-
Notifications
You must be signed in to change notification settings - Fork 26.3k
[compiled autograd] Compiled autograd configs in TLS #137821
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
[ghstack-poisoned]
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/137821
Note: Links to docs will display an error until the docs builds have been completed. ✅ No FailuresAs of commit 3ce556a with merge base 8f3efb8 ( This comment was automatically generated by Dr. CI and updates every 15 minutes. |
cc XilunWu H-Huang awgu kwen2501 wanchaol fegin fduwjj wz337 wconstab d4l3k c-p-i-o voznesenskym penguinwu EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng wenzhe-nrv jiayisunx chenyang78 kadeng chauhang amjames rec [ghstack-poisoned]
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.
Please commit the suggested changes from pytorch's linter.
cc XilunWu H-Huang awgu kwen2501 wanchaol fegin fduwjj wz337 wconstab d4l3k c-p-i-o voznesenskym penguinwu EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng wenzhe-nrv jiayisunx chenyang78 kadeng chauhang amjames rec [ghstack-poisoned]
cc XilunWu H-Huang awgu kwen2501 wanchaol fegin fduwjj wz337 wconstab d4l3k c-p-i-o voznesenskym penguinwu EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng wenzhe-nrv jiayisunx chenyang78 kadeng chauhang amjames rec [ghstack-poisoned]
cc XilunWu H-Huang awgu kwen2501 wanchaol fegin fduwjj wz337 wconstab d4l3k c-p-i-o voznesenskym penguinwu EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng wenzhe-nrv jiayisunx chenyang78 kadeng chauhang amjames rec [ghstack-poisoned]
cc XilunWu H-Huang awgu kwen2501 wanchaol fegin fduwjj wz337 wconstab d4l3k c-p-i-o voznesenskym penguinwu EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng wenzhe-nrv jiayisunx chenyang78 kadeng chauhang amjames rec [ghstack-poisoned]
cc XilunWu H-Huang awgu kwen2501 wanchaol fegin fduwjj wz337 wconstab d4l3k c-p-i-o voznesenskym penguinwu EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng wenzhe-nrv jiayisunx chenyang78 kadeng chauhang amjames rec [ghstack-poisoned]
Multithreaded doesn't work yet, this adds python side TLS for the configs passing cc XilunWu H-Huang awgu kwen2501 wanchaol fegin fduwjj wz337 wconstab d4l3k c-p-i-o voznesenskym penguinwu EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng wenzhe-nrv jiayisunx chenyang78 kadeng chauhang amjames rec [ghstack-poisoned]
Multithreaded doesn't work yet, this adds python side TLS for the configs passing cc XilunWu H-Huang awgu kwen2501 wanchaol fegin fduwjj wz337 wconstab d4l3k c-p-i-o voznesenskym penguinwu EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng wenzhe-nrv jiayisunx ipiszy yf225 chenyang78 kadeng muchulee8 ColinPeppler amjames desertfire chauhang rec [ghstack-poisoned]
…enabled` global state variable" After #137821, we will no longer be able to call the Compiled Autograd state getter under Dynamo tracing. One solution is to cache the "Compiled Autograd enabled" state outside of compile for FSDP2, and just read from the cache when we need the check. This is implemented by this PR. Fixes #138177. cc XilunWu H-Huang awgu kwen2501 wanchaol fegin fduwjj wz337 wconstab d4l3k c-p-i-o voznesenskym penguinwu EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng wenzhe-nrv jiayisunx chenyang78 kadeng chauhang amjames rec xmfan [ghstack-poisoned]
…state variable" After #137821, we will no longer be able to call the Compiled Autograd state getter under Dynamo tracing. One solution is to cache the "Compiled Autograd enabled" state outside of compile for FSDP2, and just read from the cache when we need the check. This is implemented by this PR. Fixes #138177. cc XilunWu H-Huang awgu kwen2501 wanchaol fegin fduwjj wz337 wconstab d4l3k c-p-i-o voznesenskym penguinwu EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng wenzhe-nrv jiayisunx chenyang78 kadeng chauhang amjames rec xmfan [ghstack-poisoned]
…enabled` global state variable" After #137821, we will no longer be able to call the Compiled Autograd state getter under Dynamo tracing. One solution is to cache the "Compiled Autograd enabled" state outside of compile for FSDP2, and just read from the cache when we need the check. This is implemented by this PR. Fixes #138177. cc XilunWu H-Huang awgu kwen2501 wanchaol fegin fduwjj wz337 wconstab d4l3k c-p-i-o voznesenskym penguinwu EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng wenzhe-nrv jiayisunx chenyang78 kadeng chauhang amjames rec xmfan [ghstack-poisoned]
…state variable" After #137821, we will no longer be able to call the Compiled Autograd state getter under Dynamo tracing. One solution is to cache the "Compiled Autograd enabled" state outside of compile for FSDP2, and just read from the cache when we need the check. This is implemented by this PR. Fixes #138177. cc XilunWu H-Huang awgu kwen2501 wanchaol fegin fduwjj wz337 wconstab d4l3k c-p-i-o voznesenskym penguinwu EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng wenzhe-nrv jiayisunx chenyang78 kadeng chauhang amjames rec xmfan [ghstack-poisoned]
…enabled` global state variable" After #137821, we will no longer be able to call the Compiled Autograd state getter under Dynamo tracing. One solution is to cache the "Compiled Autograd enabled" state outside of compile for FSDP2, and just read from the cache when we need the check. This is implemented by this PR. Fixes #138177. cc XilunWu H-Huang awgu kwen2501 wanchaol fegin fduwjj wz337 wconstab d4l3k c-p-i-o zhaojuanmao mrshenli rohan-varma chauhang voznesenskym penguinwu EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng wenzhe-nrv jiayisunx chenyang78 kadeng amjames rec xmfan [ghstack-poisoned]
…state variable" After #137821, we will no longer be able to call the Compiled Autograd state getter under Dynamo tracing. One solution is to cache the "Compiled Autograd enabled" state outside of compile for FSDP2, and just read from the cache when we need the check. This is implemented by this PR. Fixes #138177. cc XilunWu H-Huang awgu kwen2501 wanchaol fegin fduwjj wz337 wconstab d4l3k c-p-i-o zhaojuanmao mrshenli rohan-varma chauhang voznesenskym penguinwu EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng wenzhe-nrv jiayisunx chenyang78 kadeng amjames rec xmfan [ghstack-poisoned]
…ble (#138187) After #137821, we will no longer be able to call the Compiled Autograd state getter under Dynamo tracing. One solution is to cache the "Compiled Autograd enabled" state outside of compile for FSDP2, and just read from the cache when we need the check. This is implemented by this PR. Fixes #138177. Pull Request resolved: #138187 Approved by: https://github.com/xmfan, https://github.com/awgu ghstack dependencies: #138245, #138174
…enabled` global state variable" After #137821, we will no longer be able to call the Compiled Autograd state getter under Dynamo tracing. One solution is to cache the "Compiled Autograd enabled" state outside of compile for FSDP2, and just read from the cache when we need the check. This is implemented by this PR. Fixes #138177. cc XilunWu H-Huang awgu kwen2501 wanchaol fegin fduwjj wz337 wconstab d4l3k c-p-i-o zhaojuanmao mrshenli rohan-varma chauhang voznesenskym penguinwu EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng wenzhe-nrv jiayisunx chenyang78 kadeng amjames rec xmfan [ghstack-poisoned]
…state variable" After #137821, we will no longer be able to call the Compiled Autograd state getter under Dynamo tracing. One solution is to cache the "Compiled Autograd enabled" state outside of compile for FSDP2, and just read from the cache when we need the check. This is implemented by this PR. Fixes #138177. cc XilunWu H-Huang awgu kwen2501 wanchaol fegin fduwjj wz337 wconstab d4l3k c-p-i-o zhaojuanmao mrshenli rohan-varma chauhang voznesenskym penguinwu EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng wenzhe-nrv jiayisunx chenyang78 kadeng amjames rec xmfan [ghstack-poisoned]
…enabled` global state variable" After #137821, we will no longer be able to call the Compiled Autograd state getter under Dynamo tracing. One solution is to cache the "Compiled Autograd enabled" state outside of compile for FSDP2, and just read from the cache when we need the check. This is implemented by this PR. Fixes #138177. cc XilunWu H-Huang awgu kwen2501 wanchaol fegin fduwjj wz337 wconstab d4l3k c-p-i-o zhaojuanmao mrshenli rohan-varma chauhang voznesenskym penguinwu EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng wenzhe-nrv jiayisunx chenyang78 kadeng amjames rec xmfan [ghstack-poisoned]
…state variable" After #137821, we will no longer be able to call the Compiled Autograd state getter under Dynamo tracing. One solution is to cache the "Compiled Autograd enabled" state outside of compile for FSDP2, and just read from the cache when we need the check. This is implemented by this PR. Fixes #138177. cc XilunWu H-Huang awgu kwen2501 wanchaol fegin fduwjj wz337 wconstab d4l3k c-p-i-o zhaojuanmao mrshenli rohan-varma chauhang voznesenskym penguinwu EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng wenzhe-nrv jiayisunx chenyang78 kadeng amjames rec xmfan [ghstack-poisoned]
…enabled` global state variable" After #137821, we will no longer be able to call the Compiled Autograd state getter under Dynamo tracing. One solution is to cache the "Compiled Autograd enabled" state outside of compile for FSDP2, and just read from the cache when we need the check. This is implemented by this PR. Fixes #138177. cc XilunWu H-Huang awgu kwen2501 wanchaol fegin fduwjj wz337 wconstab d4l3k c-p-i-o zhaojuanmao mrshenli rohan-varma chauhang voznesenskym penguinwu EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng wenzhe-nrv jiayisunx chenyang78 kadeng amjames rec xmfan [ghstack-poisoned]
…state variable" After #137821, we will no longer be able to call the Compiled Autograd state getter under Dynamo tracing. One solution is to cache the "Compiled Autograd enabled" state outside of compile for FSDP2, and just read from the cache when we need the check. This is implemented by this PR. Fixes #138177. cc XilunWu H-Huang awgu kwen2501 wanchaol fegin fduwjj wz337 wconstab d4l3k c-p-i-o zhaojuanmao mrshenli rohan-varma chauhang voznesenskym penguinwu EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng wenzhe-nrv jiayisunx chenyang78 kadeng amjames rec xmfan [ghstack-poisoned]
…ble (#138187) After #137821, we will no longer be able to call the Compiled Autograd state getter under Dynamo tracing. One solution is to cache the "Compiled Autograd enabled" state outside of compile for FSDP2, and just read from the cache when we need the check. This is implemented by this PR. Fixes #138177. Pull Request resolved: #138187 Approved by: https://github.com/xmfan, https://github.com/awgu
Multithreaded doesn't work yet, this adds python side TLS only for the python side state cc XilunWu H-Huang awgu kwen2501 wanchaol fegin fduwjj wz337 wconstab d4l3k c-p-i-o voznesenskym penguinwu EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng wenzhe-nrv jiayisunx ipiszy yf225 chenyang78 kadeng muchulee8 ColinPeppler amjames desertfire chauhang rec [ghstack-poisoned]
Multithreaded doesn't work yet, this adds python side TLS only for the python side state cc XilunWu H-Huang awgu kwen2501 wanchaol fegin fduwjj wz337 wconstab d4l3k c-p-i-o voznesenskym penguinwu EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng wenzhe-nrv jiayisunx ipiszy yf225 chenyang78 kadeng muchulee8 ColinPeppler amjames desertfire chauhang rec [ghstack-poisoned]
Pull Request resolved: #138061 Approved by: https://github.com/yf225 ghstack dependencies: #137953, #137821
Summary: Multithreaded doesn't work yet, this adds python side TLS only for the python side state X-link: pytorch/pytorch#137821 Approved by: https://github.com/jansel, https://github.com/yf225 ghstack dependencies: #137953 Reviewed By: wdvr Differential Revision: D64796212 Pulled By: xmfan fbshipit-source-id: aa1d9ef8f6e61207dfb352866e37d5e7cc98df42
Multithreaded doesn't work yet, this adds python side TLS only for the python side state Pull Request resolved: #137821 Approved by: https://github.com/jansel, https://github.com/yf225 ghstack dependencies: #137953
Pull Request resolved: #138061 Approved by: https://github.com/yf225 ghstack dependencies: #137953, #137821
…out "[compiled autograd] Compiled autograd configs in TLS (#137821)" Summary: Original commit changeset: 9bf80c1492d7 Original Phabricator Diff: D64796226 Original commit changeset: aa1d9ef8f6e6 Original Phabricator Diff: D64796212 Differential Revision: D65072644
…out "[compiled autograd] Compiled autograd configs in TLS (#137821)" (#139086) Summary: Original commit changeset: 9bf80c1492d7 Original Phabricator Diff: D64796226 Original commit changeset: aa1d9ef8f6e6 Original Phabricator Diff: D64796212 Differential Revision: D65072644 Pull Request resolved: #139086 Approved by: https://github.com/malfet
|
This pull request has been reverted by fd9f4e6. To re-land this change, please open another pull request, assignthe 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). |
…d Back out "[compiled autograd] Compiled autograd configs in TLS (pytorch#137821)" (pytorch#139086) Summary: Original commit changeset: 9bf80c1492d7 Original Phabricator Diff: D64796226 Original commit changeset: aa1d9ef8f6e6 Original Phabricator Diff: D64796212 Differential Revision: D65072644 Pull Request resolved: pytorch#139086 Approved by: https://github.com/malfet
…d Back out "[compiled autograd] Compiled autograd configs in TLS (pytorch#137821)" (pytorch#139086) Summary: Original commit changeset: 9bf80c1492d7 Original Phabricator Diff: D64796226 Original commit changeset: aa1d9ef8f6e6 Original Phabricator Diff: D64796212 Differential Revision: D65072644 Pull Request resolved: pytorch#139086 Approved by: https://github.com/malfet
ghstack-source-id: d70c60a Pull Request resolved: pytorch/pytorch#137821
Multithreaded doesn't work yet, this adds python side TLS only for the python side state
Stack from ghstack (oldest at bottom):
cc @XilunWu @H-Huang @awgu @kwen2501 @wanchaol @fegin @fduwjj @wz337 @wconstab @d4l3k @c-p-i-o @voznesenskym @penguinwu @EikanWang @jgong5 @Guobing-Chen @XiaobingSuper @zhuhaozhe @blzheng @wenzhe-nrv @jiayisunx @ipiszy @yf225 @chenyang78 @kadeng @muchulee8 @ColinPeppler @amjames @desertfire @chauhang @rec