* metal : improve decoding speed for batches of 2-16
* metal : rename kernels mul_mat_ to mul_mv_
* metal : indentations
* minor
* metal : print more GPU info + disable mul_mm for MTLGPUFamiliy < Apple7
* metal : relax conditions on fast matrix multiplication kernel
* metal : revert the concurrnecy change because it was wrong
* llama : remove experimental stuff
* Minor speed gains for all quantization types
* metal: faster kernel_scale via float4
* Various other speedups for "small" kernels
* metal: faster soft_max vial float4
* metal: faster diagonal infinity
Although, to me it looks like one should simply
fuse scale + diagnonal infinity + soft_max on the
KQtensor.
* Another faster f16 x f32 matrix multiply kernel
* Reverting the diag infinity change
It does work for PP, but somehow it fails for TG.
Need to look more into it.
* metal: add back faster diagonal infinity
This time more carefully
* metal : minor (readibility)
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
* Slightly faster Q3_K and Q5_K on metal
* Another Q3_K speedup on metal
Combined with previous commit, we are now +9.6% for TG.
PP is not affected as this happens via the matrix multiplication
templates.
* Slowly progressing on Q3_K on metal
We are now 13% faster than master
* nother small improvement for Q3_K on metal
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
* metal : fix kernel_norm
ggml-ci
* metal : put warning in kernel_norm to not combine the loops
* metal : restore original F16 mat-vec multiplication
It works after the norm fixes
* common : don't do warm-up with more than n_batch tokens (close#3058)
ggml-ci
* metal : minor
* Very minor speedup via simd-group synchronization in f16 x f32
* Another very minor speedup on metal
* Quite significant PP speedup on metal
* Another attempt
* Minor
* Massive improvement for TG for fp16
* ~4-5% improvement for Q8_0 TG on metal
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
* Somewhat faster f16 x f32 matrix multiply kernel
* Better use 32 thread groups for f16 x f32
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
* metal: matrix-matrix multiplication kernel
This commit removes MPS and uses custom matrix-matrix multiplication
kernels for all quantization types. This commit also adds grouped-query
attention to support llama2 70B.
* metal: fix performance degradation from gqa
Integers are slow on the GPU, and 64-bit divides are extremely slow.
In the context of GQA, we introduce a 64-bit divide that cannot be
optimized out by the compiler, which results in a decrease of ~8% in
inference performance. This commit fixes that issue by calculating a
part of the offset with a 32-bit divide. Naturally, this limits the
size of a single matrix to ~4GB. However, this limitation should
suffice for the near future.
* metal: fix bugs for GQA and perplexity test.
I mixed up ne02 and nb02 in previous commit.
* Another speed gain for Q4_0 and Q4_1 on Metal
* Have N_DST, etc., be template parameters
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
* Faster Q2_K on Metal
* Deleting unnoticed and dangereous trailing white space
* Fixed bug in new metal Q2_K implementation
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
* metal: use uint16_t instead of uint8_t.
Apple GPU doesn't like uint8_t. For every operation on uint8_t
the gpu need to copy the uint8_t to an empty 16 bit register, then
it can issue other instructions.
For the matrix-vector multiplication kernel only, we observed a
340~350 GB/s memory read speed on M1 Max after this commit, which is
very close to the reported hardware limit.
* metal: update rms_norm kernel
This commit double the speed of rms_norm operations by using 512 threads
per threadgroup, combining with SIMD primitives to minimize the need for
thread group barriers.
* metal: use template to reduce size
Revert modifications on block_q4_0 and block_q4_1.
* Implement customizable RoPE
The original RoPE has pre-defined parameters
theta_i = 10000^(−2(i−1)/d), for i in [1, 2, ..., d/2]
Our customizable RoPE, ggml_rope_custom_inplace, uses
theta_i = scale * base^(−2(i−1)/d), for i in [1, 2, ..., d/2]
with the default matches the original
scale = 1.0
base = 10000
The new command line arguments
--rope-freq-base
--rope-freq-scale
set the two new RoPE parameter.
Recent researches show changing these two parameters extends the context limit with minimal loss.
1. Extending Context to 8K
kaiokendev
https://kaiokendev.github.io/til#extending-context-to-8k
2. Extending Context Window of Large Language Models via Positional Interpolation
Shouyuan Chen, Sherman Wong, Liangjian Chen, Yuandong Tian
https://arxiv.org/abs/2306.15595
3. NTK-Aware Scaled RoPE allows LLaMA models to have extended (8k+) context size without any fine-tuning and minimal perplexity degradation.
https://www.reddit.com/user/bloc97https://www.reddit.com/r/LocalLLaMA/comments/14lz7j5/ntkaware_scaled_rope_allows_llama_models_to_have/
For the bold, try adding the following command line parameters to your favorite model:
-c 16384 --rope-freq-base 80000 --rope-freq-scale 0.5
* ggml-metal: fix custom rope
* common: fix argument names in help
* llama: increase MEM_REQ_EVAL for MODEL_3B
It avoids crashing for quantized weights on CPU.
Better ways to calculate the required buffer size would be better.
* llama: make MEM_REQ_EVAL depend on n_ctx
* server: use proper Content-Type in curl examples
Without the header Content-Type: application/json, curl will POST with
Content-Type: application/x-www-form-urlencoded
Though our simple server doesn't care, the httplib.h used has a limit
with CPPHTTPLIB_FORM_URL_ENCODED_PAYLOAD_MAX_LENGTH 8192
With Content-Type: application/json, we can send large json data.
* style : minor fixes, mostly indentations
* ggml : fix asserts
---------
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
* 3-5% faster Q4_0 on Metal
* 7-25% faster Q4_1 on Metal
* Oops, forgot to delete the original Q4_1 kernel
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
* 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>
* 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>
* metal : 8% faster q4_0
Avoid copying into local uchar4 anf float4.
* metal : 17% faster Q4_0
Use 64 threads in a thread group.
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
* metal : add Q2_K implementation
27.1 ms / token on M2 Max 30-core GPU, so about the
same speed as Q4_0. Memory throughput is ~156 GB/s.
The access pattern used in the Q2_K
CUDA implementation resulted in significantly lower
performance (~31 ms/token).
* Fixing merge conflicts
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
* Metal implementation for Q4_K
Very slow for now:
42 ms / token, Q4_0 runs in 28 ms/token on my
30-core M2 Max GPU.
* Optimizing Q4_K on metal
The first token always takes longer, I guess because
the metal kernel is being jit-compiled.
So, using n = 128 to measure time.
At this point Q4_K takes 29.5 ms / token
compared to 27.2 ms / token for Q4_0.
Quite a bit better than the initial attempt,
but still not good enough.
* Optimizing q4_K metal dot some more
For n = 256 it is now 28.1 ms/token compared to
27 ms/token for q4_0.
* Fix after merge with master
* Metal implementation for Q6_K
Similar to the CUDA implementation.
No idea if this is the optimum for Metal, but the few
alternative variants I tried all had a lower performance.
We get 36.5 ms / token on M2 Max with 30 GPU cores.
This corresponds to ~200 GB/second throughput.
* clang-tidy : add config back
* Much better Q6_K implementation for metal
28.3 ms / token for 7B. Subtracting ~9 ms that is spent in
other compute graph operations, we are left with ~19 ms
for the matrix multiplications. The model is ~5.5 GB,
so we are getting 1000 / 19 * 5.5 = 290 GB/s!
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
* Metal implementation for Q4_K
Very slow for now:
42 ms / token, Q4_0 runs in 28 ms/token on my
30-core M2 Max GPU.
* Optimizing Q4_K on metal
The first token always takes longer, I guess because
the metal kernel is being jit-compiled.
So, using n = 128 to measure time.
At this point Q4_K takes 29.5 ms / token
compared to 27.2 ms / token for Q4_0.
Quite a bit better than the initial attempt,
but still not good enough.
* Optimizing q4_K metal dot some more
For n = 256 it is now 28.1 ms/token compared to
27 ms/token for q4_0.
* Fix after merge with master
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
* mtl : export the LLaMA computation graph
* ci : disable temporary
* mtl : adapt the MNIST example as starter
* mtl : no need for mtl-export tool, add cli arg for main instead
* mtl : export just a small part of the graph for now to make it easier
* mtl : move MSL code into separate file for easy editing
* mtl : initial get_rows_q4_0 kernel
* mtl : confirmed get_rows_q4_0 is working correctly
* mtl : add rms_norm kernel + confirm working
* mtl : add mul kernel + confirm working
* mtl : initial mul_mat Q4 kernel (wrong results)
* mtl : mul_mat fixes (still wrong)
* mtl : another mul_mat Q4 (still does not work)
* mtl : working mul_mat q4
* ggml : fix handling of "view" ops in ggml_graph_import()
* mtl : add rope kernel
* mtl : add reshape and transpose handling
* ggml : store offset as opt arg for ggml_view_xd() operators
* mtl : add cpy kernel + handle view ops
* mtl : confirm f16 x f32 attention mul mat
* mtl : add scale kernel
* mtl : add diag_mask_inf kernel
* mtl : fix soft_max kernel
* ggml : update ggml_nbytes() to handle non-contiguous tensors
* mtl : verify V tensor contents
* mtl : add f32 -> f32 cpy kernel
* mtl : add silu kernel
* mtl : add non-broadcast mul kernel
* mtl : full GPU inference of the computation graph
* mtl : optimize rms_norm and soft_max kernels
* mtl : add f16 mat x f32 vec multiplication kernel
* mtl : fix bug in f16 x f32 mul mat + speed-up computation
* mtl : faster mul_mat_q4_0_f32 kernel
* mtl : fix kernel signature + roll inner loop
* mtl : more threads for rms_norm + better timing
* mtl : remove printfs from inner loop
* mtl : simplify implementation
* mtl : add save/load vocab to ggml file
* mtl : plug Metal inference into llama.cpp (very quick-n-dirty)
* mtl : make it work with main example
Lots of hacks but at least now it generates text
* mtl : preparing for merge
* mtl : clean-up ggml mtl interface + suport scratch / inplace
* mtl : remove temp / debug code
* metal : final refactoring and simplification
* Revert "ci : disable temporary"
This reverts commit 98c267fc77.
* metal : add comments
* metal : clean-up stuff, fix typos
* readme : add Metal instructions
* readme : add example for main
