diff options
Diffstat (limited to 'CMakeLists.txt')
| -rw-r--r-- | CMakeLists.txt | 52 |
1 files changed, 45 insertions, 7 deletions
diff --git a/CMakeLists.txt b/CMakeLists.txt index d9381da..d085bc8 100644 --- a/CMakeLists.txt +++ b/CMakeLists.txt @@ -67,11 +67,13 @@ endif() option(LLAMA_ACCELERATE "llama: enable Accelerate framework" ON) option(LLAMA_BLAS "llama: use BLAS" OFF) set(LLAMA_BLAS_VENDOR "Generic" CACHE STRING "llama: BLAS library vendor") -option(LLAMA_CUBLAS "llama: use cuBLAS" OFF) +option(LLAMA_CUBLAS "llama: use CUDA" OFF) +#option(LLAMA_CUDA_CUBLAS "llama: use cuBLAS for prompt processing" OFF) +set(LLAMA_CUDA_MMQ_Y "64" CACHE STRING "llama: y tile size for mmq CUDA kernels") option(LLAMA_CUDA_FORCE_DMMV "llama: use dmmv instead of mmvq CUDA kernels" OFF) set(LLAMA_CUDA_DMMV_X "32" CACHE STRING "llama: x stride for dmmv CUDA kernels") set(LLAMA_CUDA_MMV_Y "1" CACHE STRING "llama: y block size for mmv CUDA kernels") -option(LLAMA_CUDA_DMMV_F16 "llama: use 16 bit floats for dmmv CUDA kernels" OFF) +option(LLAMA_CUDA_F16 "llama: use 16 bit floats for some calculations" OFF) set(LLAMA_CUDA_KQUANTS_ITER "2" CACHE STRING "llama: iters./thread per block for Q2_K/Q6_K") option(LLAMA_CLBLAST "llama: use CLBlast" OFF) option(LLAMA_METAL "llama: use Metal" OFF) @@ -251,6 +253,10 @@ if (LLAMA_CUBLAS) set(GGML_SOURCES_CUDA ggml-cuda.cu ggml-cuda.h) add_compile_definitions(GGML_USE_CUBLAS) +# if (LLAMA_CUDA_CUBLAS) +# add_compile_definitions(GGML_CUDA_CUBLAS) +# endif() + add_compile_definitions(GGML_CUDA_MMQ_Y=${LLAMA_CUDA_MMQ_Y}) if (LLAMA_CUDA_FORCE_DMMV) add_compile_definitions(GGML_CUDA_FORCE_DMMV) endif() @@ -259,8 +265,8 @@ if (LLAMA_CUBLAS) if (DEFINED LLAMA_CUDA_DMMV_Y) add_compile_definitions(GGML_CUDA_MMV_Y=${LLAMA_CUDA_DMMV_Y}) # for backwards compatibility endif() - if (LLAMA_CUDA_DMMV_F16) - add_compile_definitions(GGML_CUDA_DMMV_F16) + if (LLAMA_CUDA_F16 OR LLAMA_CUDA_DMMV_F16) + add_compile_definitions(GGML_CUDA_F16) endif() add_compile_definitions(K_QUANTS_PER_ITERATION=${LLAMA_CUDA_KQUANTS_ITER}) @@ -271,10 +277,14 @@ if (LLAMA_CUBLAS) endif() if (NOT DEFINED CMAKE_CUDA_ARCHITECTURES) - if (LLAMA_CUDA_DMMV_F16) - set(CMAKE_CUDA_ARCHITECTURES "60;61") # needed for f16 CUDA intrinsics + # 52 == lowest CUDA 12 standard + # 60 == f16 CUDA intrinsics + # 61 == integer CUDA intrinsics + # 70 == compute capability at which unrolling a loop in mul_mat_q kernels is faster + if (LLAMA_CUDA_F16 OR LLAMA_CUDA_DMMV_F16) + set(CMAKE_CUDA_ARCHITECTURES "60;61;70") # needed for f16 CUDA intrinsics else() - set(CMAKE_CUDA_ARCHITECTURES "52;61") # lowest CUDA 12 standard + lowest for integer intrinsics + set(CMAKE_CUDA_ARCHITECTURES "52;61;70") # lowest CUDA 12 standard + lowest for integer intrinsics endif() endif() message(STATUS "Using CUDA architectures: ${CMAKE_CUDA_ARCHITECTURES}") @@ -357,6 +367,7 @@ if (LLAMA_ALL_WARNINGS) -Wshadow -Wstrict-prototypes -Wpointer-arith + -Wmissing-prototypes ) set(cxx_flags -Wall @@ -496,6 +507,8 @@ endif() add_library(ggml OBJECT ggml.c ggml.h + ggml-alloc.c + ggml-alloc.h ${GGML_SOURCES_CUDA} ${GGML_SOURCES_OPENCL} ${GGML_SOURCES_METAL} @@ -512,6 +525,7 @@ if (BUILD_SHARED_LIBS) set_target_properties(ggml PROPERTIES POSITION_INDEPENDENT_CODE ON) add_library(ggml_shared SHARED $<TARGET_OBJECTS:ggml>) target_link_libraries(ggml_shared PUBLIC Threads::Threads ${LLAMA_EXTRA_LIBS}) + install(TARGETS ggml_shared LIBRARY) endif() add_library(llama @@ -533,8 +547,32 @@ if (BUILD_SHARED_LIBS) if (LLAMA_METAL) set_target_properties(llama PROPERTIES RESOURCE "${CMAKE_CURRENT_SOURCE_DIR}/ggml-metal.metal") endif() + install(TARGETS llama LIBRARY) endif() +include(GNUInstallDirs) +install( + FILES convert.py + PERMISSIONS + OWNER_READ + OWNER_WRITE + OWNER_EXECUTE + GROUP_READ + GROUP_EXECUTE + WORLD_READ + WORLD_EXECUTE + DESTINATION ${CMAKE_INSTALL_BINDIR}) +install( + FILES convert-lora-to-ggml.py + PERMISSIONS + OWNER_READ + OWNER_WRITE + OWNER_EXECUTE + GROUP_READ + GROUP_EXECUTE + WORLD_READ + WORLD_EXECUTE + DESTINATION ${CMAKE_INSTALL_BINDIR}) # # programs, examples and tests |