-
Notifications
You must be signed in to change notification settings - Fork 751
[ET-VK][ez] Use XNNPACK's FuseBatchNorm pass #13600
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/pytorch/executorch/13600
Note: Links to docs will display an error until the docs builds have been completed. ⏳ No Failures, 96 PendingAs of commit d2ae976 with merge base 975a4a3 ( This comment was automatically generated by Dr. CI and updates every 15 minutes. |
|
This pull request was exported from Phabricator. Differential Revision: D80741735 |
This PR needs a
|
|
This pull request was exported from Phabricator. Differential Revision: D80741735 |
1 similar comment
|
This pull request was exported from Phabricator. Differential Revision: D80741735 |
#13598) Stack from [ghstack](https://github.com/ezyang/ghstack) (oldest at bottom): * #13597 * #13596 * #13595 * #13594 * #13593 * #13600 * #13599 * __->__ #13598 Title says it all. For reduce ops, their signature are not all alike so some extra legwork needs to be done to identify specific arguments that need to be checked. Also included a small update to partitioner logging to improve debuggability. Differential Revision: [D80741737](https://our.internmc.facebook.com/intern/diff/D80741737/) --------- Co-authored-by: ssjia <[email protected]>
Stack from [ghstack](https://github.com/ezyang/ghstack) (oldest at bottom): * #13597 * #13596 * #13595 * #13594 * #13593 * #13600 * __->__ #13599 * #13598 Title says it all! Differential Revision: [D80741734](https://our.internmc.facebook.com/intern/diff/D80741734/) --------- Co-authored-by: ssjia <[email protected]>
Pull Request resolved: #13600 As title. Use XNNPACK's FuseBatchNorm pass since it can fuse into linear layers as well. ghstack-source-id: 305143837 @exported-using-ghexport Differential Revision: [D80741735](https://our.internmc.facebook.com/intern/diff/D80741735/)
e9d9c04 to
d2ae976
Compare
Stack from [ghstack](https://github.com/ezyang/ghstack) (oldest at bottom): * #13597 * #13596 * #13595 * #13594 * __->__ #13593 * #13600 * #13599 * #13598 Differential Revision: [D80800081](https://our.internmc.facebook.com/intern/diff/D80800081) Co-authored-by: ssjia <[email protected]>
…code (#13594) Stack from [ghstack](https://github.com/ezyang/ghstack) (oldest at bottom): * #13597 * #13596 * #13595 * __->__ #13594 * #13593 * #13600 * #13599 * #13598 Differential Revision: [D80800085](https://our.internmc.facebook.com/intern/diff/D80800085) Co-authored-by: ssjia <[email protected]>
…out (#13595) Stack from [ghstack](https://github.com/ezyang/ghstack) (oldest at bottom): * #13597 * #13596 * __->__ #13595 * #13594 * #13593 * #13600 * #13599 * #13598 Differential Revision: [D80800082](https://our.internmc.facebook.com/intern/diff/D80800082) Co-authored-by: ssjia <[email protected]>
Stack from [ghstack](https://github.com/ezyang/ghstack) (oldest at bottom): * #13597 * __->__ #13596 * #13595 * #13594 * #13593 * #13600 * #13599 * #13598 Differential Revision: [D80800083](https://our.internmc.facebook.com/intern/diff/D80800083) Co-authored-by: ssjia <[email protected]>
…13597) Stack from [ghstack](https://github.com/ezyang/ghstack) (oldest at bottom): * __->__ #13597 * #13596 * #13595 * #13594 * #13593 * #13600 * #13599 * #13598 Differential Revision: [D80800084](https://our.internmc.facebook.com/intern/diff/D80800084) Co-authored-by: ssjia <[email protected]>
pytorch#13598) Stack from [ghstack](https://github.com/ezyang/ghstack) (oldest at bottom): * pytorch#13597 * pytorch#13596 * pytorch#13595 * pytorch#13594 * pytorch#13593 * pytorch#13600 * pytorch#13599 * __->__ pytorch#13598 Title says it all. For reduce ops, their signature are not all alike so some extra legwork needs to be done to identify specific arguments that need to be checked. Also included a small update to partitioner logging to improve debuggability. Differential Revision: [D80741737](https://our.internmc.facebook.com/intern/diff/D80741737/) --------- Co-authored-by: ssjia <[email protected]>
Stack from [ghstack](https://github.com/ezyang/ghstack) (oldest at bottom): * pytorch#13597 * pytorch#13596 * pytorch#13595 * pytorch#13594 * pytorch#13593 * pytorch#13600 * __->__ pytorch#13599 * pytorch#13598 Title says it all! Differential Revision: [D80741734](https://our.internmc.facebook.com/intern/diff/D80741734/) --------- Co-authored-by: ssjia <[email protected]>
Stack from [ghstack](https://github.com/ezyang/ghstack) (oldest at bottom): * pytorch#13597 * pytorch#13596 * pytorch#13595 * pytorch#13594 * pytorch#13593 * __->__ pytorch#13600 * pytorch#13599 * pytorch#13598 As title. Use XNNPACK's FuseBatchNorm pass since it can fuse into linear layers as well. Differential Revision: [D80741735](https://our.internmc.facebook.com/intern/diff/D80741735/) Co-authored-by: ssjia <[email protected]>
…#13593) Stack from [ghstack](https://github.com/ezyang/ghstack) (oldest at bottom): * pytorch#13597 * pytorch#13596 * pytorch#13595 * pytorch#13594 * __->__ pytorch#13593 * pytorch#13600 * pytorch#13599 * pytorch#13598 Differential Revision: [D80800081](https://our.internmc.facebook.com/intern/diff/D80800081) Co-authored-by: ssjia <[email protected]>
…code (pytorch#13594) Stack from [ghstack](https://github.com/ezyang/ghstack) (oldest at bottom): * pytorch#13597 * pytorch#13596 * pytorch#13595 * __->__ pytorch#13594 * pytorch#13593 * pytorch#13600 * pytorch#13599 * pytorch#13598 Differential Revision: [D80800085](https://our.internmc.facebook.com/intern/diff/D80800085) Co-authored-by: ssjia <[email protected]>
…out (pytorch#13595) Stack from [ghstack](https://github.com/ezyang/ghstack) (oldest at bottom): * pytorch#13597 * pytorch#13596 * __->__ pytorch#13595 * pytorch#13594 * pytorch#13593 * pytorch#13600 * pytorch#13599 * pytorch#13598 Differential Revision: [D80800082](https://our.internmc.facebook.com/intern/diff/D80800082) Co-authored-by: ssjia <[email protected]>
…rch#13596) Stack from [ghstack](https://github.com/ezyang/ghstack) (oldest at bottom): * pytorch#13597 * __->__ pytorch#13596 * pytorch#13595 * pytorch#13594 * pytorch#13593 * pytorch#13600 * pytorch#13599 * pytorch#13598 Differential Revision: [D80800083](https://our.internmc.facebook.com/intern/diff/D80800083) Co-authored-by: ssjia <[email protected]>
…ytorch#13597) Stack from [ghstack](https://github.com/ezyang/ghstack) (oldest at bottom): * __->__ pytorch#13597 * pytorch#13596 * pytorch#13595 * pytorch#13594 * pytorch#13593 * pytorch#13600 * pytorch#13599 * pytorch#13598 Differential Revision: [D80800084](https://our.internmc.facebook.com/intern/diff/D80800084) Co-authored-by: ssjia <[email protected]>
#13598) Stack from [ghstack](https://github.com/ezyang/ghstack) (oldest at bottom): * #13597 * #13596 * #13595 * #13594 * #13593 * #13600 * #13599 * __->__ #13598 Title says it all. For reduce ops, their signature are not all alike so some extra legwork needs to be done to identify specific arguments that need to be checked. Also included a small update to partitioner logging to improve debuggability. Differential Revision: [D80741737](https://our.internmc.facebook.com/intern/diff/D80741737/) --------- Co-authored-by: ssjia <[email protected]>
Stack from [ghstack](https://github.com/ezyang/ghstack) (oldest at bottom): * #13597 * #13596 * #13595 * #13594 * #13593 * #13600 * __->__ #13599 * #13598 Title says it all! Differential Revision: [D80741734](https://our.internmc.facebook.com/intern/diff/D80741734/) --------- Co-authored-by: ssjia <[email protected]>
Stack from [ghstack](https://github.com/ezyang/ghstack) (oldest at bottom): * #13597 * #13596 * #13595 * #13594 * #13593 * __->__ #13600 * #13599 * #13598 As title. Use XNNPACK's FuseBatchNorm pass since it can fuse into linear layers as well. Differential Revision: [D80741735](https://our.internmc.facebook.com/intern/diff/D80741735/) Co-authored-by: ssjia <[email protected]>
Stack from [ghstack](https://github.com/ezyang/ghstack) (oldest at bottom): * #13597 * #13596 * #13595 * #13594 * __->__ #13593 * #13600 * #13599 * #13598 Differential Revision: [D80800081](https://our.internmc.facebook.com/intern/diff/D80800081) Co-authored-by: ssjia <[email protected]>
…code (#13594) Stack from [ghstack](https://github.com/ezyang/ghstack) (oldest at bottom): * #13597 * #13596 * #13595 * __->__ #13594 * #13593 * #13600 * #13599 * #13598 Differential Revision: [D80800085](https://our.internmc.facebook.com/intern/diff/D80800085) Co-authored-by: ssjia <[email protected]>
…out (#13595) Stack from [ghstack](https://github.com/ezyang/ghstack) (oldest at bottom): * #13597 * #13596 * __->__ #13595 * #13594 * #13593 * #13600 * #13599 * #13598 Differential Revision: [D80800082](https://our.internmc.facebook.com/intern/diff/D80800082) Co-authored-by: ssjia <[email protected]>
Stack from [ghstack](https://github.com/ezyang/ghstack) (oldest at bottom): * #13597 * __->__ #13596 * #13595 * #13594 * #13593 * #13600 * #13599 * #13598 Differential Revision: [D80800083](https://our.internmc.facebook.com/intern/diff/D80800083) Co-authored-by: ssjia <[email protected]>
…13597) Stack from [ghstack](https://github.com/ezyang/ghstack) (oldest at bottom): * __->__ #13597 * #13596 * #13595 * #13594 * #13593 * #13600 * #13599 * #13598 Differential Revision: [D80800084](https://our.internmc.facebook.com/intern/diff/D80800084) Co-authored-by: ssjia <[email protected]>
Stack from ghstack (oldest at bottom):
BufferMetadataGLSL struct to abstract tensor layout #13595As title. Use XNNPACK's FuseBatchNorm pass since it can fuse into linear layers as well.
Differential Revision: D80741735