Package Details: python-torchaudio 2.3.0-1

Git Clone URL: https://aur.archlinux.org/python-torchaudio.git (read-only, click to copy)
Package Base: python-torchaudio
Description: Data manipulation and transformation for audio signal processing, powered by PyTorch
Upstream URL: https://github.com/pytorch/audio
Licenses: BSD
Conflicts: python-torchaudio-git
Submitter: HenryJia
Maintainer: NBonaparte
Last Packager: NBonaparte
Votes: 5
Popularity: 0.000228
First Submitted: 2020-08-27 18:23 (UTC)
Last Updated: 2024-04-29 05:17 (UTC)

Latest Comments

« First ‹ Previous 1 2 3 Next › Last »

yan12125 commented on 2022-05-25 05:16 (UTC)

I switched to GCC 11 and it works with /opt/cuda. Maybe worth another try with /usr/local/cuda :)

rien333 commented on 2022-05-16 12:08 (UTC) (edited on 2022-05-16 12:09 (UTC) by rien333)

Hmm that's strange. community/cuda installs to /opt/cuda, but your logs mentioned /usr/local/cuda. Maybe some other packages or scripts installs them for you.

It's true that cuda installs to /opt, as confirmed by yay -Ql cuda. However, when I sudo rm -rf /usr/local/cuda, then yay -R cuda, and finally reinstall using yay -S cuda, /usr/local/cuda appears again! It's not a symlink to /opt/cuda, but the contents of the two folders appear to be identical (certainly, nvcc -V is the same in both folders). Hence, I think we've stumbled on a bug (a totally extraneous duplicate is installed). I'll report it in a bit.

Perhaps the hack(s) you describe don't quite apply to /usr/local/cuda, which would explain why my build keeps failing.

Good catch noticing /usr/local/cuda in my logs!

yan12125 commented on 2022-05-16 11:28 (UTC)

never messed around with installing nvidia stuff another way.

Hmm that's strange. community/cuda installs to /opt/cuda, but your logs mentioned /usr/local/cuda. Maybe some other packages or scripts installs them for you.

That this may be he root of the problem seems to accord with various reports on the web. https://gitlab.kitware.com/cmake/cmake/-/issues/23322

Thanks for the link! I have a guess now: nvcc inside /usr/local/cuda rejects gcc 12 and causes a mysterious CMake error. On the other hand, community/cuda comes with a hack [1] to accept whatever gcc versions, so CMake does not complain.

Are you sure cmake picks up on the env variable you've changed?

Before I added USE_CUDA=0, I got an error invalid type argument of unary ‘*’ when compiling compute_alphas.cu, and now it's gone for me. Somehow CMake still checks for CUDA even with USE_CUDA=0. Maybe that flag only disables invoking nvcc for compiling *.cu files.

[1] https://github.com/archlinux/svntogit-community/blob/packages/cuda/trunk/PKGBUILD#L124

rien333 commented on 2022-05-16 09:03 (UTC) (edited on 2022-05-16 09:05 (UTC) by rien333)

I have community/cuda installed, never messed around with installing nvidia stuff another way. I reinstalled it for the fun of it, but no luck.

CUDA 11.7 ... is still incompatible with GCC 12

That this may be he root of the problem seems to accord with various reports on the web. See https://gitlab.kitware.com/cmake/cmake/-/issues/23322

I disabled CUDA altogether for now

Perhaps this doesn't quite work yet? Are you sure cmake picks up on the env variable you've changed?

yan12125 commented on 2022-05-16 03:32 (UTC)

To be honest, I got a different error than yours and I thought my fix works for both, and that seems not the case. Anyway, your CUDA copy inside /usr/local/cuda seems broken. Could you try removing /usr/local/cuda, installing https://archlinux.org/packages/community/x86_64/cuda/ and see if that works?

rien333 commented on 2022-05-15 16:38 (UTC)

I still get the same error. Was the package properly updated? Like did you make a new release and stuff?

yan12125 commented on 2022-05-12 13:45 (UTC)

rien333: With some changes to PKGBUILD, this error can be avoided, but you're right - I tested nvcc from CUDA 11.7, and it is still incompatible with GCC 12. I disabled CUDA altogether for now.

rien333 commented on 2022-05-12 07:06 (UTC) (edited on 2022-05-12 07:13 (UTC) by rien333)

I'm getting the following build error. I think the PKGBUILD is missing a dependency or something. Googling shows something about nvcc currently being incompatible with certain gcc versions? Arch is on like 12 now.

- Caffe2: Protobuf version 3.20.1
-- Found CUDA: /usr/local/cuda (found version "11.6")
-- The CUDA compiler identification is unknown
CMake Error at /usr/share/cmake/Modules/CMakeDetermineCUDACompiler.cmake:633 (message):
  Failed to detect a default CUDA architecture.



  Compiler output:

Call Stack (most recent call first):
  /usr/lib64/cmake/Caffe2/public/cuda.cmake:41 (enable_language)
  /usr/lib64/cmake/Caffe2/Caffe2Config.cmake:88 (include)
  /usr/lib64/cmake/Torch/TorchConfig.cmake:68 (find_package)
  CMakeLists.txt:87 (find_package)


-- Configuring incomplete, errors occurred!

yan12125 commented on 2022-05-11 04:38 (UTC)

Thanks for the update and adding me as a co-maintainer. Seems new torchaudio has fixed the version string and my previous BUILD_VERSION trick is no longer necessary. How do you think about using arch=('x86_64' 'i686')? (I can push it directly, but I prefer a consensus)

yan12125 commented on 2022-05-03 10:30 (UTC)

FWIW, here is a working patch for updating to 0.11.0 with various fixes: http://fars.ee/v4qE

Also, may I become a co-maintainer of this package? I'd like to keep this package up-to-date as best as I can.

« First ‹ Previous 1 2 3 Next › Last »