-
Notifications
You must be signed in to change notification settings - Fork 21.7k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Merge Inductor perf smoke test with other inductor CI tests #93395
Conversation
Summary: Now the smoke test can also be triggered with the ciflow/inductor label. [ghstack-poisoned]
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/93395
Note: Links to docs will display an error until the docs builds have been completed. ❌ 1 FailuresAs of commit 45c5cb5: NEW FAILURES - The following jobs have failed:
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
Summary: Now the smoke test can also be triggered with the ciflow/inductor label. ghstack-source-id: b785a72c295336fb534d22d65b5fc0c81d79ba4f Pull Request resolved: #93395
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM. I will keep an eye on the queue time of A100, and increase runners if needed.
@malfet, let me know if this is what you had in mind when you said "running on master". |
@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 mandatory check(s) failed (Rule Dig deeper by viewing the failures on hud Details for Dev Infra teamRaised by workflow job |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Would there be enough capacity?
@pytorchbot merge -f "Doesn't really affect trunk, but newly added config is green" |
Merge startedYour change will be merged immediately since you used the force (-f) flag, bypassing any CI checks (ETA: 1-5 minutes). Learn more about merging in the wiki. Questions? Feedback? Please reach out to the PyTorch DevX Team |
The standalone action was already running for every push to the master, the capacity is enough. |
Perf smoke test takes ~1min, it's very little compared to all the other accuracy sweeps we are running |
Good point, we might just have identified a BE work. This smoke test job spent most of its time checking out torchbench. There might be a way to skip that. |
…n-dev-setup * origin: (898 commits) Move dynamo.optimizations.distributed to backends (pytorch#93408) Remove cuda 11.6 from nightly (pytorch#93979) Refactor dynamo register_backend/BACKENDS (pytorch#93389) Remove cuda 11.6 from CI replace with 11.7 (pytorch#93406) [Dynamo] Rename `GuardBuilder.guarded_code` -> `check_fn_manager` (pytorch#93934) Revert "Remove CUDA 11.6 from nightly builds (pytorch#93404)" Revert "[inductor] fix crash issue when input is a view tensor (pytorch#90150)" Basic Validation for FSDP `state_dict` transformations of modules with persistent buffers (pytorch#93396) Merge Inductor perf smoke test with other inductor CI tests (pytorch#93395) [inductor] Don't import torchvision (pytorch#93027) [FSDP][3/N] Refactor `summon_full_params` unit tests (pytorch#92298) [FSDP][2/N] `_summon_full_params` -> `_unshard_params` (pytorch#92297) Remove CUDA 11.6 from nightly builds (pytorch#93404) Mark buffers that reuse other buffers (pytorch#93329) Refactor to allow reuse of SchedulerNode.allocate (pytorch#93328) retire sparse_mask_helper (pytorch#91714) update fbgemm third party (pytorch#93907) [inductor] fix crash issue when input is a view tensor (pytorch#90150) [Inductor] add config for weight prepacking (pytorch#93811) Check for none for NNModuleVariable.__module__ (pytorch#93326) ...
Addresses (#93395 (comment)) The perf smoke test is supposed to be around one minute. But the torchbench checkout process is taking more than 15 minutes. This PR explores a way to just checkout torchbench with only needed models that are later used to do perf smoke test and memory compression ratio check. Torchbench installation has "python install.py models model1 model 2 model3" support to just install model1 model2 and model3, not providing "models model1 model2 model3" would install all models by default. Before this PR, inductor job takes about 27 minutes (21 minutes spent in testing phase) https://github.com/pytorch/pytorch/actions/runs/4149154553/jobs/7178024253 After this PR, inductor job takes about 19 minutes (12 minutes spent in testing phase), pytorch checkout and docker image pull takes about 5 - 6 minutes total. https://github.com/pytorch/pytorch/actions/runs/4149155814/jobs/7178735494 Pull Request resolved: #94578 Approved by: https://github.com/orionr, https://github.com/malfet, https://github.com/desertfire
Stack from ghstack (oldest at bottom):
Summary: Now the smoke test can also be triggered with the
ciflow/inductor label.