mirror of
https://github.com/ggerganov/llama.cpp.git
synced 2024-12-31 22:04:35 +00:00
2025fa67e9
* kompute: op_unary: reject unsupported parameters Signed-off-by: Sergio Lopez <slp@redhat.com> * kompute: softmax: implement ALiBi support Signed-off-by: Sergio Lopez <slp@redhat.com> * kompute: rope: implement neox and phi3 support Signed-off-by: Sergio Lopez <slp@redhat.com> * kompute: op_mul_mat_q4_k permutted support Signed-off-by: Sergio Lopez <slp@redhat.com> * kompute: op_mul_mat_[q4_0|q4_1|q8_0] permutted support Signed-off-by: Sergio Lopez <slp@redhat.com> * kompute: op_mul_mat_f16 permutted support Signed-off-by: Sergio Lopez <slp@redhat.com> * kompute: op_mul_mat_q6_k permutted support Signed-off-by: Sergio Lopez <slp@redhat.com> --------- Signed-off-by: Sergio Lopez <slp@redhat.com>
167 lines
7.3 KiB
CMake
167 lines
7.3 KiB
CMake
|
|
find_package(Vulkan COMPONENTS glslc REQUIRED)
|
|
find_program(glslc_executable NAMES glslc HINTS Vulkan::glslc)
|
|
|
|
if (NOT glslc_executable)
|
|
message(FATAL_ERROR "glslc not found")
|
|
endif()
|
|
|
|
ggml_add_backend_library(ggml-kompute
|
|
ggml-kompute.cpp
|
|
../../include/ggml-kompute.h
|
|
)
|
|
|
|
target_link_libraries(ggml-kompute PRIVATE ggml-base kompute)
|
|
target_include_directories(ggml-kompute PRIVATE ${CMAKE_CURRENT_BINARY_DIR})
|
|
|
|
add_compile_definitions(VULKAN_HPP_DISPATCH_LOADER_DYNAMIC=1)
|
|
|
|
function(compile_shader)
|
|
set(options)
|
|
set(oneValueArgs)
|
|
set(multiValueArgs SOURCES)
|
|
cmake_parse_arguments(compile_shader "${options}" "${oneValueArgs}" "${multiValueArgs}" ${ARGN})
|
|
foreach(source ${compile_shader_SOURCES})
|
|
get_filename_component(filename ${source} NAME)
|
|
set(spv_file ${filename}.spv)
|
|
add_custom_command(
|
|
OUTPUT ${spv_file}
|
|
DEPENDS ${CMAKE_CURRENT_SOURCE_DIR}/${source}
|
|
${CMAKE_CURRENT_SOURCE_DIR}/kompute-shaders/common.comp
|
|
${CMAKE_CURRENT_SOURCE_DIR}/kompute-shaders/op_getrows.comp
|
|
${CMAKE_CURRENT_SOURCE_DIR}/kompute-shaders/op_mul_mv_q_n_pre.comp
|
|
${CMAKE_CURRENT_SOURCE_DIR}/kompute-shaders/op_mul_mv_q_n.comp
|
|
COMMAND ${glslc_executable} --target-env=vulkan1.2 -o ${spv_file} ${CMAKE_CURRENT_SOURCE_DIR}/${source}
|
|
COMMENT "Compiling ${source} to ${spv_file}"
|
|
)
|
|
|
|
get_filename_component(RAW_FILE_NAME ${spv_file} NAME)
|
|
set(FILE_NAME "shader${RAW_FILE_NAME}")
|
|
string(REPLACE ".comp.spv" ".h" HEADER_FILE ${FILE_NAME})
|
|
string(TOUPPER ${HEADER_FILE} HEADER_FILE_DEFINE)
|
|
string(REPLACE "." "_" HEADER_FILE_DEFINE "${HEADER_FILE_DEFINE}")
|
|
set(OUTPUT_HEADER_FILE "${HEADER_FILE}")
|
|
message(STATUS "${HEADER_FILE} generating ${HEADER_FILE_DEFINE}")
|
|
if(CMAKE_GENERATOR MATCHES "Visual Studio")
|
|
add_custom_command(
|
|
OUTPUT ${OUTPUT_HEADER_FILE}
|
|
COMMAND ${CMAKE_COMMAND} -E echo "/*THIS FILE HAS BEEN AUTOMATICALLY GENERATED - DO NOT EDIT*/" > ${OUTPUT_HEADER_FILE}
|
|
COMMAND ${CMAKE_COMMAND} -E echo \"\#ifndef ${HEADER_FILE_DEFINE}\" >> ${OUTPUT_HEADER_FILE}
|
|
COMMAND ${CMAKE_COMMAND} -E echo \"\#define ${HEADER_FILE_DEFINE}\" >> ${OUTPUT_HEADER_FILE}
|
|
COMMAND ${CMAKE_COMMAND} -E echo "namespace kp {" >> ${OUTPUT_HEADER_FILE}
|
|
COMMAND ${CMAKE_COMMAND} -E echo "namespace shader_data {" >> ${OUTPUT_HEADER_FILE}
|
|
COMMAND ${CMAKE_BINARY_DIR}/bin/$<CONFIG>/xxd -i ${RAW_FILE_NAME} >> ${OUTPUT_HEADER_FILE}
|
|
COMMAND ${CMAKE_COMMAND} -E echo "}}" >> ${OUTPUT_HEADER_FILE}
|
|
COMMAND ${CMAKE_COMMAND} -E echo \"\#endif // define ${HEADER_FILE_DEFINE}\" >> ${OUTPUT_HEADER_FILE}
|
|
DEPENDS ${spv_file} xxd
|
|
COMMENT "Converting to hpp: ${FILE_NAME} ${CMAKE_BINARY_DIR}/bin/$<CONFIG>/xxd"
|
|
)
|
|
else()
|
|
add_custom_command(
|
|
OUTPUT ${OUTPUT_HEADER_FILE}
|
|
COMMAND ${CMAKE_COMMAND} -E echo "/*THIS FILE HAS BEEN AUTOMATICALLY GENERATED - DO NOT EDIT*/" > ${OUTPUT_HEADER_FILE}
|
|
COMMAND ${CMAKE_COMMAND} -E echo \"\#ifndef ${HEADER_FILE_DEFINE}\" >> ${OUTPUT_HEADER_FILE}
|
|
COMMAND ${CMAKE_COMMAND} -E echo \"\#define ${HEADER_FILE_DEFINE}\" >> ${OUTPUT_HEADER_FILE}
|
|
COMMAND ${CMAKE_COMMAND} -E echo "namespace kp {" >> ${OUTPUT_HEADER_FILE}
|
|
COMMAND ${CMAKE_COMMAND} -E echo "namespace shader_data {" >> ${OUTPUT_HEADER_FILE}
|
|
COMMAND ${CMAKE_BINARY_DIR}/bin/xxd -i ${RAW_FILE_NAME} >> ${OUTPUT_HEADER_FILE}
|
|
COMMAND ${CMAKE_COMMAND} -E echo "}}" >> ${OUTPUT_HEADER_FILE}
|
|
COMMAND ${CMAKE_COMMAND} -E echo \"\#endif // define ${HEADER_FILE_DEFINE}\" >> ${OUTPUT_HEADER_FILE}
|
|
DEPENDS ${spv_file} xxd
|
|
COMMENT "Converting to hpp: ${FILE_NAME} ${CMAKE_BINARY_DIR}/bin/xxd"
|
|
)
|
|
endif()
|
|
endforeach()
|
|
endfunction()
|
|
|
|
if (EXISTS "${CMAKE_CURRENT_SOURCE_DIR}/kompute/CMakeLists.txt")
|
|
message(STATUS "Kompute found")
|
|
set(KOMPUTE_OPT_LOG_LEVEL Error CACHE STRING "Kompute log level")
|
|
add_subdirectory(kompute)
|
|
|
|
# Compile our shaders
|
|
compile_shader(SOURCES
|
|
kompute-shaders/op_scale.comp
|
|
kompute-shaders/op_scale_8.comp
|
|
kompute-shaders/op_add.comp
|
|
kompute-shaders/op_addrow.comp
|
|
kompute-shaders/op_mul.comp
|
|
kompute-shaders/op_silu.comp
|
|
kompute-shaders/op_relu.comp
|
|
kompute-shaders/op_gelu.comp
|
|
kompute-shaders/op_softmax.comp
|
|
kompute-shaders/op_norm.comp
|
|
kompute-shaders/op_rmsnorm.comp
|
|
kompute-shaders/op_diagmask.comp
|
|
kompute-shaders/op_mul_mat_mat_f32.comp
|
|
kompute-shaders/op_mul_mat_f16.comp
|
|
kompute-shaders/op_mul_mat_q8_0.comp
|
|
kompute-shaders/op_mul_mat_q4_0.comp
|
|
kompute-shaders/op_mul_mat_q4_1.comp
|
|
kompute-shaders/op_mul_mat_q4_k.comp
|
|
kompute-shaders/op_mul_mat_q6_k.comp
|
|
kompute-shaders/op_getrows_f32.comp
|
|
kompute-shaders/op_getrows_f16.comp
|
|
kompute-shaders/op_getrows_q4_0.comp
|
|
kompute-shaders/op_getrows_q4_1.comp
|
|
kompute-shaders/op_getrows_q6_k.comp
|
|
kompute-shaders/op_rope_norm_f16.comp
|
|
kompute-shaders/op_rope_norm_f32.comp
|
|
kompute-shaders/op_rope_neox_f16.comp
|
|
kompute-shaders/op_rope_neox_f32.comp
|
|
kompute-shaders/op_cpy_f16_f16.comp
|
|
kompute-shaders/op_cpy_f16_f32.comp
|
|
kompute-shaders/op_cpy_f32_f16.comp
|
|
kompute-shaders/op_cpy_f32_f32.comp
|
|
)
|
|
|
|
# Create a custom target for our generated shaders
|
|
add_custom_target(generated_shaders DEPENDS
|
|
shaderop_scale.h
|
|
shaderop_scale_8.h
|
|
shaderop_add.h
|
|
shaderop_addrow.h
|
|
shaderop_mul.h
|
|
shaderop_silu.h
|
|
shaderop_relu.h
|
|
shaderop_gelu.h
|
|
shaderop_softmax.h
|
|
shaderop_norm.h
|
|
shaderop_rmsnorm.h
|
|
shaderop_diagmask.h
|
|
shaderop_mul_mat_mat_f32.h
|
|
shaderop_mul_mat_f16.h
|
|
shaderop_mul_mat_q8_0.h
|
|
shaderop_mul_mat_q4_0.h
|
|
shaderop_mul_mat_q4_1.h
|
|
shaderop_mul_mat_q4_k.h
|
|
shaderop_mul_mat_q6_k.h
|
|
shaderop_getrows_f32.h
|
|
shaderop_getrows_f16.h
|
|
shaderop_getrows_q4_0.h
|
|
shaderop_getrows_q4_1.h
|
|
shaderop_getrows_q6_k.h
|
|
shaderop_rope_norm_f16.h
|
|
shaderop_rope_norm_f32.h
|
|
shaderop_rope_neox_f16.h
|
|
shaderop_rope_neox_f32.h
|
|
shaderop_cpy_f16_f16.h
|
|
shaderop_cpy_f16_f32.h
|
|
shaderop_cpy_f32_f16.h
|
|
shaderop_cpy_f32_f32.h
|
|
)
|
|
|
|
# Create a custom command that depends on the generated_shaders
|
|
add_custom_command(
|
|
OUTPUT ${CMAKE_CURRENT_BINARY_DIR}/ggml-kompute.stamp
|
|
COMMAND ${CMAKE_COMMAND} -E touch ${CMAKE_CURRENT_BINARY_DIR}/ggml-kompute.stamp
|
|
DEPENDS generated_shaders
|
|
COMMENT "Ensuring shaders are generated before compiling ggml-kompute.cpp"
|
|
)
|
|
|
|
# Add the stamp to the main sources to ensure dependency tracking
|
|
target_sources(ggml-kompute PRIVATE ${CMAKE_CURRENT_BINARY_DIR}/ggml-kompute.stamp)
|
|
else()
|
|
message(WARNING "Kompute not found")
|
|
endif()
|