Bug 2539092

Summary: Review Request: python-torchcodec - A library for audio and signal processing with PyTorch
Product: [Fedora] Fedora Reporter: Peter Robinson <pbrobinson>
Component: Package ReviewAssignee: Neal Gompa <ngompa13>
Status: ASSIGNED --- QA Contact: Fedora Extras Quality Assurance <extras-qa>
Severity: unspecified Docs Contact:
Priority: unspecified    
Version: rawhideCC: ngompa13, package-review
Target Milestone: ---   
Target Release: ---   
Hardware: Unspecified   
OS: Unspecified   
Whiteboard:
Fixed In Version: Doc Type: ---
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: Type: Bug
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Embargoed:
Bug Depends On:    
Bug Blocks: 2547877    

Description Peter Robinson 2026-09-23 05:04:12 UTC
SPEC: https://pbrobinson.fedorapeople.org/python-torchcodec.spec
SRPM: https://pbrobinson.fedorapeople.org/python-torchcodec-0.16.0-1.fc44.src.rpm

Description:
TorchCodec is a Python library for decoding video and audio data into
PyTorch tensors, on CPU and CUDA GPU. It also supports video and audio
encoding on CPU! It aims to be fast, easy to use, and well integrated
into the PyTorch ecosystem. If you want to use PyTorch to train ML models
on videos and audio, TorchCodec is how you turn these into data.

We achieve these capabilities through:
* Pythonic APIs that mirror Python and PyTorch conventions.
* Relying on FFmpeg to do the decoding and encoding. TorchCodec uses the
version of FFmpeg you already have installed. FFmpeg is a mature library
with broad coverage available on most systems. It is, however, not easy
to use. TorchCodec abstracts FFmpeg's complexity to ensure it is used
correctly and efficiently.
* Returning data as PyTorch tensors, ready to be fed into PyTorch
transforms or used directly to train models.

FAS: pbrobinson

koji: https://koji.fedoraproject.org/koji/taskinfo?taskID=150522178

Comment 1 Neal Gompa 2026-09-23 14:00:07 UTC
Taking this review.

Comment 2 Neal Gompa 2026-10-29 15:58:23 UTC
I can't build it locally, so fedora-review continues to fail. This is the error:

  Building PyTorch for GPU arch: gfx900;gfx906;gfx908;gfx90a;gfx942;gfx1030;gfx1100
  CMake Error at /usr/lib64/python3.15/site-packages/torch/share/cmake/Caffe2/public/LoadHIP.cmake:103 (message):
    HIP compiler not found at /usr/lib/llvm/bin/clang++
  Call Stack (most recent call first):
    /usr/lib64/python3.15/site-packages/torch/share/cmake/Caffe2/Caffe2Config.cmake:74 (include)
    /usr/lib64/python3.15/site-packages/torch/share/cmake/Torch/TorchConfig.cmake:62 (find_package)
    src/torchcodec/_core/CMakeLists.txt:21 (find_package)
  -- Configuring incomplete, errors occurred!
  *** CMake configuration failed
  error: subprocess-exited-with-error

  × Building wheel for torchcodec (pyproject.toml) did not run successfully.
  │ exit code: 1
  ╰─> No available output.

  note: This error originates from a subprocess, and is likely not a problem with pip.
  Building wheel for torchcodec (pyproject.toml): finished with status 'error'
  full command: /usr/bin/python3 /usr/lib/python3.15/site-packages/pip/_vendor/pyproject_hooks/_in_process/_in_process.py build_wheel /builddir/build/BUILD/python-torchcodec-0.16.0-build/.pyproject-builddir/tmpsa601rp6
  cwd: /builddir/build/BUILD/python-torchcodec-0.16.0-build/torchcodec-0.16.0
  ERROR: Failed building wheel for torchcodec

Comment 3 Peter Robinson 2026-10-29 16:02:22 UTC
(In reply to Neal Gompa from comment #2)
> I can't build it locally, so fedora-review continues to fail. This is the
> error:

Which release are you using locally? Presumably on x86 given the rocm failure.

Comment 4 Peter Robinson 2026-10-29 16:15:30 UTC
Actually that fails on rawhide/x86 in koji, but builds on rawhide/aarch64 and both on f45. There was a ROCm rebase recently.