diff options
| author | Kawrakow <48489457+ikawrakow@users.noreply.github.com> | 2023-06-16 20:08:44 +0300 |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2023-06-16 20:08:44 +0300 |
| commit | 3d0112261042b356621e93db3fa4c6798a5d098f (patch) | |
| tree | 3634baa70ed23142f86c5a44701bbf4b0971c2fd /Makefile | |
| parent | 602c748863e15270d80d74aa2c3bf86ab8139e07 (diff) | |
CUDA : faster k-quant dot kernels (#1862)
* cuda : faster k-quant dot kernels
* Imrove Q2_K dot kernel on older GPUs
We now have a K_QUANTS_PER_ITERATION macro, which should be
set to 1 on older and to 2 on newer GPUs.
With this, we preserve the performance of the original
PR on RTX-4080, and are faster compared to master on
GTX-1660.
* Imrove Q6_K dot kernel on older GPUs
Using the same K_QUANTS_PER_ITERATION macro as last commit,
we preserve performance on RTX-4080 and speed up
Q6_K on a GTX-1660.
* Add LLAMA_CUDA_KQUANTS_ITER to CMakeLists.txt and Makefile
Allowed values are 1 or 2. 2 gives the best performance on
modern GPUs and is set as default. On older GPUs 1 may work
better.
* PR comments
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
Diffstat (limited to 'Makefile')
| -rw-r--r-- | Makefile | 5 |
1 files changed, 5 insertions, 0 deletions
@@ -171,6 +171,11 @@ ifdef LLAMA_CUDA_DMMV_Y else NVCCFLAGS += -DGGML_CUDA_DMMV_Y=1 endif # LLAMA_CUDA_DMMV_Y +ifdef LLAMA_CUDA_KQUANTS_ITER + NVCCFLAGS += -DK_QUANTS_PER_ITERATION=$(LLAMA_CUDA_KQUANTS_ITER) +else + NVCCFLAGS += -DK_QUANTS_PER_ITERATION=2 +endif ggml-cuda.o: ggml-cuda.cu ggml-cuda.h $(NVCC) $(NVCCFLAGS) $(CXXFLAGS) -Wno-pedantic -c $< -o $@ endif # LLAMA_CUBLAS |