perf: pad small attention heads to 64 for MMA Flash Attention (#1992)

This commit is contained in:
leejet authored and GitHub committed 2026-09-18 23:59:58 +08:00
1 parent 656a1354c3
commit adcac69650
5 files changed
+95

No files matched your search

+5
View File
@@ -343,6 +343,11 @@ add_subdirectory(thirdparty)
target_sources(${SD_LIB} PRIVATE $<TARGET_OBJECTS:zip>)
target_link_libraries(${SD_LIB} PUBLIC ggml)
target_link_libraries(${SD_LIB} PRIVATE onig sd-utf8proc)
if (SD_CUDA)
find_package(CUDAToolkit REQUIRED)
target_link_libraries(${SD_LIB} PRIVATE CUDA::cuda_driver)
set_property(SOURCE src/core/ggml_extend_backend.cpp APPEND PROPERTY COMPILE_DEFINITIONS SD_USE_CUDA)
endif()
target_include_directories(${SD_LIB} PUBLIC . src include)
target_include_directories(${SD_LIB} PRIVATE src/core)
target_include_directories(${SD_LIB} PUBLIC . thirdparty)