rapidsai / cuml

cuML - RAPIDS Machine Learning Library
https://docs.rapids.ai/api/cuml/stable/
Apache License 2.0
4.22k stars 530 forks source link

[QST]CMake Error at build/_deps/rapids-cmake-src/rapids-cmake/test/generate_resource_spec.cmake:82 #6027

Open suhansu opened 2 months ago

suhansu commented 2 months ago

What is your question? ~hi, gay I got an error while building cuML, which has been bothering me for a long time 图片

suhansu commented 2 months ago

I'm at cmake... He will pull a cumlprims_mg-src repository, but there seems to be some problems with this repository. When I annotate cumlprims_mg-related annotations in CMakeLists.txt, I will report a lot of problems in make where the header file is not found. I don't understand why we put cumlprims_mg-src in CMakeLists.txt.

dantegd commented 2 months ago

Hi @suhansu, thanks for the issue. cumlprims-mg contains some of the multigpu code for certain algorithms. You can install it from conda rapidsai and rapidsai-nightly channels: https://anaconda.org/rapidsai-nightly/libcumlprims and https://anaconda.org/rapidsai/libcumlprims. You can also avoid the need for this dependency when building with the singlegpu option: https://github.com/rapidsai/cuml/blob/c7f53ef92d80f604b04829406b1c0e16ba563823/build.sh#L43.

How are you building the repository? I would highly encourage using our conda developer environments here: https://github.com/rapidsai/cuml/blob/branch-24.10/conda/environments/all_cuda-125_arch-x86_64.yaml

suhansu commented 2 months ago

Thank you for your reply, I will try it, have a good day

dantegd commented 2 months ago

Let me know if you run into any further issue @suhansu, glad to try to help!