[CPU] Bump test_complex_2d thresholds for LBFGS on complex64

[eqy]

Is this supposed to be bitwise identical? Wasn't sure how to interpret the comment but it seems to be giving mismatches like:

Mismatched elements: 1 / 2 (50.0%)
Greatest absolute difference: 4.6372413635253906e-05 at index (1,) (up to 1e-05 allowed)
Greatest relative difference: 3.4600801882334054e-05 at index (1,) (up to 1.3e-06 allowed)

To execute this test, run the following from the base repo dir:
     python test/test_optim.py -k test_complex_2d_LBFGS_cpu_complex64

on Neoverse-N2 SBSA ARM CPUs.

cc @vincentqb @jbschlosser @albanD @janeyx99 @crcrpar @jgong5 @mingfeima @XiaobingSuper @sanchitintel @ashokei @jingxu10 @ezyang @anjali411 @dylanbespalko @mruberry @lezcano @nikitaved @amjames

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/126358

• 📄 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.

✅ You can merge normally! (1 Unrelated Failure)

As of commit 3900c22 with merge base 636e799 ():

FLAKY - The following job failed but was likely due to flakiness present on trunk:

• pull / linux-focal-cuda12.1-py3.10-gcc9-sm86 / test (default, 2, 5, linux.g5.4xlarge.nvidia.gpu) (gh) (similar failure)
  test_foreach.py::TestForeachCUDA::test_binary_op_list_error_cases__foreach_add_cuda_bfloat16

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

[lezcano]

I reckon that one vectorises and the other one doesn't, hence the error, but would be good to confirm it.

[albanD]

What is the default tolerance for complex64? Souds like it should be at that level already?

[crcrpar]

rtol of 1.3e-6 and atol of 1e-5

ref: https://pytorch.org/docs/stable/testing.html#torch.testing.assert_close

[albanD]

Ho right! Sorry I confused myself with halving the bitlength. This is expected to match fp32.
The general rule here is that for single op, we expect these to hold. If you test a bunch of ops chained one after the other, then we might have to increase the tolerance yes.

[janeyx99]

https://github.com/pytorch/pytorch/blob/main/test/test_optim.py#L518-L527 for why it's not bitwise equal

[eqy]

@pytorchmergebot rebase

[pytorchmergebot]

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

[pytorchmergebot]

Successfully rebased eqy-patch-3 onto refs/remotes/origin/viable/strict, please pull locally before adding more changes (for example, via git checkout eqy-patch-3 && git pull --rebase)

[janeyx99]

@pytorchbot merge

[pytorchmergebot]

Merge started

Your change will be merged once all checks pass (ETA 0-4 Hours).

Learn more about merging in the wiki.

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

Advanced Debugging

Check the merge workflow status
here

[pytorchmergebot]

Merge failed

Reason: Not merging any PRs at the moment because there is a merge blocking https://github.com/pytorch/pytorch/labels/ci:%20sev issue open at:
#126896

Details for Dev Infra team

Raised by workflow job

[eqy]

@pytorchmergebot merge

[pytorchmergebot]

Merge started

Your change will be merged once all checks pass (ETA 0-4 Hours).

Learn more about merging in the wiki.

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

Advanced Debugging

Check the merge workflow status
here