#meeting-2:fedoraproject.org: fedora-pytorch

Meeting started by @tflink:fedora.im at 17:30:54 UTC

Meeting summary

  1. TOPIC:welcome and roll call (@tflink:fedora.im, 17:31:01)
    1. LINK: agenda document https://board.net/p/fedora-pytorch-meeting (@tflink:fedora.im, 17:31:37)
  2. TOPIC:pytorch 2.3 (@tflink:fedora.im, 17:35:17)
    1. INFO: trix is considering epel packages for pytorch and is looking for feedback on whether those are desired (@tflink:fedora.im, 17:40:07)
    2. INFO: pytorch 2.3 with rocm support can be built but for the moment, support is limited to a few cards and has only been tested on mi210 (@tflink:fedora.im, 17:43:24)
    3. INFO: more tesitng for rccl and pytorch distributed is needed, those features may be enabled for pytorch 2.3 (@tflink:fedora.im, 17:50:57)
    4. INFO: caffe2 and openmp support are planned to be added for the pytorch 2.3 package (@tflink:fedora.im, 17:52:24)
  3. TOPIC:fesco update (@tflink:fedora.im, 17:55:04)
    1. LINK: https://pagure.io/fesco/issue/3175 (@tflink:fedora.im, 17:55:41)
    2. LINK: https://lists.fedoraproject.org/archives/list/legal@lists.fedoraproject.org/thread/PIPILJCMDEO67ORL4SAKB3NPHHVMFDJE/#PIPILJCMDEO67ORL4SAKB3NPHHVMFDJE (@tflink:fedora.im, 17:56:35)
    3. INFO: FESCo said that the issue on including pre-trained weights is an issue for legal so we started a public conversation with Fedora legal. once that conversation is concluded, we will re-open the FESCo issue (@tflink:fedora.im, 17:58:28)
  4. TOPIC:rocm splitting and pytorch (@tflink:fedora.im, 18:00:10)
    1. INFO: some rocm components generate libraries which are too large to package if all supported gpus are enabled so they have been split up into multiple libraries by gpu family. this causes problems with downstream applications (like pytorch) which need to link against a single library (@tflink:fedora.im, 18:04:36)
    2. LINK: https://src.fedoraproject.org/rpms/python-torch/tree/rhel-test (@tflink:fedora.im, 18:11:09)
    3. INFO: the current strategy is to build pytorch multiple times within the package to support multiple gpu families - this does mean that for not-current-gen GPUs (outside of gfx10 and gfx11 at the moment), users would have to use a custom PYTHONPATH to get rocm accelerated pytorch to work (@tflink:fedora.im, 18:28:22)
    4. INFO: this solution isn't ideal but until things that are outside of our control change, it's either this or to farther restrict support for gpu families like amd does for the binaries that they distribute (@tflink:fedora.im, 18:29:21)
  5. TOPIC:open floor (@tflink:fedora.im, 18:30:03)


Meeting ended at 18:33:17 UTC

Action items

  1. (none)


People present (lines said)

  1. @tflink:fedora.im (98)
  2. @trix:fedora.im (53)
  3. @kaitlynabdo:fedora.im (8)
  4. @zodbot:fedora.im (2)
  5. @meetbot:fedora.im (2)
  6. @conan_kudo:matrix.org (2)
  7. @davide:cavalca.name (1)