[QST] Error when build from source

[Thaumaturge2020]

when I try to build the project I encounter error when compiling linear_programming/solve.cu:

FAILED: CMakeFiles/cuopt.dir/src/linear_programming/solve.cu.o
/usr/bin/nvcc -forward-unknown-to-host-compiler -DCCCL_DISABLE_PDL -DCUB_DISABLE_NAMESPACE_MAGIC -DCUB_IGNORE_NAMESPACE_MAGIC_ERROR -DCUDSS_MT_LIB_FILE_NAME="libcudss_mtlayer_gomp.so.0" -DCUOPT_CPU_ARCHITECTURE="x86_64" -DCUOPT_CUDA_ARCHITECTURES="89-real" -DCUOPT_LOG_ACTIVE_LEVEL=RAPIDS_LOGGER_LOG_LEVEL_INFO -DCUTLASS_NAMESPACE=raft_cutlass -DLIBCUDACXX_ENABLE_EXPERIMENTAL_MEMORY_RESOURCE -DNVTX_ENABLED -DRAFT_LOG_ACTIVE_LEVEL=RAPIDS_LOGGER_LOG_LEVEL_INFO -DRAFT_SYSTEM_LITTLE_ENDIAN=1 -DTHRUST_DEVICE_SYSTEM=THRUST_DEVICE_SYSTEM_CUDA -DTHRUST_DISABLE_ABI_NAMESPACE -DTHRUST_HOST_SYSTEM=THRUST_HOST_SYSTEM_CPP -DTHRUST_IGNORE_ABI_NAMESPACE_ERROR -Dcuopt_EXPORTS -I/home/youmu/GIT/cuopt/cpp/../thirdparty -I/home/youmu/GIT/cuopt/cpp/src -I/home/youmu/GIT/cuopt/cpp/include -I/home/youmu/GIT/cuopt/cpp/build/include -I/home/youmu/GIT/cuopt/cpp/libmps_parser/include -I/home/youmu/GIT/cuopt/cpp/build/_deps/rmm-src/cpp/include -I/home/youmu/GIT/cuopt/cpp/build/_deps/rmm-build/include -I/home/youmu/GIT/cuopt/cpp/build/_deps/cccl-src/lib/cmake/thrust/../../../thrust -I/home/youmu/GIT/cuopt/cpp/build/_deps/cccl-src/lib/cmake/libcudacxx/../../../libcudacxx/include -I/home/youmu/GIT/cuopt/cpp/build/_deps/cccl-src/lib/cmake/cub/../../../cub -I/home/youmu/GIT/cuopt/cpp/build/_deps/nvtx3-src/c/include -I/home/youmu/GIT/cuopt/cpp/build/_deps/rapids_logger-src/include -I/home/youmu/GIT/cuopt/cpp/build/_deps/raft-src/cpp/include -I/home/youmu/GIT/cuopt/cpp/build/_deps/raft-build/include -I/home/youmu/GIT/cuopt/cpp/build/_deps/cuco-src/include -I/home/youmu/GIT/cuopt/cpp/build/_deps/nvidiacutlass-src/include -I/home/youmu/GIT/cuopt/cpp/build/_deps/nvidiacutlass-build/include -isystem /home/youmu/GIT/cuopt/cpp/build/_deps/papilo-src/src -isystem /home/youmu/GIT/cuopt/cpp/build/_deps/papilo-build --expt-relaxed-constexpr --expt-extended-lambda -O3 -DNDEBUG -lineinfo -std=c++17 "--generate-code=arch=compute_89,code=[sm_89]" -Xcompiler=-fPIC -Werror=cross-execution-space-call -Wno-deprecated-declarations -Xcompiler=-Werror --default-stream=per-thread -Xcompiler=-Wall -Wno-error=non-template-friend -Xfatbin=-compress-all -fopenmp -lineinfo --expt-extended-lambda --expt-relaxed-constexpr -MD -MT CMakeFiles/cuopt.dir/src/linear_programming/solve.cu.o -MF CMakeFiles/cuopt.dir/src/linear_programming/solve.cu.o.d -x cu -c /home/youmu/GIT/cuopt/cpp/src/linear_programming/solve.cu -o CMakeFiles/cuopt.dir/src/linear_programming/solve.cu.o
/home/youmu/GIT/cuopt/cpp/src/utilities/driver_helpers.cuh(20): error: identifier "cudaGetDriverEntryPointByVersion" is undefined

my g++/gcc version is 13.3
cuda-toolkit 12.0
CMake version 4.2.0

anyone encounter the situation like me and know how to solve it?

[aliceb-nv]

Hi!
cudaGetDriverEntryPointByVersion was introduced in CUDA 12.5. It seems our CUDA toolkit requirements in the documentation are outdated @rgsl888prabhu
Updating your CUDA Toolkit version to the latest 12.x or 13.x should fix the issue.

I see you're using your system's nvcc - following the instructions in our CONTRIBUTING.md and installing cuopt's conda environment should ensure you get a smooth build.
Could you confirm if this resolves your issue?
.
Thank you for your interest!

[Thaumaturge2020]

Thanks,update the cuda toolkit to 13.0 will help.