-
Notifications
You must be signed in to change notification settings - Fork 608
Add fuse_dq_q_pass in exir/passes and also add it to HTP backend #2295
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/2295
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 1a5f23d with merge base 091e524 ( FLAKY - The following job failed but was likely due to flakiness present on trunk:
BROKEN TRUNK - The following job failed but was present on the merge base:👉 Rebase onto the `viable/strict` branch to avoid these failures
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
This pull request was exported from Phabricator. Differential Revision: D54543323 |
Summary: There are passes such as these https://fburl.com/code/vs6n4vcv that remove noops from the graph. The problem is that after this pass runs it still leaves in the dq->q nodes in the graph which then potentially get delegated to the backend causing perf regressions. This pass will remove the dq->q ops if their qparams are of the same value. If not it won't touch them. Differential Revision: D54543323
6a93f2d
to
f004294
Compare
This pull request was exported from Phabricator. Differential Revision: D54543323 |
1 similar comment
This pull request was exported from Phabricator. Differential Revision: D54543323 |
Summary: There are passes such as these https://fburl.com/code/vs6n4vcv that remove noops from the graph. The problem is that after this pass runs it still leaves in the dq->q nodes in the graph which then potentially get delegated to the backend causing perf regressions. This pass will remove the dq->q ops if their qparams are of the same value. If not it won't touch them. Differential Revision: D54543323
f004294
to
e860f35
Compare
e860f35
to
68d29f7
Compare
Summary: There are passes such as these https://fburl.com/code/vs6n4vcv that remove noops from the graph. The problem is that after this pass runs it still leaves in the dq->q nodes in the graph which then potentially get delegated to the backend causing perf regressions. This pass will remove the dq->q ops if their qparams are of the same value. If not it won't touch them. Differential Revision: D54543323
This pull request was exported from Phabricator. Differential Revision: D54543323 |
1 similar comment
This pull request was exported from Phabricator. Differential Revision: D54543323 |
Summary: There are passes such as these https://fburl.com/code/vs6n4vcv that remove noops from the graph. The problem is that after this pass runs it still leaves in the dq->q nodes in the graph which then potentially get delegated to the backend causing perf regressions. This pass will remove the dq->q ops if their qparams are of the same value. If not it won't touch them. Reviewed By: kimishpatel Differential Revision: D54543323
ec18ba3
to
7f74c06
Compare
This pull request was exported from Phabricator. Differential Revision: D54543323 |
Summary: There are passes such as these https://fburl.com/code/vs6n4vcv that remove noops from the graph. The problem is that after this pass runs it still leaves in the dq->q nodes in the graph which then potentially get delegated to the backend causing perf regressions. This pass will remove the dq->q ops if their qparams are of the same value. If not it won't touch them. Reviewed By: kimishpatel Differential Revision: D54543323
68d29f7
to
ec18ba3
Compare
Summary: There are passes such as these https://fburl.com/code/vs6n4vcv that remove noops from the graph. The problem is that after this pass runs it still leaves in the dq->q nodes in the graph which then potentially get delegated to the backend causing perf regressions. This pass will remove the dq->q ops if their qparams are of the same value. If not it won't touch them. Reviewed By: kimishpatel Differential Revision: D54543323
7f74c06
to
1e12ffa
Compare
This pull request was exported from Phabricator. Differential Revision: D54543323 |
1 similar comment
This pull request was exported from Phabricator. Differential Revision: D54543323 |
Summary: Pull Request resolved: #2295 There are passes such as these https://fburl.com/code/vs6n4vcv that remove noops from the graph. The problem is that after this pass runs it still leaves in the dq->q nodes in the graph which then potentially get delegated to the backend causing perf regressions. This pass will remove the dq->q ops if their qparams are of the same value. If not it won't touch them. Reviewed By: kimishpatel Differential Revision: D54543323
1e12ffa
to
ba03b6b
Compare
Summary: There are passes such as these https://fburl.com/code/vs6n4vcv that remove noops from the graph. The problem is that after this pass runs it still leaves in the dq->q nodes in the graph which then potentially get delegated to the backend causing perf regressions. This pass will remove the dq->q ops if their qparams are of the same value. If not it won't touch them. Reviewed By: kimishpatel Differential Revision: D54543323
ba03b6b
to
daafbc7
Compare
This pull request was exported from Phabricator. Differential Revision: D54543323 |
Summary: There are passes such as these https://fburl.com/code/vs6n4vcv that remove noops from the graph. The problem is that after this pass runs it still leaves in the dq->q nodes in the graph which then potentially get delegated to the backend causing perf regressions. This pass will remove the dq->q ops if their qparams are of the same value. If not it won't touch them. Reviewed By: kimishpatel Differential Revision: D54543323
daafbc7
to
fda6d5e
Compare
This pull request was exported from Phabricator. Differential Revision: D54543323 |
Summary: There are passes such as these https://fburl.com/code/vs6n4vcv that remove noops from the graph. The problem is that after this pass runs it still leaves in the dq->q nodes in the graph which then potentially get delegated to the backend causing perf regressions. This pass will remove the dq->q ops if their qparams are of the same value. If not it won't touch them. Reviewed By: kimishpatel Differential Revision: D54543323
fda6d5e
to
10e03b1
Compare
This pull request was exported from Phabricator. Differential Revision: D54543323 |
Summary: There are passes such as these https://fburl.com/code/vs6n4vcv that remove noops from the graph. The problem is that after this pass runs it still leaves in the dq->q nodes in the graph which then potentially get delegated to the backend causing perf regressions. This pass will remove the dq->q ops if their qparams are of the same value. If not it won't touch them. Reviewed By: kimishpatel Differential Revision: D54543323
This pull request was exported from Phabricator. Differential Revision: D54543323 |
10e03b1
to
1a5f23d
Compare
This pull request has been merged in 24477df. |
Summary:
There are passes such as these https://fburl.com/code/vs6n4vcv that remove noops from the graph. The problem is that after this pass runs it still leaves in the dq->q nodes in the graph which then potentially get delegated to the backend causing perf regressions.
This pass will remove the dq->q ops if their qparams are of the same value. If not it won't touch them.
Differential Revision: D54543323