Skip to content

HIP: add GGML_HIP_MMQ_MFMA option to allow disableing the MFMA path. #14930

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

Merged
merged 2 commits into from
Jul 29, 2025
Merged

HIP: add GGML_HIP_MMQ_MFMA option to allow disableing the MFMA path. #14930

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
8d19ff4
HIP: add GGML_HIP_MMQ_MFMA option to allow disableing the MFMA path. …
IMbackK Jul 28, 2025
3edd0ed
Reword the option description
IMbackK Jul 28, 2025
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
1 change: 1 addition & 0 deletions ggml/CMakeLists.txt
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -174,6 +174,7 @@ option(GGML_HIP_GRAPHS "ggml: use HIP graph, experimental,
option(GGML_HIP_NO_VMM "ggml: do not try to use HIP VMM" ON)
option(GGML_HIP_ROCWMMA_FATTN "ggml: enable rocWMMA for FlashAttention" OFF)
option(GGML_HIP_FORCE_ROCWMMA_FATTN_GFX12 "ggml: enable rocWMMA FlashAttention on GFX12" OFF)
option(GGML_HIP_MMQ_MFMA "ggml: enable MFMA MMA for CDNA in MMQ" ON)
option(GGML_MUSA_GRAPHS "ggml: use MUSA graph, experimental, unstable" OFF)
option(GGML_MUSA_MUDNN_COPY "ggml: enable muDNN for accelerated copy" OFF)
option(GGML_VULKAN "ggml: use Vulkan" OFF)
Expand Down
8 changes: 6 additions & 2 deletions ggml/src/ggml-cuda/common.cuh
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -227,7 +227,7 @@ typedef float2 dfloat2;
#define FP16_MMA_AVAILABLE
#endif // defined(GGML_HIP_ROCWMMA_FATTN) && (defined(CDNA) || defined(RDNA3) || (defined(GGML_HIP_ROCWMMA_FATTN_GFX12) && defined(RDNA4)))

#if defined(GGML_USE_HIP) && defined(__HIP_PLATFORM_AMD__) && defined(CDNA3)
#if defined(GGML_USE_HIP) && defined(__HIP_PLATFORM_AMD__) && defined(CDNA3) && !defined(GGML_HIP_NO_MMQ_MFMA)
#define AMD_MFMA_AVAILABLE
#endif // defined(GGML_USE_HIP) && defined(__HIP_PLATFORM_AMD__) && defined(CDNA3)

Expand Down Expand Up @@ -295,7 +295,11 @@ static bool fp32_mma_hardware_available(const int cc) {

// AMD CDNA3 matrix cores.. Will add support for other CDNA generations later.
static bool amd_mfma_available(const int cc) {
return cc >= GGML_CUDA_CC_OFFSET_AMD && GGML_CUDA_CC_IS_CDNA3(cc);
#if !defined(GGML_HIP_NO_MMQ_MFMA)
return GGML_CUDA_CC_IS_CDNA3(cc);
#else
return false;
#endif //!defined(GGML_HIP_NO_MMQ_MFMA)
}

// Volta technically had FP16 tensor cores but they work very differently compared to Turing and later.
Expand Down
4 changes: 4 additions & 0 deletions ggml/src/ggml-hip/CMakeLists.txt
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -113,6 +113,10 @@ if (GGML_HIP_ROCWMMA_FATTN)
add_compile_definitions(GGML_HIP_ROCWMMA_FATTN)
endif()

if (NOT GGML_HIP_MMQ_MFMA)
add_compile_definitions(GGML_HIP_NO_MMQ_MFMA)
endif()

if (GGML_HIP_FORCE_ROCWMMA_FATTN_GFX12 OR ${hip_VERSION} VERSION_GREATER_EQUAL 7.0)
add_compile_definitions(GGML_HIP_ROCWMMA_FATTN_GFX12)
endif()
Expand Down
Loading