ensure user modes are disabled when running aot_compile_joint_with_descriptors#169023
Open
bdhirsh wants to merge 17 commits intogh/bdhirsh/679/basefrom
Open
ensure user modes are disabled when running aot_compile_joint_with_descriptors#169023bdhirsh wants to merge 17 commits intogh/bdhirsh/679/basefrom
bdhirsh wants to merge 17 commits intogh/bdhirsh/679/basefrom
Conversation
…scriptors [ghstack-poisoned]
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/169023
Note: Links to docs will display an error until the docs builds have been completed. ✅ You can merge normally! (2 Unrelated Failures)As of commit 9e704ce with merge base c49edf7 ( FLAKY - The following job failed but was likely due to flakiness present on trunk:
UNSTABLE - The following job is marked as unstable, possibly due to flakiness on trunk:
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
yushangdi
approved these changes
Nov 24, 2025
…int_with_descriptors" In vanilla torch.compile, dynamo disables all active modes during compilation: [https://github.com/pytorch/pytorch/blob/main/torch/_dynamo/convert_frame.py#L2151](https://github.com/pytorch/pytorch/blob/main/torch/_dynamo/convert_frame.py?brid=t6D9McqWPJ2TbxhsrYj_dQ#L2151) When using the compile pipeline flow that splits graph capture from compilation, since we split graph capture and compilation into two stages, we only end up disabling modes during the graph capture step but not compilation. We already have a notion of an ExitStack, so one option would have been for me to arrange for the ExitStack to include a call to disable modes. I found it a bit unwieldy, though, since today only the AOTAutograd code handles the exit stack, and the context manager that I need is used by dynamo, so instead I just manually popped modes in `aot_compile_joint_with_descriptors` directly. [ghstack-poisoned]
Collaborator
Author
|
@pytorchbot merge |
…int_with_descriptors" In vanilla torch.compile, dynamo disables all active modes during compilation: [https://github.com/pytorch/pytorch/blob/main/torch/_dynamo/convert_frame.py#L2151](https://github.com/pytorch/pytorch/blob/main/torch/_dynamo/convert_frame.py?brid=t6D9McqWPJ2TbxhsrYj_dQ#L2151) When using the compile pipeline flow that splits graph capture from compilation, since we split graph capture and compilation into two stages, we only end up disabling modes during the graph capture step but not compilation. We already have a notion of an ExitStack, so one option would have been for me to arrange for the ExitStack to include a call to disable modes. I found it a bit unwieldy, though, since today only the AOTAutograd code handles the exit stack, and the context manager that I need is used by dynamo, so instead I just manually popped modes in `aot_compile_joint_with_descriptors` directly. [ghstack-poisoned]
…int_with_descriptors" In vanilla torch.compile, dynamo disables all active modes during compilation: [https://github.com/pytorch/pytorch/blob/main/torch/_dynamo/convert_frame.py#L2151](https://github.com/pytorch/pytorch/blob/main/torch/_dynamo/convert_frame.py?brid=t6D9McqWPJ2TbxhsrYj_dQ#L2151) When using the compile pipeline flow that splits graph capture from compilation, since we split graph capture and compilation into two stages, we only end up disabling modes during the graph capture step but not compilation. We already have a notion of an ExitStack, so one option would have been for me to arrange for the ExitStack to include a call to disable modes. I found it a bit unwieldy, though, since today only the AOTAutograd code handles the exit stack, and the context manager that I need is used by dynamo, so instead I just manually popped modes in `aot_compile_joint_with_descriptors` directly. [ghstack-poisoned]
Collaborator
Author
|
@pytorchbot merge |
Collaborator
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 |
Collaborator
Merge failedReason: 1 jobs have failed, first few of them are: trunk / linux-jammy-rocm-py3.10 / test (default, 3, 6, linux.rocm.gpu.gfx942.1) Details for Dev Infra teamRaised by workflow job |
…int_with_descriptors" In vanilla torch.compile, dynamo disables all active modes during compilation: [https://github.com/pytorch/pytorch/blob/main/torch/_dynamo/convert_frame.py#L2151](https://github.com/pytorch/pytorch/blob/main/torch/_dynamo/convert_frame.py?brid=t6D9McqWPJ2TbxhsrYj_dQ#L2151) When using the compile pipeline flow that splits graph capture from compilation, since we split graph capture and compilation into two stages, we only end up disabling modes during the graph capture step but not compilation. We already have a notion of an ExitStack, so one option would have been for me to arrange for the ExitStack to include a call to disable modes. I found it a bit unwieldy, though, since today only the AOTAutograd code handles the exit stack, and the context manager that I need is used by dynamo, so instead I just manually popped modes in `aot_compile_joint_with_descriptors` directly. [ghstack-poisoned]
tiendatngcs
pushed a commit
to tiendatngcs/pytorch-Dec25
that referenced
this pull request
Dec 10, 2025
…scriptors ghstack-source-id: 30f140a Pull Request resolved: pytorch/pytorch#169023
…int_with_descriptors" In vanilla torch.compile, dynamo disables all active modes during compilation: [https://github.com/pytorch/pytorch/blob/main/torch/_dynamo/convert_frame.py#L2151](https://github.com/pytorch/pytorch/blob/main/torch/_dynamo/convert_frame.py?brid=t6D9McqWPJ2TbxhsrYj_dQ#L2151) When using the compile pipeline flow that splits graph capture from compilation, since we split graph capture and compilation into two stages, we only end up disabling modes during the graph capture step but not compilation. We already have a notion of an ExitStack, so one option would have been for me to arrange for the ExitStack to include a call to disable modes. I found it a bit unwieldy, though, since today only the AOTAutograd code handles the exit stack, and the context manager that I need is used by dynamo, so instead I just manually popped modes in `aot_compile_joint_with_descriptors` directly. [ghstack-poisoned]
…int_with_descriptors" In vanilla torch.compile, dynamo disables all active modes during compilation: [https://github.com/pytorch/pytorch/blob/main/torch/_dynamo/convert_frame.py#L2151](https://github.com/pytorch/pytorch/blob/main/torch/_dynamo/convert_frame.py?brid=t6D9McqWPJ2TbxhsrYj_dQ#L2151) When using the compile pipeline flow that splits graph capture from compilation, since we split graph capture and compilation into two stages, we only end up disabling modes during the graph capture step but not compilation. We already have a notion of an ExitStack, so one option would have been for me to arrange for the ExitStack to include a call to disable modes. I found it a bit unwieldy, though, since today only the AOTAutograd code handles the exit stack, and the context manager that I need is used by dynamo, so instead I just manually popped modes in `aot_compile_joint_with_descriptors` directly. [ghstack-poisoned]
…int_with_descriptors" In vanilla torch.compile, dynamo disables all active modes during compilation: [https://github.com/pytorch/pytorch/blob/main/torch/_dynamo/convert_frame.py#L2151](https://github.com/pytorch/pytorch/blob/main/torch/_dynamo/convert_frame.py?brid=t6D9McqWPJ2TbxhsrYj_dQ#L2151) When using the compile pipeline flow that splits graph capture from compilation, since we split graph capture and compilation into two stages, we only end up disabling modes during the graph capture step but not compilation. We already have a notion of an ExitStack, so one option would have been for me to arrange for the ExitStack to include a call to disable modes. I found it a bit unwieldy, though, since today only the AOTAutograd code handles the exit stack, and the context manager that I need is used by dynamo, so instead I just manually popped modes in `aot_compile_joint_with_descriptors` directly. [ghstack-poisoned]
…int_with_descriptors" In vanilla torch.compile, dynamo disables all active modes during compilation: [https://github.com/pytorch/pytorch/blob/main/torch/_dynamo/convert_frame.py#L2151](https://github.com/pytorch/pytorch/blob/main/torch/_dynamo/convert_frame.py?brid=t6D9McqWPJ2TbxhsrYj_dQ#L2151) When using the compile pipeline flow that splits graph capture from compilation, since we split graph capture and compilation into two stages, we only end up disabling modes during the graph capture step but not compilation. We already have a notion of an ExitStack, so one option would have been for me to arrange for the ExitStack to include a call to disable modes. I found it a bit unwieldy, though, since today only the AOTAutograd code handles the exit stack, and the context manager that I need is used by dynamo, so instead I just manually popped modes in `aot_compile_joint_with_descriptors` directly. [ghstack-poisoned]
…int_with_descriptors" In vanilla torch.compile, dynamo disables all active modes during compilation: [https://github.com/pytorch/pytorch/blob/main/torch/_dynamo/convert_frame.py#L2151](https://github.com/pytorch/pytorch/blob/main/torch/_dynamo/convert_frame.py?brid=t6D9McqWPJ2TbxhsrYj_dQ#L2151) When using the compile pipeline flow that splits graph capture from compilation, since we split graph capture and compilation into two stages, we only end up disabling modes during the graph capture step but not compilation. We already have a notion of an ExitStack, so one option would have been for me to arrange for the ExitStack to include a call to disable modes. I found it a bit unwieldy, though, since today only the AOTAutograd code handles the exit stack, and the context manager that I need is used by dynamo, so instead I just manually popped modes in `aot_compile_joint_with_descriptors` directly. [ghstack-poisoned]
…int_with_descriptors" In vanilla torch.compile, dynamo disables all active modes during compilation: [https://github.com/pytorch/pytorch/blob/main/torch/_dynamo/convert_frame.py#L2151](https://github.com/pytorch/pytorch/blob/main/torch/_dynamo/convert_frame.py?brid=t6D9McqWPJ2TbxhsrYj_dQ#L2151) When using the compile pipeline flow that splits graph capture from compilation, since we split graph capture and compilation into two stages, we only end up disabling modes during the graph capture step but not compilation. We already have a notion of an ExitStack, so one option would have been for me to arrange for the ExitStack to include a call to disable modes. I found it a bit unwieldy, though, since today only the AOTAutograd code handles the exit stack, and the context manager that I need is used by dynamo, so instead I just manually popped modes in `aot_compile_joint_with_descriptors` directly. [ghstack-poisoned]
…int_with_descriptors" In vanilla torch.compile, dynamo disables all active modes during compilation: [https://github.com/pytorch/pytorch/blob/main/torch/_dynamo/convert_frame.py#L2151](https://github.com/pytorch/pytorch/blob/main/torch/_dynamo/convert_frame.py?brid=t6D9McqWPJ2TbxhsrYj_dQ#L2151) When using the compile pipeline flow that splits graph capture from compilation, since we split graph capture and compilation into two stages, we only end up disabling modes during the graph capture step but not compilation. We already have a notion of an ExitStack, so one option would have been for me to arrange for the ExitStack to include a call to disable modes. I found it a bit unwieldy, though, since today only the AOTAutograd code handles the exit stack, and the context manager that I need is used by dynamo, so instead I just manually popped modes in `aot_compile_joint_with_descriptors` directly. [ghstack-poisoned]
…int_with_descriptors" In vanilla torch.compile, dynamo disables all active modes during compilation: [https://github.com/pytorch/pytorch/blob/main/torch/_dynamo/convert_frame.py#L2151](https://github.com/pytorch/pytorch/blob/main/torch/_dynamo/convert_frame.py?brid=t6D9McqWPJ2TbxhsrYj_dQ#L2151) When using the compile pipeline flow that splits graph capture from compilation, since we split graph capture and compilation into two stages, we only end up disabling modes during the graph capture step but not compilation. We already have a notion of an ExitStack, so one option would have been for me to arrange for the ExitStack to include a call to disable modes. I found it a bit unwieldy, though, since today only the AOTAutograd code handles the exit stack, and the context manager that I need is used by dynamo, so instead I just manually popped modes in `aot_compile_joint_with_descriptors` directly. [ghstack-poisoned]
…int_with_descriptors" In vanilla torch.compile, dynamo disables all active modes during compilation: [https://github.com/pytorch/pytorch/blob/main/torch/_dynamo/convert_frame.py#L2151](https://github.com/pytorch/pytorch/blob/main/torch/_dynamo/convert_frame.py?brid=t6D9McqWPJ2TbxhsrYj_dQ#L2151) When using the compile pipeline flow that splits graph capture from compilation, since we split graph capture and compilation into two stages, we only end up disabling modes during the graph capture step but not compilation. We already have a notion of an ExitStack, so one option would have been for me to arrange for the ExitStack to include a call to disable modes. I found it a bit unwieldy, though, since today only the AOTAutograd code handles the exit stack, and the context manager that I need is used by dynamo, so instead I just manually popped modes in `aot_compile_joint_with_descriptors` directly. [ghstack-poisoned]
…int_with_descriptors" In vanilla torch.compile, dynamo disables all active modes during compilation: [https://github.com/pytorch/pytorch/blob/main/torch/_dynamo/convert_frame.py#L2151](https://github.com/pytorch/pytorch/blob/main/torch/_dynamo/convert_frame.py?brid=t6D9McqWPJ2TbxhsrYj_dQ#L2151) When using the compile pipeline flow that splits graph capture from compilation, since we split graph capture and compilation into two stages, we only end up disabling modes during the graph capture step but not compilation. We already have a notion of an ExitStack, so one option would have been for me to arrange for the ExitStack to include a call to disable modes. I found it a bit unwieldy, though, since today only the AOTAutograd code handles the exit stack, and the context manager that I need is used by dynamo, so instead I just manually popped modes in `aot_compile_joint_with_descriptors` directly. [ghstack-poisoned]
…int_with_descriptors" In vanilla torch.compile, dynamo disables all active modes during compilation: [https://github.com/pytorch/pytorch/blob/main/torch/_dynamo/convert_frame.py#L2151](https://github.com/pytorch/pytorch/blob/main/torch/_dynamo/convert_frame.py?brid=t6D9McqWPJ2TbxhsrYj_dQ#L2151) When using the compile pipeline flow that splits graph capture from compilation, since we split graph capture and compilation into two stages, we only end up disabling modes during the graph capture step but not compilation. We already have a notion of an ExitStack, so one option would have been for me to arrange for the ExitStack to include a call to disable modes. I found it a bit unwieldy, though, since today only the AOTAutograd code handles the exit stack, and the context manager that I need is used by dynamo, so instead I just manually popped modes in `aot_compile_joint_with_descriptors` directly. [ghstack-poisoned]
…int_with_descriptors" In vanilla torch.compile, dynamo disables all active modes during compilation: [https://github.com/pytorch/pytorch/blob/main/torch/_dynamo/convert_frame.py#L2151](https://github.com/pytorch/pytorch/blob/main/torch/_dynamo/convert_frame.py?brid=t6D9McqWPJ2TbxhsrYj_dQ#L2151) When using the compile pipeline flow that splits graph capture from compilation, since we split graph capture and compilation into two stages, we only end up disabling modes during the graph capture step but not compilation. We already have a notion of an ExitStack, so one option would have been for me to arrange for the ExitStack to include a call to disable modes. I found it a bit unwieldy, though, since today only the AOTAutograd code handles the exit stack, and the context manager that I need is used by dynamo, so instead I just manually popped modes in `aot_compile_joint_with_descriptors` directly. [ghstack-poisoned]
Collaborator
|
Starting merge as part of PR stack under #169295 |
1 similar comment
Collaborator
|
Starting merge as part of PR stack under #169295 |
SergeyTyshkevich
pushed a commit
to SergeyTyshkevich/chart2
that referenced
this pull request
Jan 19, 2026
…scriptors ghstack-source-id: 2230e15 Pull Request resolved: pytorch/pytorch#169023
Contributor
|
Looks like this PR hasn't been updated in a while so we're going to go ahead and mark this as |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
In vanilla torch.compile, dynamo disables all active modes during compilation: https://github.com/pytorch/pytorch/blob/main/torch/_dynamo/convert_frame.py#L2151
When using the compile pipeline flow that splits graph capture from compilation, since we split graph capture and compilation into two stages, we only end up disabling modes during the graph capture step but not compilation.
We already have a notion of an ExitStack, so one option would have been for me to arrange for the ExitStack to include a call to disable modes. I found it a bit unwieldy, though, since today only the AOTAutograd code handles the exit stack, and the context manager that I need is used by dynamo, so instead I just manually popped modes in
aot_compile_joint_with_descriptorsdirectly.Stack from ghstack (oldest at bottom):