akifcorduk
commented
3 years ago The minimum standalone possible Cmake configuration that works for me:
#
# Copyright (c) 2019, NVIDIA CORPORATION.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
#
cmake_minimum_required(VERSION 3.8 FATAL_ERROR)
project(kmeans_example LANGUAGES CXX CUDA)
set(CMAKE_CXX_STANDARD 14)
set(CMAKE_CXX_STANDARD_REQUIRED ON)
set(CMAKE_CUDA_FLAGS "${CMAKE_CUDA_FLAGS} --expt-extended-lambda --expt-relaxed-constexpr")
set(CMAKE_CUDA_FLAGS "${CMAKE_CUDA_FLAGS} -Werror=cross-execution-space-call -Wno-deprecated-declarations -Xptxas --disable-warnings")
set(CMAKE_CUDA_FLAGS "${CMAKE_CUDA_FLAGS} -Xcompiler -Wall,-Wno-error=sign-compare,-Wno-error=unused-but-set-variable")
include(FetchContent)
# - THRUST/CUB
message("Fetching Thrust")
FetchContent_Declare(
thrust
GIT_REPOSITORY https://github.com/thrust/thrust.git
# August 28, 2020
GIT_TAG 52a8bda46c5c2128414d1d47f546b486ff0be2f0
)
FetchContent_GetProperties(thrust)
if(NOT thrust_POPULATED)
FetchContent_Populate(thrust)
# We are not using the thrust CMake targets, so no need to call `add_subdirectory()`.
endif()
set(THRUST_INCLUDE_DIR "${thrust_SOURCE_DIR}")
if(IS_DIRECTORY ${CUML_INCLUDE_DIR})
# CMAKE_CUDA_TOOLKIT_INCLUDE_DIRECTORIES is needed so that cuda_runtime.h is found
# CUML_INCLUDE_DIR is needed so that kmeans/kmeans_c.h is found
include_directories(${CMAKE_CUDA_TOOLKIT_INCLUDE_DIRECTORIES} ${CUML_INCLUDE_DIR} ${THRUST_INCLUDE_DIR})
else()
message(FATAL_ERROR "CUML_INCLUDE_DIR not specified.")
endif(IS_DIRECTORY ${CUML_INCLUDE_DIR})
if(IS_DIRECTORY ${CUML_LIBRARY_DIR})
# CUML_LIBRARY_DIR point to the director where libcuml++.so lives
link_directories(${CUML_LIBRARY_DIR})
else()
message(FATAL_ERROR "CUML_LIBRARY_DIR not specified.")
endif(IS_DIRECTORY ${CUML_LIBRARY_DIR})
find_package(OpenMP)
set(FAISS_GPU_ARCHS "${FAISS_GPU_ARCHS} -gencode arch=compute_70,code=compute_70")
include(ExternalProject)
set(FAISS_DIR ${CMAKE_CURRENT_BINARY_DIR}/faiss CACHE STRING
"Path to FAISS source directory")
ExternalProject_Add(faiss
GIT_REPOSITORY https://github.com/facebookresearch/faiss.git
GIT_TAG a5b850dec6f1cd6c88ab467bfd5e87b0cac2e41d
CONFIGURE_COMMAND LIBS=-pthread
CPPFLAGS=-w
LDFLAGS=-L${CMAKE_INSTALL_PREFIX}/lib
${CMAKE_CURRENT_BINARY_DIR}/faiss/src/faiss/configure
--prefix=${CMAKE_CURRENT_BINARY_DIR}/faiss
--with-blas=${BLAS_LIBRARIES}
--with-cuda=${CUDA_TOOLKIT_ROOT_DIR}
--with-cuda-arch=${FAISS_GPU_ARCHS}
-v
PREFIX ${FAISS_DIR}
BUILD_COMMAND make -j${PARALLEL_LEVEL} VERBOSE=1
BUILD_BYPRODUCTS ${FAISS_DIR}/lib/libfaiss.a
BUILD_ALWAYS 1
INSTALL_COMMAND make -s install > /dev/null
UPDATE_COMMAND ""
BUILD_IN_SOURCE 1
PATCH_COMMAND patch -p1 -N < ${CMAKE_CURRENT_SOURCE_DIR}/cmake/faiss_cuda11.patch || true)
ExternalProject_Get_Property(faiss install_dir)
add_library(FAISS::FAISS STATIC IMPORTED)
add_dependencies(FAISS::FAISS faiss)
set_property(TARGET FAISS::FAISS PROPERTY
IMPORTED_LOCATION ${FAISS_DIR}/lib/libfaiss.a)
set(FAISS_INCLUDE_DIRS "${FAISS_DIR}/src")
add_executable(tsp_test
src/tsp.cu
src/tsp_test.cu)
set_property(TARGET tsp_test PROPERTY CUDA_ARCHITECTURES 70)
# Need to set linker language to CUDA to link the CUDA Runtime
set_target_properties(tsp_test PROPERTIES LINKER_LANGUAGE "CUDA")
# Link cuml
target_link_libraries(tsp_test cublas cusolver cusparse cudart cuda cuml++ FAISS::FAISS blas)
if(OpenMP_CXX_FOUND)
target_link_libraries(tsp_test
${OpenMP_CXX_LIB_NAMES})
endif(OpenMP_CXX_FOUND)Blas package found by cmake's find_package was incompatible with Faiss library. So it is required to give the blas library path of the conda environment as a cmake argument. I also had to copy raft include directory(which was under include/cuml/) to one level higher(include/) to be able to find raft header files in my project.
The following is the cmake command to build the above cmake script:
cmake .. -DCUML_LIBRARY_DIR="$CONDA_PREFIX/lib" -DBLAS_LIBRARIES=$CONDA_PREFIX/lib/libopenblas.so.0 -DCUML_INCLUDE_DIR="$CONDA_PREFIX/include"
github-actions[bot]
ITFUHENG
We currently have 2 examples dbscan and kmeans showing how to setup a C++ project and use libcuml++ inside them. They also come with
CMakeLists_standalone.txtwhich are supposed to be used by such folks to start right-off-the-bat. However, they don't seem to be updated, especially after all the library-related changes that have went into the cmake files of cuML project. We should keep these standalone files updated as per the changes made in the main project.We discovered this issue while @akifcorduk was trying to setup a standalone C++ project to debug his TSP-related work. @akifcorduk please share the cmake file showing the changes you had to make while trying to set this one up.