hf_text-generation-inference/backends/trtllm/cmake/trtllm.cmake

33 lines
1.1 KiB
CMake

set(USE_CXX11_ABI ON)
set(NVTX_DISABLE OFF)
set(BUILD_PYT OFF)
set(BUILD_PYBIND OFF)
set(BUILD_MICRO_BENCHMARKS OFF)
set(BUILD_BENCHMARKS OFF)
set(BUILD_TESTS OFF)
set(TRT_INCLUDE_DIR ${TGI_TRTLLM_BACKEND_TRT_INCLUDE_DIR})
set(TRT_LIB_DIR ${TGI_TRTLLM_BACKEND_TRT_LIB_DIR})
set(CMAKE_CUDA_ARCHITECTURES ${TGI_TRTLLM_BACKEND_TARGET_CUDA_ARCH_LIST})
message(STATUS "Building for CUDA Architectures: ${CMAKE_CUDA_ARCHITECTURES}")
if (${CMAKE_BUILD_TYPE} STREQUAL "Debug")
set(FAST_BUILD ON)
else ()
set(FAST_BUILD OFF)
endif ()
# This line turn off DEBUG in TRTLLM logger which is quite spammy
add_compile_definitions(NDEBUG OFF)
fetchcontent_declare(
trtllm
GIT_REPOSITORY https://github.com/nvidia/tensorrt-llm.git
GIT_TAG a96cccafcf6365c128f004f779160951f8c0801c
GIT_SHALLOW TRUE
)
fetchcontent_makeavailable(trtllm)
message(STATUS "Found TensorRT-LLM: ${trtllm_SOURCE_DIR}")
execute_process(COMMAND git lfs install WORKING_DIRECTORY "${trtllm_SOURCE_DIR}/")
execute_process(COMMAND git lfs pull WORKING_DIRECTORY "${trtllm_SOURCE_DIR}/")