-
Notifications
You must be signed in to change notification settings - Fork 25.2k
[inductor][Optimus]Improve logging for group batch fusion #109314
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
Conversation
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/109314
Note: Links to docs will display an error until the docs builds have been completed. ✅ No FailuresAs of commit 800ef8f with merge base aed9bee ( This comment was automatically generated by Dr. CI and updates every 15 minutes. |
This pull request was exported from Phabricator. Differential Revision: D49284640 |
This PR needs a
|
d1783a2
to
02e4bbb
Compare
…9314) Summary: Log graph with Everpaste for debug and find more patterns to fuse Test Plan: to add logs Reviewed By: yanboliang Differential Revision: D49284640
This pull request was exported from Phabricator. Differential Revision: D49284640 |
02e4bbb
to
02378c8
Compare
This pull request was exported from Phabricator. Differential Revision: D49284640 |
02378c8
to
b5ce65c
Compare
This pull request was exported from Phabricator. Differential Revision: D49284640 |
…9314) Summary: Log graph with Everpaste for debug and find more patterns to fuse Test Plan: ``` [2023-09-14 22:26:53,449] [3/0] torch._inductor.fx_passes.group_batch_fusion: [INFO] Apply fusion BatchLinearFusion. Print graph: https://www.internalfb.com/intern/everpaste/?color=0&handle=GCfrCBNt9SP9x4MDAHouROPiHlkxbr0LAAAz [2023-09-14 22:26:54,817] [3/0] torch._inductor.fx_passes.group_batch_fusion: [INFO] Apply fusion BatchLinearLHSFusion. Print graph: https://www.internalfb.com/intern/everpaste/?color=0&handle=GP8UmhbRVpKo6DECAGZjVkwwnj0pbr0LAAAz [2023-09-14 22:26:54,897] [3/0] torch._inductor.fx_passes.group_batch_fusion: [INFO] BatchLayernormFusion: key = ('batch_layernorm', 'torch.Size([2048, 96])', 'torch.Size([96])', 'torch.Size([96])', '(96,)', '1e-05'); subset size = 60 [2023-09-14 22:26:54,982] [3/0] torch._inductor.fx_passes.group_batch_fusion: [INFO] BatchLayernormFusion: key = ('batch_layernorm', 'torch.Size([2048, 48])', 'torch.Size([48])', 'torch.Size([48])', '(48,)', '1e-05'); subset size = 10 [2023-09-14 22:26:56,271] [3/0] torch._inductor.fx_passes.group_batch_fusion: [INFO] Apply fusion BatchLayernormFusion. Print graph: https://www.internalfb.com/intern/everpaste/?color=0&handle=GGFLmBbZbscX2uoCAPOeQh04RQwrbr0LAAAz [2023-09-14 22:26:57,969] [3/0] torch._inductor.fx_passes.group_batch_fusion: [INFO] Apply fusion BatchTanhFusion. Print graph: https://www.internalfb.com/intern/everpaste/?color=0&handle=GEiljRbUagk50vkAAESJWARlM1Unbr0LAAAz [2023-09-14 22:26:58,194] [3/0] torch._inductor.fx_passes.split_cat: [WARNING] example value absent for node: cat_45 [2023-09-14 22:27:35,905] [3/0] torch._inductor.compile_fx: [INFO] Step 3: torchinductor compiling FORWARDS graph 0 [2023-09-14 22:27:41,707] [3/0] torch._inductor.fx_passes.group_batch_fusion: [INFO] Before group_batch fusion in post grads pass. Print graph: https://www.internalfb.com/intern/everpaste/?color=0&handle=GMa2RRYT4H7Vt78AAP4Th57LhxMZbr0LAAAz ``` Reviewed By: yanboliang Differential Revision: D49284640
b5ce65c
to
800ef8f
Compare
This pull request was exported from Phabricator. Differential Revision: D49284640 |
@pytorchbot merge (Initiating merge automatically since Phabricator Diff has merged) |
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 |
@pytorchbot revert -m="Diff reverted internally" -c="ghfirst" This Pull Request has been reverted by a revert inside Meta. To re-land this change, please open another pull request, assign the same reviewers, fix the CI failures that caused the revert and make sure that the failing CI runs on the PR by applying the proper ciflow label (e.g., ciflow/trunk).) |
@pytorchbot successfully started a revert job. Check the current status here. |
@jackiexu1992 your PR has been successfully reverted. |
…09314)" This reverts commit afad0d0. Reverted #109314 on behalf of https://github.com/facebook-github-bot due to Diff reverted internally ([comment](#109314 (comment)))
Summary: Log graph with Everpaste for debug and find more patterns to fuse
Test Plan: to add logs
Differential Revision: D49284640
cc @voznesenskym @penguinwu @EikanWang @jgong5 @Guobing-Chen @XiaobingSuper @zhuhaozhe @blzheng @Xia-Weiwen @wenzhe-nrv @jiayisunx @peterbell10 @ipiszy @ngimel @yf225 @chenyang78 @kadeng @muchulee8 @aakhundov