Relax tolerance for test_quick_baddbmm_cpu_complex64 #152424
Conversation
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/152424
Note: Links to docs will display an error until the docs builds have been completed. ✅ No FailuresAs of commit b489969 with merge base 6737e2c ( This comment was automatically generated by Dr. CI and updates every 15 minutes. |
pytorch-bot
bot
added
the
release notes: python_frontend
Flamefire
added
topic: not user facing
|
@pytorchbot merge |
Merge startedYour 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 |
Merge failedReason: 1 jobs have failed, first few of them are: trunk / linux-focal-rocm-py3.10 / test (default, 1, 2, linux.rocm.gpu.2) Details for Dev Infra teamRaised by workflow job |
|
This is the failure:
hanging in the ROCM setup step. Can we ignore that for the merge? |
|
@pytorchbot merge |
Merge startedYour 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 |
Merge failedReason: 1 jobs have failed, first few of them are: trunk / linux-focal-rocm-py3.10 / test (default, 1, 2, linux.rocm.gpu.2) Details for Dev Infra teamRaised by workflow job |
|
@Skylion007 The failure seems to be unrelated and there is no message what exactly failed. What shall we do with that? |
|
Looks like this PR hasn't been updated in a while so we're going to go ahead and mark this as |
|
Any updates here? |
|
@pytorchbot rebase |
|
@pytorchbot started a rebase job onto refs/remotes/origin/viable/strict. Check the current status here |
On Zen 2 (AMD EPYC) and Intel Sapphire Rapids this fails with small differences when compiled with native targeted optimizations. I.e. it fails with `-march=znver2` but succeeds with `-march=znver1`.
|
Successfully rebased |
pytorchmergebot
force-pushed
the
decomp-tolerance
branch
from
19c6b89 to
b489969
Compare
|
@pytorchbot merge |
Merge startedYour 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 |
On Zen 2 (AMD EPYC) and Intel Sapphire Rapids this fails with small differences when compiled with native targeted optimizations. I.e. it fails with `-march=znver2` but succeeds with `-march=znver1`. I assume some operator fusing is being used by GCC. Small differences like using `vmovdqa` can be seen in the minimized code of the baddbmm kernel: https://godbolt.org/z/jsxMa91Wb The greatest differences are consistent and the same on both CPU architectures: ``` Greatest absolute difference: 3.43852152582258e-05 at index (1, 2, 1) (up to 1e-05 allowed) Greatest relative difference: 3.6034286949870875e-06 at index (1, 2, 1) (up to 1.3e-06 allowed) ``` Hence I assume this is in the expected tolerances especially as `complex128` and all other types pass. Pull Request resolved: pytorch#152424 Approved by: https://github.com/malfet
On Zen 2 (AMD EPYC) and Intel Sapphire Rapids this fails with small differences when compiled with native targeted optimizations. I.e. it fails with `-march=znver2` but succeeds with `-march=znver1`. I assume some operator fusing is being used by GCC. Small differences like using `vmovdqa` can be seen in the minimized code of the baddbmm kernel: https://godbolt.org/z/jsxMa91Wb The greatest differences are consistent and the same on both CPU architectures: ``` Greatest absolute difference: 3.43852152582258e-05 at index (1, 2, 1) (up to 1e-05 allowed) Greatest relative difference: 3.6034286949870875e-06 at index (1, 2, 1) (up to 1.3e-06 allowed) ``` Hence I assume this is in the expected tolerances especially as `complex128` and all other types pass. Pull Request resolved: pytorch#152424 Approved by: https://github.com/malfet
On Zen 2 (AMD EPYC) and Intel Sapphire Rapids this fails with small differences when compiled with native targeted optimizations. I.e. it fails with `-march=znver2` but succeeds with `-march=znver1`. I assume some operator fusing is being used by GCC. Small differences like using `vmovdqa` can be seen in the minimized code of the baddbmm kernel: https://godbolt.org/z/jsxMa91Wb The greatest differences are consistent and the same on both CPU architectures: ``` Greatest absolute difference: 3.43852152582258e-05 at index (1, 2, 1) (up to 1e-05 allowed) Greatest relative difference: 3.6034286949870875e-06 at index (1, 2, 1) (up to 1.3e-06 allowed) ``` Hence I assume this is in the expected tolerances especially as `complex128` and all other types pass. Pull Request resolved: pytorch#152424 Approved by: https://github.com/malfet
5 participants
On Zen 2 (AMD EPYC) and Intel Sapphire Rapids this fails with small differences when compiled with native targeted optimizations. I.e. it fails with
-march=znver2but succeeds with-march=znver1.I assume some operator fusing is being used by GCC. Small differences like using
vmovdqacan be seen in the minimized code of the baddbmm kernel: https://godbolt.org/z/jsxMa91WbThe greatest differences are consistent and the same on both CPU architectures:
Hence I assume this is in the expected tolerances especially as
complex128and all other types pass.