2023-09-16 14:57:49 -04:00
|
|
|
cmake_minimum_required(VERSION 3.21)
|
2023-03-23 13:54:14 -04:00
|
|
|
|
|
|
|
|
project(llama_cpp)
|
|
|
|
|
|
2023-09-12 18:29:18 -04:00
|
|
|
option(LLAMA_BUILD "Build llama.cpp shared library and install alongside python package" ON)
|
2023-11-28 04:55:21 -05:00
|
|
|
option(LLAVA_BUILD "Build llava shared library and install alongside python package" ON)
|
2023-04-25 01:36:37 -04:00
|
|
|
|
2024-07-01 23:11:49 -04:00
|
|
|
function(llama_cpp_python_install_target target)
|
|
|
|
|
install(
|
|
|
|
|
TARGETS ${target}
|
|
|
|
|
LIBRARY DESTINATION ${CMAKE_CURRENT_SOURCE_DIR}/llama_cpp/lib
|
|
|
|
|
RUNTIME DESTINATION ${CMAKE_CURRENT_SOURCE_DIR}/llama_cpp/lib
|
|
|
|
|
ARCHIVE DESTINATION ${CMAKE_CURRENT_SOURCE_DIR}/llama_cpp/lib
|
|
|
|
|
FRAMEWORK DESTINATION ${CMAKE_CURRENT_SOURCE_DIR}/llama_cpp/lib
|
|
|
|
|
RESOURCE DESTINATION ${CMAKE_CURRENT_SOURCE_DIR}/llama_cpp/lib
|
|
|
|
|
)
|
|
|
|
|
install(
|
|
|
|
|
TARGETS ${target}
|
|
|
|
|
LIBRARY DESTINATION ${SKBUILD_PLATLIB_DIR}/llama_cpp/lib
|
|
|
|
|
RUNTIME DESTINATION ${SKBUILD_PLATLIB_DIR}/llama_cpp/lib
|
|
|
|
|
ARCHIVE DESTINATION ${SKBUILD_PLATLIB_DIR}/llama_cpp/lib
|
|
|
|
|
FRAMEWORK DESTINATION ${SKBUILD_PLATLIB_DIR}/llama_cpp/lib
|
|
|
|
|
RESOURCE DESTINATION ${SKBUILD_PLATLIB_DIR}/llama_cpp/lib
|
|
|
|
|
)
|
|
|
|
|
set_target_properties(${target} PROPERTIES
|
|
|
|
|
INSTALL_RPATH "$ORIGIN"
|
|
|
|
|
BUILD_WITH_INSTALL_RPATH TRUE
|
|
|
|
|
)
|
2024-07-01 23:17:02 -04:00
|
|
|
if(UNIX)
|
|
|
|
|
if(APPLE)
|
|
|
|
|
set_target_properties(${target} PROPERTIES
|
|
|
|
|
INSTALL_RPATH "@loader_path"
|
|
|
|
|
BUILD_WITH_INSTALL_RPATH TRUE
|
|
|
|
|
)
|
|
|
|
|
else()
|
|
|
|
|
set_target_properties(${target} PROPERTIES
|
|
|
|
|
INSTALL_RPATH "$ORIGIN"
|
|
|
|
|
BUILD_WITH_INSTALL_RPATH TRUE
|
|
|
|
|
)
|
|
|
|
|
endif()
|
|
|
|
|
endif()
|
2024-07-01 23:11:49 -04:00
|
|
|
endfunction()
|
2024-07-01 22:03:19 -04:00
|
|
|
|
2023-09-12 18:29:18 -04:00
|
|
|
if (LLAMA_BUILD)
|
2023-04-08 02:39:17 -04:00
|
|
|
set(BUILD_SHARED_LIBS "On")
|
2023-11-10 16:32:36 -05:00
|
|
|
|
2024-07-01 21:28:11 -04:00
|
|
|
set(CMAKE_SKIP_BUILD_RPATH FALSE)
|
|
|
|
|
|
|
|
|
|
# When building, don't use the install RPATH already
|
|
|
|
|
# (but later on when installing)
|
|
|
|
|
set(CMAKE_BUILD_WITH_INSTALL_RPATH FALSE)
|
|
|
|
|
|
|
|
|
|
# Add the automatically determined parts of the RPATH
|
|
|
|
|
# which point to directories outside the build tree to the install RPATH
|
|
|
|
|
set(CMAKE_INSTALL_RPATH_USE_LINK_PATH TRUE)
|
2024-07-01 23:03:26 -04:00
|
|
|
set(CMAKE_SKIP_RPATH FALSE)
|
2024-07-01 21:28:11 -04:00
|
|
|
|
2023-11-10 16:32:36 -05:00
|
|
|
# Building llama
|
2023-09-14 18:25:44 -04:00
|
|
|
if (APPLE AND NOT CMAKE_SYSTEM_PROCESSOR MATCHES "arm64")
|
|
|
|
|
# Need to disable these llama.cpp flags on Apple x86_64,
|
2023-09-12 18:41:00 -04:00
|
|
|
# otherwise users may encounter invalid instruction errors
|
2024-07-09 00:29:29 -04:00
|
|
|
set(GGML_AVX "Off" CACHE BOOL "ggml: enable AVX" FORCE)
|
|
|
|
|
set(GGML_AVX2 "Off" CACHE BOOL "ggml: enable AVX2" FORCE)
|
|
|
|
|
set(GGML_FMA "Off" CACHE BOOL "gml: enable FMA" FORCE)
|
|
|
|
|
set(GGML_F16C "Off" CACHE BOOL "gml: enable F16C" FORCE)
|
2023-09-12 18:25:31 -04:00
|
|
|
endif()
|
2024-03-20 17:46:09 +01:00
|
|
|
|
2024-04-06 01:36:53 -04:00
|
|
|
if (APPLE)
|
2024-07-09 00:29:29 -04:00
|
|
|
set(GGML_METAL_EMBED_LIBRARY "On" CACHE BOOL "llama: embed metal library" FORCE)
|
2024-03-20 17:46:09 +01:00
|
|
|
endif()
|
|
|
|
|
|
2023-09-12 18:29:18 -04:00
|
|
|
add_subdirectory(vendor/llama.cpp)
|
2024-07-01 23:11:49 -04:00
|
|
|
llama_cpp_python_install_target(llama)
|
|
|
|
|
llama_cpp_python_install_target(ggml)
|
2024-07-01 23:28:19 -04:00
|
|
|
|
2023-09-16 14:57:49 -04:00
|
|
|
# Workaround for Windows + CUDA https://github.com/abetlen/llama-cpp-python/issues/563
|
2024-07-01 23:28:19 -04:00
|
|
|
if (WIN32)
|
2024-06-03 22:42:34 -06:00
|
|
|
install(
|
|
|
|
|
FILES $<TARGET_RUNTIME_DLLS:llama>
|
2024-07-01 23:28:19 -04:00
|
|
|
DESTINATION ${CMAKE_CURRENT_SOURCE_DIR}/llama_cpp/lib
|
|
|
|
|
)
|
|
|
|
|
install(
|
|
|
|
|
FILES $<TARGET_RUNTIME_DLLS:llama>
|
|
|
|
|
DESTINATION ${SKBUILD_PLATLIB_DIR}/llama_cpp/lib
|
|
|
|
|
)
|
|
|
|
|
install(
|
|
|
|
|
FILES $<TARGET_RUNTIME_DLLS:ggml>
|
|
|
|
|
DESTINATION ${CMAKE_CURRENT_SOURCE_DIR}/llama_cpp/lib
|
|
|
|
|
)
|
|
|
|
|
install(
|
|
|
|
|
FILES $<TARGET_RUNTIME_DLLS:ggml>
|
|
|
|
|
DESTINATION ${SKBUILD_PLATLIB_DIR}/llama_cpp/lib
|
2024-06-03 22:42:34 -06:00
|
|
|
)
|
|
|
|
|
endif()
|
2023-11-10 16:32:36 -05:00
|
|
|
|
2023-11-28 04:55:21 -05:00
|
|
|
if (LLAVA_BUILD)
|
2024-05-10 10:18:47 -04:00
|
|
|
if (LLAMA_CUBLAS OR LLAMA_CUDA)
|
2024-02-11 12:53:59 -06:00
|
|
|
add_compile_definitions(GGML_USE_CUBLAS)
|
2024-05-10 10:18:47 -04:00
|
|
|
add_compile_definitions(GGML_USE_CUDA)
|
2024-02-11 12:53:59 -06:00
|
|
|
endif()
|
|
|
|
|
|
|
|
|
|
if (LLAMA_METAL)
|
|
|
|
|
add_compile_definitions(GGML_USE_METAL)
|
|
|
|
|
endif()
|
|
|
|
|
|
2023-11-28 04:55:21 -05:00
|
|
|
# Building llava
|
|
|
|
|
add_subdirectory(vendor/llama.cpp/examples/llava)
|
|
|
|
|
set_target_properties(llava_shared PROPERTIES OUTPUT_NAME "llava")
|
|
|
|
|
# Set CUDA_ARCHITECTURES to OFF on windows
|
|
|
|
|
if (WIN32)
|
|
|
|
|
set_target_properties(llava_shared PROPERTIES CUDA_ARCHITECTURES OFF)
|
|
|
|
|
endif()
|
2024-07-01 23:11:49 -04:00
|
|
|
llama_cpp_python_install_target(llava_shared)
|
2024-07-01 23:28:19 -04:00
|
|
|
if (WIN32)
|
|
|
|
|
install(
|
|
|
|
|
FILES $<TARGET_RUNTIME_DLLS:llava_shared>
|
|
|
|
|
DESTINATION ${CMAKE_CURRENT_SOURCE_DIR}/llama_cpp/lib
|
|
|
|
|
)
|
|
|
|
|
install(
|
|
|
|
|
FILES $<TARGET_RUNTIME_DLLS:llava_shared>
|
|
|
|
|
DESTINATION ${SKBUILD_PLATLIB_DIR}/llama_cpp/lib
|
|
|
|
|
)
|
|
|
|
|
endif()
|
2023-11-10 16:32:36 -05:00
|
|
|
endif()
|
2024-07-09 00:29:29 -04:00
|
|
|
endif()
|
