Getting Started

Runtime requirements

  • CUDA Toolkit version 13.0

    • Only the NVRTC redistributable component is required from the CUDA Toolkit. CUDA Toolkit Documentation Installation Guide can be used for guidance.

    • Note that the NVRTC component in the Toolkit can be obtained via PYPI, Conda or Local Installer.

  • Python version >= 3.9

Installing from PyPI

$ pip install cupti-python

Setting for CUPTI C Library

  • CUPTI Python has a dependency on nvidia-cuda-cupti. It uses the CUPTI C library libcupti.so* from nvidia-cuda-cupti.

  • However, if nvidia-cuda-cupti is uninstalled, the LD_LIBRARY_PATH environment variable needs to be set to the directory in which libcupti.so* is present. When the CUDA Toolkit is installed, libcupti.so* is present under $CUDA_TOOLKIT_INSTALL_PATH/extras/CUPTI/lib64.

  • e.g. If the CUDA_TOOLKIT_INSTALL_PATH is /usr/local/cuda-13.0:

$ export LD_LIBRARY_PATH=/usr/local/cuda-13.0/extras/CUPTI/lib64

NVTX Configuration

  • CUPTI Python provides support for NVTX through the activity kinds cupti.cupti.ActivityKind.MARKER, cupti.cupti.ActivityKind.MARKER_DATA, and cupti.cupti.ActivityKind.NAME.

  • To enable proper collection of NVTX-related activity records, ensure the following setup:

  • Set the NVTX_INJECTION64_PATH environment variable to point to the same CUPTI shared library that CUPTI Python is using.

  • e.g. If CUPTI Python is using the CUPTI shared library from the nvidia-cuda-cupti package, the path should be:

$ export NVTX_INJECTION64_PATH=<site-packages>/nvidia/cu13/lib/libcupti.so.13