🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/142827

• 📄 Preview Python docs built from this PR
• 📄 Preview C++ docs built from this PR
• ❓ Need help or want to give feedback on the CI? Visit the bot commands wiki or our office hours

Note: Links to docs will display an error until the docs builds have been completed.

❗ 1 Active SEVs

There are 1 currently active SEVs. If your PR is affected, please view them below:

• Network outage on ROCm runners

❌ 1 New Failure, 3 Unrelated Failures

As of commit f6b347e with merge base 2b105de ():

This comment was automatically generated by Dr. CI and updates every 15 minutes.

Libtorch and manywheel docker images built successfully with PYTORCH_ROCM_ARCH not containing gfx900 or gfx906.

@pytorchbot merge -f "Unrelated CI failures". ROCm manywheel/libtorch docker images built successfully"

❌ 🤖 pytorchbot command failed:

Got EOF while in a quoted string```
Try `@pytorchbot --help` for more info.

@pytorchbot merge -f "Unrelated CI failures. ROCm manywheel/libtorch docker images built successfully"

Merge started

Your change will be merged immediately since you used the force (-f) flag, bypassing any CI checks (ETA: 1-5 minutes). Please use -f as last resort and instead consider -i/--ignore-current to continue the merge ignoring current failures. This will allow currently pending tests to finish and report signal before the merge.

Learn more about merging in the wiki.

Questions? Feedback? Please reach out to the PyTorch DevX Team

I can’t say anything other than that I’m disappointed AMD doesn’t want to compete with Nvidia on software support.

• Support your GPUs for 8+ years, like Nvidia does, including gfx906 GPUs ROCm/ROCm#3893

@pytorchbot revert

❌ 🤖 pytorchbot command failed:

@pytorchbot revert: error: the following arguments are required: -m/--message, -c/--classification

usage: @pytorchbot revert -m MESSAGE -c
                          {nosignal,ignoredsignal,landrace,weird,ghfirst}

Try @pytorchbot --help for more info.

@pytorchbot revert -m "prematurely dropped support for gfx900/gfx906" -c weird

@pytorchbot successfully started a revert job. Check the current status here.
Questions? Feedback? Please reach out to the PyTorch DevX Team

@jithunnair-amd your PR has been successfully reverted.

@jeffdaily @jithunnair-amd maybe pursue using offload compression support added to llvm recently as an alternative?

suggestion: generally try focus a bit on things other than forcibly reducing your user base.

#143986 added --offload-compress to our builds to help reduce our binary size without removing gfx arches.

There is effort underway to support generic targets, as well.

@jithunnair-amd Who cares about the increasing size of ROCm binaries? Have you listened to your users?

People want long-term support so that prosumer and consumer use can be fulfilled. The compute is there, 6700XT is a modern card, and you shouldn't be ending support so soon - especially since this is an argument for people to switch to NVIDIA, since they supported Kepler GPUs (released 2012) till September 2024.