AMD offers wheel files for following Python libraries to help users get optimal performance on AMD EPYC™ processors.

  • NumPy with AOCL
  • SciPy with AOCL
  • PyTorch with AOCL

For more details, refer to the user guide.

For details of the AOCL Data Analytics patch for scikit-learn view the AOCL Data-Analytics page.

Download the files using the links in Downloads section. 

Resources and Technical Support

Documentation

Python Libraries with AOCL User Guide

Support

For support options, refer to Technical Support.

AMD Community

For moderated forums, refer to the AMD community.

Download with End User License Agreement

File Name Python Library Version Python Version Size Launch Date OS Bitness Description
numpy-1.26.4-cp311-cp311-linux_x86_64.whl 1.26.4 3.11 19MB 1/15/2025 Rocky 9.x, Ubuntu 22.04 64-bit GCC compiled NumPy with AOCL-5.0
sha256 Checksum: d0d6814e29cfcf2762a656583c7b96a4c28a0013e1c65972ac4bcbc152878254
numpy-1.26.4-cp312-cp312-linux_x86_64.whl 1.26.4 3.12 19MB 1/15/2025 Rocky 9.x, Ubuntu 22.04 64-bit GCC compiled NumPy with AOCL-5.0
sha256 Checksum: 1a65b56dba8fdaafaafea39207132c34253ab79a94c1a3a633f4a36b70984596
SciPy-1.11.3-cp311-cp311-linux_x86_64.whl 1.11.3 3.11 38MB 1/15/2025 Rocky 9.x, Ubuntu 22.04 64-bit GCC compiled SciPy with AOCL-5.0
sha256 Checksum: 78e31b6074ddb4c37f4b7a765421ac0bbee1414eaafd26253fc381a376eec348
SciPy-1.11.3-cp312-cp312-linux_x86_64.whl 1.11.3 3.12 38MB 1/15/2025 Rocky 9.x, Ubuntu 22.04 64-bit GCC compiled SciPy with AOCL-5.0
sha256 Checksum: 5428374ca7eadbad2623ceb00e5870966c78fbcf9fc6c0f98267d974bc1635b6
torch-2.4.0a0+gitd990dad-cp311-cp311-linux_x86_64.whl 2.4.0 3.11 150MB 1/15/2025 Rocky 9.x, Ubuntu 22.04 64-bit GCC compiled Pytorch with AOCL-5.0
sha256 Checksum: 82dbb5e7861486ba523b8c6245e03f3f03c45e132b63dede10fabd69e7c7114a
torch-2.4.0a0+gitd990dad-cp312-cp312-linux_x86_64.whl 2.4.0 3.12 150MB 1/15/2025 Rocky 9.x, Ubuntu 22.04 64-bit GCC compiled Pytorch with AOCL-5.0
sha256 Checksum: 8b61fc7df832cf640266fe17dd1d178f0e99833bcc50978f24d6c09be05dd57f