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

CUDA: tuned mul_mat_q kernels (#2546)

Browse Source
This commit is contained in:
Johannes Gäßler 2023-08-09 09:42:34 +02:00 committed by GitHub
parent f5bfea0580
commit 25d43e0eb5
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
3 changed files with 676 additions and 386 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
Makefile
View File

@ -253,11 +253,6 @@ ifdef LLAMA_CUDA_KQUANTS_ITER
else else
NVCCFLAGS += -DK_QUANTS_PER_ITERATION=2 NVCCFLAGS += -DK_QUANTS_PER_ITERATION=2
endif endif
ifdef LLAMA_CUDA_MMQ_Y
NVCCFLAGS += -DGGML_CUDA_MMQ_Y=$(LLAMA_CUDA_MMQ_Y)
else
NVCCFLAGS += -DGGML_CUDA_MMQ_Y=64
endif # LLAMA_CUDA_MMQ_Y
#ifdef LLAMA_CUDA_CUBLAS #ifdef LLAMA_CUDA_CUBLAS
# NVCCFLAGS += -DGGML_CUDA_CUBLAS # NVCCFLAGS += -DGGML_CUDA_CUBLAS
#endif # LLAMA_CUDA_CUBLAS #endif # LLAMA_CUDA_CUBLAS

1
README.md
View File

@ -406,7 +406,6 @@ Building the program with BLAS support may lead to some performance improvements
---> --->
| Option | Legal values | Default | Description | | Option | Legal values | Default | Description |
|-------------------------|------------------------|---------|-------------| |-------------------------|------------------------|---------|-------------|
| LLAMA_CUDA_MMQ_Y | Positive integer >= 32 | 64 | Tile size in y direction when using the custom CUDA kernels for prompt processing. Higher values can be faster depending on the amount of shared memory available. Power of 2 heavily recommended. |
| LLAMA_CUDA_FORCE_DMMV | Boolean | false | Force the use of dequantization + matrix vector multiplication kernels instead of using kernels that do matrix vector multiplication on quantized data. By default the decision is made based on compute capability (MMVQ for 6.1/Pascal/GTX 1000 or higher). Does not affect k-quants. | | LLAMA_CUDA_FORCE_DMMV | Boolean | false | Force the use of dequantization + matrix vector multiplication kernels instead of using kernels that do matrix vector multiplication on quantized data. By default the decision is made based on compute capability (MMVQ for 6.1/Pascal/GTX 1000 or higher). Does not affect k-quants. |
| LLAMA_CUDA_DMMV_X | Positive integer >= 32 | 32 | Number of values in x direction processed by the CUDA dequantization + matrix vector multiplication kernel per iteration. Increasing this value can improve performance on fast GPUs. Power of 2 heavily recommended. Does not affect k-quants. | | LLAMA_CUDA_DMMV_X | Positive integer >= 32 | 32 | Number of values in x direction processed by the CUDA dequantization + matrix vector multiplication kernel per iteration. Increasing this value can improve performance on fast GPUs. Power of 2 heavily recommended. Does not affect k-quants. |
| LLAMA_CUDA_MMV_Y | Positive integer | 1 | Block size in y direction for the CUDA mul mat vec kernels. Increasing this value can improve performance on fast GPUs. Power of 2 recommended. Does not affect k-quants. | | LLAMA_CUDA_MMV_Y | Positive integer | 1 | Block size in y direction for the CUDA mul mat vec kernels. Increasing this value can improve performance on fast GPUs. Power of 2 recommended. Does not affect k-quants. |

980
ggml-cuda.cu
View File

File diff suppressed because it is too large Load Diff