Skip to content

Add option to disable MMA support on Turing (again) #15468

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
wants to merge 1 commit into from
Closed

Add option to disable MMA support on Turing (again) #15468

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 2 additions & 0 deletions ggml/CMakeLists.txt
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -158,6 +158,8 @@ option(GGML_CUDA "ggml: use CUDA"
option(GGML_MUSA "ggml: use MUSA" OFF)
option(GGML_CUDA_FORCE_MMQ "ggml: use mmq kernels instead of cuBLAS" OFF)
option(GGML_CUDA_FORCE_CUBLAS "ggml: always use cuBLAS instead of mmq kernels" OFF)
option(GGML_CUDA_NO_TURING_MMA "ggml: disable the use of mma in mmq kernels" OFF)

set (GGML_CUDA_PEER_MAX_BATCH_SIZE "128" CACHE STRING
"ggml: max. batch size for using peer access")
option(GGML_CUDA_NO_PEER_COPY "ggml: do not use peer to peer copies" OFF)
Expand Down
5 changes: 5 additions & 0 deletions ggml/src/ggml-cuda/CMakeLists.txt
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -77,6 +77,11 @@ if (CUDAToolkit_FOUND)
add_compile_definitions(GGML_CUDA_FORCE_CUBLAS)
endif()

if (GGML_CUDA_NO_TURING_MMA)
add_compile_definitions(GGML_CUDA_NO_TURING_MMA)
add_compile_definitions(GGML_CUDA_FORCE_MMQ)
endif()

if (GGML_CUDA_NO_VMM)
add_compile_definitions(GGML_CUDA_NO_VMM)
endif()
Expand Down
4 changes: 4 additions & 0 deletions ggml/src/ggml-cuda/ggml-cuda.cu
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -71,6 +71,10 @@
#include <string>
#include <vector>

#ifdef GGML_CUDA_NO_TURING_MMA
#define CUBLAS_COMPUTE_16F CUBLAS_COMPUTE_16F_PEDANTIC
#endif

static_assert(sizeof(half) == sizeof(ggml_fp16_t), "wrong fp16 size");

[[noreturn]]
Expand Down
3 changes: 3 additions & 0 deletions ggml/src/ggml-cuda/mmq.cuh
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -9,6 +9,9 @@

using namespace ggml_cuda_mma;

#ifdef GGML_CUDA_NO_TURING_MMA
#undef TURING_MMA_AVAILABLE
#endif
#define MMQ_DP4A_MAX_BATCH_SIZE 64 // Max. batch size to use for dp4a MMQ kernels when FP16 tensor cores are available.
#define MMQ_ITER_K 256
#define MMQ_NWARPS 8
Expand Down