[Inductor] Add the possible fusions group by priority #123067
Closed
Conversation
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
[ghstack-poisoned]
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/123067
Note: Links to docs will display an error until the docs builds have been completed. ✅ You can merge normally! (4 Unrelated Failures)As of commit b7a05be with merge base d8717c2 ( FLAKY - The following jobs failed but were likely due to flakiness present on trunk:
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
leslie-fang-intel
changed the title
cc voznesenskym penguinwu EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng wenzhe-nrv jiayisunx peterbell10 ipiszy yf225 chenyang78 kadeng muchulee8 aakhundov ColinPeppler amjames desertfire chauhang [ghstack-poisoned]
cc voznesenskym penguinwu EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng wenzhe-nrv jiayisunx peterbell10 ipiszy yf225 chenyang78 kadeng muchulee8 aakhundov ColinPeppler amjames desertfire chauhang [ghstack-poisoned]
cc voznesenskym penguinwu EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng wenzhe-nrv jiayisunx peterbell10 ipiszy yf225 chenyang78 kadeng muchulee8 aakhundov ColinPeppler amjames desertfire chauhang [ghstack-poisoned]
|
Hi @lezcano @jgong5, for the discussion in #121625 (comment). Here is one possible abstract comes to my mind. Could you kindly help to take a look? |
lezcano
approved these changes
Apr 2, 2024
**Summary** Refactor the `Scheduler.fuse_nodes` changes in #121625. In the previous implementation of `Scheduler.fuse_nodes` in #121625, we use the `enable_outer_loop_fusion` context to ensure `OuterLoopFusion` happens after all the norm fusions. And there is a discussion in https://github.com/pytorch/pytorch/pull/121625/files#r1527177141 to reuse current `score_fusion` mechanism. However, given that [fuse_nodes](https://github.com/pytorch/pytorch/blob/f4ff063c333f286d4384523bac67c047aca4d7b9/torch/_inductor/scheduler.py#L1679-L1698) will invoke `fuse_nodes_once` 10 times. We are concerned that the score approach may potentially disrupt pairs of regular fusion nodes in the 2rd invocation of `fuse_nodes_once` if they have been pick up by the outer loop fusion in the 1st invocation of `fuse_nodes_once`. In this PR, we propose adding an abstract of `filter_possible_fusions_by_priority`. In each invoking of `fuse_nodes_once`, the possible fusions will be grouped by their priority from the backend. And only the group of possible fusions with highest priority will be fused in this invocation. In this way, we can ensure `OuterLoopFusion` happens after all the norm fusions. cc voznesenskym penguinwu EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng wenzhe-nrv jiayisunx peterbell10 ipiszy yf225 chenyang78 kadeng muchulee8 aakhundov ColinPeppler amjames desertfire chauhang [ghstack-poisoned]
**Summary** Refactor the `Scheduler.fuse_nodes` changes in #121625. In the previous implementation of `Scheduler.fuse_nodes` in #121625, we use the `enable_outer_loop_fusion` context to ensure `OuterLoopFusion` happens after all the norm fusions. And there is a discussion in https://github.com/pytorch/pytorch/pull/121625/files#r1527177141 to reuse current `score_fusion` mechanism. However, given that [fuse_nodes](https://github.com/pytorch/pytorch/blob/f4ff063c333f286d4384523bac67c047aca4d7b9/torch/_inductor/scheduler.py#L1679-L1698) will invoke `fuse_nodes_once` 10 times. We are concerned that the score approach may potentially disrupt pairs of regular fusion nodes in the 2rd invocation of `fuse_nodes_once` if they have been pick up by the outer loop fusion in the 1st invocation of `fuse_nodes_once`. In this PR, we propose adding an abstract of `filter_possible_fusions_by_priority`. In each invoking of `fuse_nodes_once`, the possible fusions will be grouped by their priority from the backend. And only the group of possible fusions with highest priority will be fused in this invocation. In this way, we can ensure `OuterLoopFusion` happens after all the norm fusions. cc voznesenskym penguinwu EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng wenzhe-nrv jiayisunx peterbell10 ipiszy yf225 chenyang78 kadeng muchulee8 aakhundov ColinPeppler amjames desertfire chauhang [ghstack-poisoned]
jgong5
approved these changes
Apr 4, 2024
**Summary** Refactor the `Scheduler.fuse_nodes` changes in #121625. In the previous implementation of `Scheduler.fuse_nodes` in #121625, we use the `enable_outer_loop_fusion` context to ensure `OuterLoopFusion` happens after all the norm fusions. And there is a discussion in https://github.com/pytorch/pytorch/pull/121625/files#r1527177141 to reuse current `score_fusion` mechanism. However, given that [fuse_nodes](https://github.com/pytorch/pytorch/blob/f4ff063c333f286d4384523bac67c047aca4d7b9/torch/_inductor/scheduler.py#L1679-L1698) will invoke `fuse_nodes_once` 10 times. We are concerned that the score approach may potentially disrupt pairs of regular fusion nodes in the 2rd invocation of `fuse_nodes_once` if they have been pick up by the outer loop fusion in the 1st invocation of `fuse_nodes_once`. In this PR, we propose adding an abstract of `filter_possible_fusions_by_priority`. In each invoking of `fuse_nodes_once`, the possible fusions will be grouped by their priority from the backend. And only the group of possible fusions with highest priority will be fused in this invocation. In this way, we can ensure `OuterLoopFusion` happens after all the norm fusions. cc voznesenskym penguinwu EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng wenzhe-nrv jiayisunx peterbell10 ipiszy yf225 chenyang78 kadeng muchulee8 aakhundov ColinPeppler amjames desertfire chauhang [ghstack-poisoned]
**Summary** Refactor the `Scheduler.fuse_nodes` changes in #121625. In the previous implementation of `Scheduler.fuse_nodes` in #121625, we use the `enable_outer_loop_fusion` context to ensure `OuterLoopFusion` happens after all the norm fusions. And there is a discussion in https://github.com/pytorch/pytorch/pull/121625/files#r1527177141 to reuse current `score_fusion` mechanism. However, given that [fuse_nodes](https://github.com/pytorch/pytorch/blob/f4ff063c333f286d4384523bac67c047aca4d7b9/torch/_inductor/scheduler.py#L1679-L1698) will invoke `fuse_nodes_once` 10 times. We are concerned that the score approach may potentially disrupt pairs of regular fusion nodes in the 2rd invocation of `fuse_nodes_once` if they have been pick up by the outer loop fusion in the 1st invocation of `fuse_nodes_once`. In this PR, we propose adding an abstract of `filter_possible_fusions_by_priority`. In each invoking of `fuse_nodes_once`, the possible fusions will be grouped by their priority from the backend. And only the group of possible fusions with highest priority will be fused in this invocation. In this way, we can ensure `OuterLoopFusion` happens after all the norm fusions. cc voznesenskym penguinwu EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng wenzhe-nrv jiayisunx peterbell10 ipiszy yf225 chenyang78 kadeng muchulee8 aakhundov ColinPeppler amjames desertfire chauhang [ghstack-poisoned]
**Summary** Refactor the `Scheduler.fuse_nodes` changes in #121625. In the previous implementation of `Scheduler.fuse_nodes` in #121625, we use the `enable_outer_loop_fusion` context to ensure `OuterLoopFusion` happens after all the norm fusions. And there is a discussion in https://github.com/pytorch/pytorch/pull/121625/files#r1527177141 to reuse current `score_fusion` mechanism. However, given that [fuse_nodes](https://github.com/pytorch/pytorch/blob/f4ff063c333f286d4384523bac67c047aca4d7b9/torch/_inductor/scheduler.py#L1679-L1698) will invoke `fuse_nodes_once` 10 times. We are concerned that the score approach may potentially disrupt pairs of regular fusion nodes in the 2rd invocation of `fuse_nodes_once` if they have been pick up by the outer loop fusion in the 1st invocation of `fuse_nodes_once`. In this PR, we propose adding an abstract of `filter_possible_fusions_by_priority`. In each invoking of `fuse_nodes_once`, the possible fusions will be grouped by their priority from the backend. And only the group of possible fusions with highest priority will be fused in this invocation. In this way, we can ensure `OuterLoopFusion` happens after all the norm fusions. cc voznesenskym penguinwu EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng wenzhe-nrv jiayisunx peterbell10 ipiszy yf225 chenyang78 kadeng muchulee8 aakhundov ColinPeppler amjames desertfire chauhang [ghstack-poisoned]
|
@pytorchbot merge |
Merge failedReason: This PR needs a If not, please add the To add a label, you can comment to pytorchbot, for example For more information, see Details for Dev Infra teamRaised by workflow job |
|
@pytorchbot merge |
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 |
sanketpurandare
pushed a commit
to sanketpurandare/pytorch
that referenced
this pull request
Apr 22, 2024
**Summary** Refactor the `Scheduler.fuse_nodes` changes in pytorch#121625. In the previous implementation of `Scheduler.fuse_nodes` in pytorch#121625, we use the `enable_outer_loop_fusion` context to ensure `OuterLoopFusion` happens after all the norm fusions. And there is a discussion in https://github.com/pytorch/pytorch/pull/121625/files#r1527177141 to reuse current `score_fusion` mechanism. However, given that [fuse_nodes](https://github.com/pytorch/pytorch/blob/f4ff063c333f286d4384523bac67c047aca4d7b9/torch/_inductor/scheduler.py#L1679-L1698) will invoke `fuse_nodes_once` 10 times. We are concerned that the score approach may potentially disrupt pairs of regular fusion nodes in the 2rd invocation of `fuse_nodes_once` if they have been pick up by the outer loop fusion in the 1st invocation of `fuse_nodes_once`. In this PR, we propose adding an abstract of `filter_possible_fusions_by_priority`. In each invoking of `fuse_nodes_once`, the possible fusions will be grouped by their priority from the backend. And only the group of possible fusions with highest priority will be fused in this invocation. In this way, we can ensure `OuterLoopFusion` happens after all the norm fusions. Pull Request resolved: pytorch#123067 Approved by: https://github.com/lezcano, https://github.com/jgong5 ghstack dependencies: pytorch#121625
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
5 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.
Stack from ghstack (oldest at bottom):
Summary
Refactor the
Scheduler.fuse_nodeschanges in #121625. In the previous implementation ofScheduler.fuse_nodesin #121625, we use theenable_outer_loop_fusioncontext to ensureOuterLoopFusionhappens after all the norm fusions.And there is a discussion in https://github.com/pytorch/pytorch/pull/121625/files#r1527177141 to reuse current
score_fusionmechanism. However, given that fuse_nodes will invokefuse_nodes_once10 times. We are concerned that the score approach may potentially disrupt pairs of regular fusion nodes in the 2rd invocation offuse_nodes_onceif they have been pick up by the outer loop fusion in the 1st invocation offuse_nodes_once.In this PR, we propose adding an abstract of
filter_possible_fusions_by_priority. In each invoking offuse_nodes_once, the possible fusions will be grouped by their priority from the backend. And only the group of possible fusions with highest priority will be fused in this invocation. In this way, we can ensureOuterLoopFusionhappens after all the norm fusions.cc @voznesenskym @penguinwu @EikanWang @jgong5 @Guobing-Chen @XiaobingSuper @zhuhaozhe @blzheng @wenzhe-nrv @jiayisunx @peterbell10 @ipiszy @yf225 @chenyang78 @kadeng @muchulee8 @aakhundov @ColinPeppler @amjames @desertfire @chauhang