vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Merge branch 'concedo' into concedo_experimental

Browse source
This commit is contained in:
Concedo 2023-06-30 23:06:21 +08:00
commit d16926dff4
1 changed files with 12 additions and 5 deletions
Download patch file Download diff file Expand all files Collapse all files

15
CMakeLists.txt
View file

@ -42,7 +42,11 @@ endif()
# 3rd party libs
option(LLAMA_CUBLAS "llama: use cuBLAS" ON)
set(LLAMA_CUDA_DMMV_X "32" CACHE STRING "llama: x stride for dmmv CUDA kernels")
set(LLAMA_CUDA_DMMV_Y "1" CACHE STRING "llama: y block size for dmmv CUDA kernels")
option(LLAMA_CUDA_DMMV_F16 "llama: use 16 bit floats for dmmv CUDA kernels" OFF)
set(LLAMA_CUDA_KQUANTS_ITER "2" CACHE STRING "llama: iters./thread per block for Q2_K/Q6_K")
option(LLAMA_K_QUANTS "llama: use k-quants" ON)
#
@ -72,6 +76,12 @@ if (LLAMA_CUBLAS)
set(GGML_V2_LEGACY_CUDA_SOURCES otherarch/ggml_v2-cuda-legacy.cu otherarch/ggml_v2-cuda-legacy.h)
add_compile_definitions(GGML_USE_CUBLAS)
add_compile_definitions(GGML_CUDA_DMMV_X=${LLAMA_CUDA_DMMV_X})
add_compile_definitions(GGML_CUDA_DMMV_Y=${LLAMA_CUDA_DMMV_Y})
if (LLAMA_CUDA_DMMV_F16)
add_compile_definitions(GGML_CUDA_DMMV_F16)
endif()
add_compile_definitions(K_QUANTS_PER_ITERATION=${LLAMA_CUDA_KQUANTS_ITER})
if (LLAMA_STATIC)
set(LLAMA_EXTRA_LIBS ${LLAMA_EXTRA_LIBS} CUDA::cudart_static CUDA::cublas_static CUDA::cublasLt_static)
@ -84,8 +94,6 @@ if (LLAMA_CUBLAS)
endif()
endif()
if (LLAMA_ALL_WARNINGS)
if (NOT MSVC)
set(c_flags
@ -298,4 +306,3 @@ set_target_properties(${TARGET} PROPERTIES OUTPUT_NAME "koboldcpp_cublas")
set_target_properties(${TARGET} PROPERTIES POSITION_INDEPENDENT_CODE ON)
target_link_libraries(${TARGET} PUBLIC ggml ggml_v1 ggml_v2 common2 gpttype_adapter ${CMAKE_THREAD_LIBS_INIT})
target_compile_features(${TARGET} PRIVATE cxx_std_11)