🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/107339

• 📄 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.

❌ 130 New Failures, 6 Unrelated Failures

As of commit 59b327c with merge base 5531a23 ():

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

This PR needs a release notes: label

If your changes are user facing and intended to be a part of release notes, please use a label starting with release notes:.

If not, please add the topic: not user facing label.

To add a label, you can comment to pytorchbot, for example
@pytorchbot label "topic: not user facing"

For more information, see
https://github.com/pytorch/pytorch/wiki/PyTorch-AutoLabel-Bot#why-categorize-for-release-notes-and-how-does-it-work.

@pytorchbot label ciflow/periodic

❌ 🤖 pytorchbot command failed:

@pytorchbot: error: argument command: invalid choice: 'tag' (choose from 'merge', 'revert', 'rebase', 'label', 'drci')

usage: @pytorchbot [-h] {merge,revert,rebase,label,drci} ...

Try @pytorchbot --help for more info.

@pytorchbot label ciflow/slow

@pytorchbot label ciflow/trunk

@pytorchbot label ciflow/inductor

@izaitsevfb has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator.

I've imported the PR to get the signal from internal meta builds ahead of the merge. Will let you know the results as soon as they are ready.

Unfortunately, the internal builds are still failing. I'm trying to see how to forward fix them them on meta's side.

Please don't merge yet, I'm still working on the internal build fix.

@izaitsevfb What is the situation?

@izaitsevfb What is the situation?

I'm sorry for the delay, unfortunately, the fix is not trivial. We don't have a precedent yet of including third-party libraries into c10 core internally, and my attempts so far were unsuccessful.

But I'm still on it, and I'll try to solve this EoW. If not, I'll escalate.

@pytorchbot label ciflow/binaries

@izaitsevfb If it is hard to add cpuinfo to c10, may we remove the cpuinfo test temporarily and try that in another PR?

@cyyever

@izaitsevfb If it is hard to add cpuinfo to c10, may we remove the cpuinfo test temporarily and try that in another PR?

I think I finally found everything that needs to be patched internally.

The following changes should be made to your PR (I can do this myself if you're ok with that) (see #109079):

• in c10/ovrsource_defs.bzl "//third-party/cpuinfo:cpuinfo", should be added.
• in c10/core/build.bzl I changed dependency to ":cpuinfo", but your original variant might also work (testing this now).

I'll try to merge this PR tomorrow. Apologies again for the delay.

@izaitsevfb Thank you so much for the efforts. I will add the changes.

@pytorchbot started a rebase job onto refs/remotes/origin/viable/strict. Check the current status here

Successfully rebased cpu_core_count2 onto refs/remotes/origin/viable/strict, please pull locally before adding more changes (for example, via git checkout cpu_core_count2 && git pull --rebase)

@izaitsevfb has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator.

Ok, unfortunately :cpuinfo in c10/core/build.bzl doesn't work in OSS bazel build (I missed the build failures on my test PR, as they are suppressed as flaky). I'm trying to figure out the hack to have a consolidated dependency syntax that works both internally and in OSS.

@izaitsevfb I had a local bazel build and "//third_party/cpuinfo" could work.

@izaitsevfb I had a local bazel build and "//third_party/cpuinfo" could work.

Yep, that's the way to go. I found how to support this variant internally as well, and all internal tests are passing now on the patched version.

So the plan is following (mostly for the record):

• import the PR
• patch the diff with internal-only changes (they are already prepared)
• land both versions simultaneously in a co-dev style

@izaitsevfb has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator.

The corresponding diff was merged internally, so our tooling should merge this PR shortly (if it doesn't, I'll force-merge it).

@pytorchbot merge -f 'Landed internally as D48443523'

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

@izaitsevfb Thank you!

@cyyever thank you for you patience, that was quite an adventure :)