René's URL Explorer Experiment

{"@context":"https://schema.org","@type":"DiscussionForumPosting","headline":"[FEA]: pathfinder support for `cufftMp` wheel","articleBody":"### Is this a duplicate?\n\n- [x] I confirmed there appear to be no [duplicate issues](https://github.com/NVIDIA/cuda-python/issues) for this request and that I agree to the [Code of Conduct](CODE_OF_CONDUCT.md)\n\n### Area\n\ncuda.pathfinder\n\n### Is your feature request related to a problem? Please describe.\n\nReproducer (using pathfinder version under #855):\n\n```\n+ python3 -m venv TempVenv\n+ set +x\n+ set +x\nRequirement already satisfied: pip in ./TempVenv/lib/python3.12/site-packages (24.0)\nCollecting pip\n  Using cached pip-25.2-py3-none-any.whl.metadata (4.7 kB)\nUsing cached pip-25.2-py3-none-any.whl (1.8 MB)\nInstalling collected packages: pip\n  Attempting uninstall: pip\n    Found existing installation: pip 24.0\n    Uninstalling pip-24.0:\n      Successfully uninstalled pip-24.0\nSuccessfully installed pip-25.2\n+ set +x\nProcessing /home/rgrossekunst/forked/cuda-python/cuda_pathfinder\n  Installing build dependencies: started\n  Installing build dependencies: finished with status 'done'\n  Getting requirements to build wheel: started\n  Getting requirements to build wheel: finished with status 'done'\n  Preparing metadata (pyproject.toml): started\n  Preparing metadata (pyproject.toml): finished with status 'done'\nBuilding wheels for collected packages: cuda-pathfinder\n  Building wheel for cuda-pathfinder (pyproject.toml): started\n  Building wheel for cuda-pathfinder (pyproject.toml): finished with status 'done'\n  Created wheel for cuda-pathfinder: filename=cuda_pathfinder-1.1.1a2-py3-none-any.whl size=18404 sha256=55eecd8eb958d7db3794295b039ff938c83aa541ef1ed7a45780a5df78045aab\n  Stored in directory: /tmp/pip-ephem-wheel-cache-b33fxgsq/wheels/d1/04/8c/c856bb3f85f3511d35e8d69acc9ca63f99f4d52d879aba968a\nSuccessfully built cuda-pathfinder\nInstalling collected packages: cuda-pathfinder\nSuccessfully installed cuda-pathfinder-1.1.1a2\n+ set +x\nCollecting nvidia-cufftmp-cu12\n  Using cached nvidia_cufftmp_cu12-11.4.0.6.post1-py3-none-manylinux_2_28_x86_64.whl.metadata (1.5 kB)\nCollecting nvidia-nvshmem-cu12 (from nvidia-cufftmp-cu12)\n  Using cached nvidia_nvshmem_cu12-3.3.20-py3-none-manylinux2014_x86_64.manylinux_2_17_x86_64.whl.metadata (2.1 kB)\nUsing cached nvidia_cufftmp_cu12-11.4.0.6.post1-py3-none-manylinux_2_28_x86_64.whl (207.1 MB)\nUsing cached nvidia_nvshmem_cu12-3.3.20-py3-none-manylinux2014_x86_64.manylinux_2_17_x86_64.whl (124.7 MB)\nInstalling collected packages: nvidia-nvshmem-cu12, nvidia-cufftmp-cu12\n\nSuccessfully installed nvidia-cufftmp-cu12-11.4.0.6.post1 nvidia-nvshmem-cu12-3.3.20\n+ find TempVenv\n+ grep libcufftMp.so\nTempVenv/lib/python3.12/site-packages/nvidia/cufftmp/cu12/lib/libcufftMp.so.11\n+ python -c 'from cuda import pathfinder as pf; print(pf.load_nvidia_dynamic_lib('\\''cufftMp'\\''))'\nTraceback (most recent call last):\n  File \"\u003cstring>\", line 1, in \u003cmodule>\n  File \"/home/rgrossekunst/forked/cuda-python/cuda_pathfinder/cuda/pathfinder/_dynamic_libs/load_nvidia_dynamic_lib.py\", line 74, in load_nvidia_dynamic_lib\n    return _load_lib_no_cache(libname)\n           ^^^^^^^^^^^^^^^^^^^^^^^^^^^\n  File \"/home/rgrossekunst/forked/cuda-python/cuda_pathfinder/cuda/pathfinder/_dynamic_libs/load_nvidia_dynamic_lib.py\", line 48, in _load_lib_no_cache\n    found.raise_if_abs_path_is_None()\n  File \"/home/rgrossekunst/forked/cuda-python/cuda_pathfinder/cuda/pathfinder/_dynamic_libs/find_nvidia_dynamic_lib.py\", line 194, in raise_if_abs_path_is_None\n    raise DynamicLibNotFoundError(f'Failure finding \"{self.lib_searched_for}\": {err}\\n{att}')\ncuda.pathfinder._dynamic_libs.load_dl_common.DynamicLibNotFoundError: Failure finding \"libcufftMp.so\": No such file: libcufftMp.so*\n  listdir(\"/home/rgrossekunst/forked/cuda-python/cuda_pathfinder/TempVenv/lib/python3.12/site-packages/nvidia/nvshmem/lib\"):\n    libnvshmem_device.a\n    libnvshmem_device.bc\n    libnvshmem_host.so.3\n    nvshmem_bootstrap_mpi.so.3\n    nvshmem_bootstrap_pmi.so.3\n    nvshmem_bootstrap_pmi2.so.3\n    nvshmem_bootstrap_pmix.so.3\n    nvshmem_bootstrap_shmem.so.3\n    nvshmem_bootstrap_uid.so.3\n    nvshmem_transport_ibdevx.so.3\n    nvshmem_transport_ibgda.so.3\n    nvshmem_transport_ibrc.so.3\n    nvshmem_transport_libfabric.so.3\n    nvshmem_transport_ucx.so.3\n```\n\n### Describe the solution you'd like\n\n`pf.load_nvidia_dynamic_lib('cufftMp')` above should succeed.\n\n### Describe alternatives you've considered\n\n_No response_\n\n### Additional context\n\n_No response_","author":{"url":"https://github.com/rwgk","@type":"Person","name":"rwgk"},"datePublished":"2025-08-18T21:54:55Z","url":"https://github.com/NVIDIA/cuda-python/issues/856"}