alihan/whisper.cpp
1
0
Fork 0
mirror of https://github.com/ggerganov/whisper.cpp.git synced 2023-11-04 02:52:44 +03:00
Code Issues Packages Projects Releases Wiki Activity

whisper : add GPU support via cuBLAS (#834)

Browse Source 
* make : add WHISPER_CUBLAS

* make : fix CUBLAS build

* whisper : disable Flash Attention + adjust memory buffers

* whisper : remove old commented code

* readme : add cuBLAS instructions

* cmake : add WHISPER_CUBLAS option

* gitignore : ignore build-cublas
This commit is contained in:
Georgi Gerganov
2023-04-30 12:14:33 +03:00
committed by GitHub
parent 0ccd6746c9
commit 5fd1bdd7fc
10 changed files with 97 additions and 46 deletions
Download Patch File Download Diff File Expand all files Collapse all files

39
CMakeLists.txt
View File

@@ -51,7 +51,7 @@ option(WHISPER_SANITIZE_UNDEFINED "whisper: enable undefined sanitizer" OFF)
option(WHISPER_BUILD_TESTS "whisper: build tests" ${WHISPER_STANDALONE})
option(WHISPER_BUILD_EXAMPLES "whisper: build examples" ${WHISPER_STANDALONE})
option(WHISPER_SUPPORT_SDL2 "whisper: support for libSDL2" OFF)
option(WHISPER_SDL2 "whisper: support for libSDL2" OFF)
if (APPLE)
option(WHISPER_NO_ACCELERATE "whisper: disable Accelerate framework" OFF)
@@ -62,7 +62,8 @@ if (APPLE)
option(WHISPER_COREML "whisper: enable Core ML framework" OFF)
option(WHISPER_COREML_ALLOW_FALLBACK "whisper: allow non-CoreML fallback" OFF)
else()
option(WHISPER_SUPPORT_OPENBLAS "whisper: support for OpenBLAS" OFF)
option(WHISPER_OPENBLAS "whisper: support for OpenBLAS" OFF)
option(WHISPER_CUBLAS "whisper: support for cuBLAS" OFF)
endif()
option(WHISPER_PERF "whisper: enable perf timings" OFF)
@@ -127,7 +128,7 @@ if (APPLE)
endif()
endif()
if (WHISPER_SUPPORT_OPENBLAS)
if (WHISPER_OPENBLAS)
find_library(OPENBLAS_LIB
NAMES openblas libopenblas
)
@@ -141,6 +142,31 @@ if (WHISPER_SUPPORT_OPENBLAS)
endif()
endif()
if (WHISPER_CUBLAS)
cmake_minimum_required(VERSION 3.17)
find_package(CUDAToolkit)
if (CUDAToolkit_FOUND)
message(STATUS "cuBLAS found")
enable_language(CUDA)
set(GGML_CUDA_SOURCES ggml-cuda.cu ggml-cuda.h)
add_compile_definitions(GGML_USE_CUBLAS)
if (WHISPER_STATIC)
set(WHISPER_EXTRA_LIBS ${WHISPER_EXTRA_LIBS} CUDA::cudart_static CUDA::cublas_static CUDA::cublasLt_static)
else()
set(WHISPER_EXTRA_LIBS ${WHISPER_EXTRA_LIBS} CUDA::cudart CUDA::cublas CUDA::cublasLt)
endif()
else()
message(WARNING "cuBLAS not found")
endif()
endif()
# compiler flags
if (NOT CMAKE_BUILD_TYPE AND NOT CMAKE_CONFIGURATION_TYPES)
@@ -247,6 +273,7 @@ set(TARGET whisper)
add_library(${TARGET}
ggml.h
ggml.c
${GGML_CUDA_SOURCES}
whisper.h
whisper.cpp
)
@@ -279,6 +306,12 @@ if (BUILD_SHARED_LIBS)
)
endif()
if (GGML_CUDA_SOURCES)
message(STATUS "GGML CUDA sources found, configuring CUDA architecture")
set_property(TARGET whisper PROPERTY CUDA_ARCHITECTURES OFF)
set_property(TARGET whisper PROPERTY CUDA_SELECT_NVCC_ARCH_FLAGS "Auto")
endif()
if (EMSCRIPTEN)
set_target_properties(${TARGET} PROPERTIES COMPILE_FLAGS "-msimd128")
endif()