[ib] >> 12) & 7) + 1;
const int delta = qh[ib] & 0x8000 ? -1 : 1;
int lsum = 0; int lsum = 0;
for (int l = 0; l < 4; ++l) { for (int l = 0; l < 4; ++l) {
const int8_t * grid = (const int8_t *)(iq1s_grid + (qs[l] | (((qh[ib] >> 3*l) & 7) << 8))); const int8_t * grid = (const int8_t *)(iq1s_grid + (qs[l] | (((qh[ib] >> 3*l) & 7) << 8)));
@ -9665,10 +9679,11 @@ void ggml_vec_dot_iq1_s_q8_K (int n, float * restrict s, size_t bs, const void
q8 += 8; q8 += 8;
} }
sumi += ls * lsum; sumi += ls * lsum;
sumi1 += ls * delta * (y[i].bsums[2*ib+0] + y[i].bsums[2*ib+1]);
qs += 4; qs += 4;
} }
sumf += GGML_FP16_TO_FP32(x[i].d) * y[i].d * sumi; sumf += GGML_FP16_TO_FP32(x[i].d) * y[i].d * (sumi + IQ1S_DELTA * sumi1);
} }
*s = sumf; *s = sumf;
@ -11438,7 +11453,7 @@ static int iq1_find_best_neighbour(const uint16_t * restrict neighbours, const u
} }
static int iq1_find_best_neighbour2(const uint16_t * restrict neighbours, const uint64_t * restrict grid, static int iq1_find_best_neighbour2(const uint16_t * restrict neighbours, const uint64_t * restrict grid,
const float * restrict xval, const float * restrict weight, float scale, int8_t * restrict L, int ngrid) { const float * restrict xval, const float * restrict weight, float scale, const float * restrict xg, int8_t * restrict L, int ngrid) {
int num_neighbors = neighbours[0]; int num_neighbors = neighbours[0];
GGML_ASSERT(num_neighbors > 0); GGML_ASSERT(num_neighbors > 0);
float best_score = FLT_MAX; float best_score = FLT_MAX;
@ -11447,7 +11462,7 @@ static int iq1_find_best_neighbour2(const uint16_t * restrict neighbours, const
const int8_t * pg = (const int8_t *)(grid + neighbours[j]); const int8_t * pg = (const int8_t *)(grid + neighbours[j]);
float d2 = 0; float d2 = 0;
for (int i = 0; i < 8; ++i) { for (int i = 0; i < 8; ++i) {
float q = (pg[i] - 3)/2; float q = xg[(pg[i] - 1)/2];
float w = weight[i]; float w = weight[i];
float diff = scale*q - xval[i]; float diff = scale*q - xval[i];
d2 += w*diff*diff; d2 += w*diff*diff;
@ -11463,7 +11478,7 @@ static int iq1_find_best_neighbour2(const uint16_t * restrict neighbours, const
float d2 = 0; float d2 = 0;
for (int j = 0; j < 8; ++j) { for (int j = 0; j < 8; ++j) {
float w = weight[j]; float w = weight[j];
float q = (grid_i[j] - 3)/2; float q = xg[(grid_i[j] - 1)/2];
float diff = scale*q - xval[i]; float diff = scale*q - xval[i];
d2 += w*diff*diff; d2 += w*diff*diff;
} }
@ -11480,7 +11495,7 @@ static int iq1_find_best_neighbour2(const uint16_t * restrict neighbours, const
const int8_t * pg = (const int8_t *)(grid + neighbours[j]); const int8_t * pg = (const int8_t *)(grid + neighbours[j]);
float sumqx = 0, sumq2 = 0; float sumqx = 0, sumq2 = 0;
for (int i = 0; i < 8; ++i) { for (int i = 0; i < 8; ++i) {
float q = (pg[i] - 3)/2; float q = xg[(pg[i] - 1)/2];
float w = weight[i]; float w = weight[i];
sumqx += w*q*xval[i]; sumqx += w*q*xval[i];
sumq2 += w*q*q; sumq2 += w*q*q;
@ -11519,6 +11534,9 @@ static void quantize_row_iq1_s_impl(const float * restrict x, void * restrict vy
block_iq1_s * y = vy; block_iq1_s * y = vy;
const float x_p[3] = {-1 + IQ1S_DELTA, IQ1S_DELTA, 1 + IQ1S_DELTA};
const float x_m[3] = {-1 - IQ1S_DELTA, -IQ1S_DELTA, 1 - IQ1S_DELTA};
float scales[QK_K/IQ1S_BLOCK_SIZE]; float scales[QK_K/IQ1S_BLOCK_SIZE];
float weight[IQ1S_BLOCK_SIZE]; float weight[IQ1S_BLOCK_SIZE];
int8_t L[IQ1S_BLOCK_SIZE]; int8_t L[IQ1S_BLOCK_SIZE];
@ -11527,6 +11545,7 @@ static void quantize_row_iq1_s_impl(const float * restrict x, void * restrict vy
float pairs[2*IQ1S_BLOCK_SIZE]; float pairs[2*IQ1S_BLOCK_SIZE];
int * idx = (int *)(pairs + 1); int * idx = (int *)(pairs + 1);
uint16_t index[IQ1S_BLOCK_SIZE/8]; uint16_t index[IQ1S_BLOCK_SIZE/8];
int8_t shifts[QK_K/IQ1S_BLOCK_SIZE];
for (int ibl = 0; ibl < nbl; ++ibl) { for (int ibl = 0; ibl < nbl; ++ibl) {
@ -11572,25 +11591,33 @@ static void quantize_row_iq1_s_impl(const float * restrict x, void * restrict vy
} }
} }
float best_score = 0, scale = max; float best_score = 0, scale = max;
int besti1 = 0, besti2 = 0; int besti1 = -1, besti2 = -1, best_shift = 0;
for (int i1 = 0; i1 <= IQ1S_BLOCK_SIZE; ++i1) { for (int i1 = 0; i1 <= IQ1S_BLOCK_SIZE; ++i1) {
for (int i2 = i1; i2 <= IQ1S_BLOCK_SIZE; ++i2) { for (int i2 = i1; i2 <= IQ1S_BLOCK_SIZE; ++i2) {
float sumqx = -(sumx[i1] - sumx[0]) + (sumx[IQ1S_BLOCK_SIZE] - sumx[i2]); float sumqx = (sumx[i1] - sumx[0])*x_p[0] + (sumx[i2] - sumx[i1])*x_p[1] + (sumx[IQ1S_BLOCK_SIZE] - sumx[i2])*x_p[2];
float sumq2 = (sumw[i1] - sumw[0]) + (sumw[IQ1S_BLOCK_SIZE] - sumw[i2]); float sumq2 = (sumw[i1] - sumw[0])*x_p[0]*x_p[0] + (sumw[i2] - sumw[i1])*x_p[1]*x_p[1] + (sumw[IQ1S_BLOCK_SIZE] - sumw[i2])*x_p[2]*x_p[2];
if (sumq2 > 0 && sumqx*sumqx > best_score*sumq2) { if (sumq2 > 0 && sumqx*sumqx > best_score*sumq2) {
scale = sumqx/sumq2; best_score = scale*sumqx; scale = sumqx/sumq2; best_score = scale*sumqx;
besti1 = i1; besti2 = i2; besti1 = i1; besti2 = i2; best_shift = 1;
}
sumqx = (sumx[i1] - sumx[0])*x_m[0] + (sumx[i2] - sumx[i1])*x_m[1] + (sumx[IQ1S_BLOCK_SIZE] - sumx[i2])*x_m[2];
sumq2 = (sumw[i1] - sumw[0])*x_m[0]*x_m[0] + (sumw[i2] - sumw[i1])*x_m[1]*x_m[1] + (sumw[IQ1S_BLOCK_SIZE] - sumw[i2])*x_m[2]*x_m[2];
if (sumq2 > 0 && sumqx*sumqx > best_score*sumq2) {
scale = sumqx/sumq2; best_score = scale*sumqx;
besti1 = i1; besti2 = i2; best_shift = -1;
} }
} }
} }
GGML_ASSERT(besti1 >= 0 && besti2 >= 0 && best_shift != 0);
for (int j = 0; j < besti1; ++j) L[idx[2*j]] = 0; for (int j = 0; j < besti1; ++j) L[idx[2*j]] = 0;
for (int j = besti1; j < besti2; ++j) L[idx[2*j]] = 1; for (int j = besti1; j < besti2; ++j) L[idx[2*j]] = 1;
for (int j = besti2; j < IQ1S_BLOCK_SIZE; ++j) L[idx[2*j]] = 2; for (int j = besti2; j < IQ1S_BLOCK_SIZE; ++j) L[idx[2*j]] = 2;
if (scale < 0) { if (scale < 0) {
for (int j = 0; j < IQ1S_BLOCK_SIZE; ++j) L[j] = 2 - L[j]; for (int j = 0; j < IQ1S_BLOCK_SIZE; ++j) L[j] = 2 - L[j];
scale = -scale; scale = -scale; best_shift = -best_shift;
} }
bool all_on_grid = true; bool all_on_grid = true;
const float * xx = best_shift == 1 ? x_p : x_m;
for (int k = 0; k < IQ1S_BLOCK_SIZE/8; ++k) { for (int k = 0; k < IQ1S_BLOCK_SIZE/8; ++k) {
uint16_t u = 0; uint16_t u = 0;
for (int j = 0; j < 8; ++j) u |= (L[8*k+j] << 2*j); for (int j = 0; j < 8; ++j) u |= (L[8*k+j] << 2*j);
@ -11598,7 +11625,7 @@ static void quantize_row_iq1_s_impl(const float * restrict x, void * restrict vy
if (grid_index < 0) { if (grid_index < 0) {
all_on_grid = false; all_on_grid = false;
const uint16_t * neighbours = kneighbors_q2xs - kmap_q2xs[u] - 1; const uint16_t * neighbours = kneighbors_q2xs - kmap_q2xs[u] - 1;
grid_index = iq1_find_best_neighbour2(neighbours, kgrid_q2xs, xb + 8*k, weight + 8*k, scale, L + 8*k, NGRID_IQ1S); grid_index = iq1_find_best_neighbour2(neighbours, kgrid_q2xs, xb + 8*k, weight + 8*k, scale, xx, L + 8*k, NGRID_IQ1S);
GGML_ASSERT(grid_index >= 0); GGML_ASSERT(grid_index >= 0);
} }
index[k] = grid_index; index[k] = grid_index;
@ -11609,7 +11636,7 @@ static void quantize_row_iq1_s_impl(const float * restrict x, void * restrict vy
const int8_t * pg = (const int8_t *)(kgrid_q2xs + index[k]); const int8_t * pg = (const int8_t *)(kgrid_q2xs + index[k]);
for (int j = 0; j < 8; ++j) { for (int j = 0; j < 8; ++j) {
float w = weight[8*k + j]; float w = weight[8*k + j];
float q = (pg[j] - 3)/2; float q = xx[(pg[j] - 1)/2];
sumqx += w*q*xb[8*k+j]; sumqx += w*q*xb[8*k+j];
sumq2 += w*q*q; sumq2 += w*q*q;
} }
@ -11624,6 +11651,7 @@ static void quantize_row_iq1_s_impl(const float * restrict x, void * restrict vy
y[ibl].qh[ib] = h; y[ibl].qh[ib] = h;
GGML_ASSERT(scale >= 0); GGML_ASSERT(scale >= 0);
scales[ib] = scale; scales[ib] = scale;
shifts[ib] = best_shift;
max_scale = MAX(max_scale, scale); max_scale = MAX(max_scale, scale);
} }
@ -11632,12 +11660,13 @@ static void quantize_row_iq1_s_impl(const float * restrict x, void * restrict vy
continue; continue;
} }
float d = max_scale/31; float d = max_scale/15;
y[ibl].d = GGML_FP32_TO_FP16(d*1.125f); // 1.085f is another fudge factor. Don't ask me why it is needed. y[ibl].d = GGML_FP32_TO_FP16(d*1.125f); // 1.085f is another fudge factor. Don't ask me why it is needed.
float id = 1/d; float id = 1/d;
for (int ib = 0; ib < QK_K/IQ1S_BLOCK_SIZE; ++ib) { for (int ib = 0; ib < QK_K/IQ1S_BLOCK_SIZE; ++ib) {
int l = nearest_int(0.5f*(id*scales[ib]-1)); int l = nearest_int(0.5f*(id*scales[ib]-1));
l = MAX(0, MIN(15, l)); l = MAX(0, MIN(7, l));
if (shifts[ib] == -1) l |= 8;
y[ibl].qh[ib] |= (l << 12); y[ibl].qh[ib] |= (l << 12);
} }
} }