[Unity][MetaSchedule] BlockCollector focusing on current func #14595
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
PR apache#11999 introduces the sugar method `work_on` to TIR Schedule, with a field `func_working_on_` newly added to the ScheduleNode. In some cases we may want to know which function a ScheduleNode is working on, which is not supported previously. Therefore, this PR introduces a method to ScheduleNode that returns the function (more accurately, GlobalVar) currently being worked on. With this we are able to know the function being worked on.
PR apache#14182 introduces the default GPU schedule pass, which makes uses of the BlockCollector in Meta-Schedule. When requested, the BlockCollector goes through every function in the input IRModule, while in the case of applying default GPU schedule where we apply schedule to one PrimFunc at a time, going through every PrimFunc is wasteful. Therefore, this PR enhances the BlockCollector so that it only focuses on the function being worked on.
Collaborator
|
Thanks for contributing to TVM! Please refer to the contributing guidelines https://tvm.apache.org/docs/contribute/ for useful information and tips. Please request code reviews from Reviewers by @-ing them in a comment.
Generated by tvm-bot |
1 similar comment
Collaborator
|
Thanks for contributing to TVM! Please refer to the contributing guidelines https://tvm.apache.org/docs/contribute/ for useful information and tips. Please request code reviews from Reviewers by @-ing them in a comment.
Generated by tvm-bot |
Contributor
Author
|
Depending on #14594 |
junrushao
approved these changes
Apr 12, 2023
Member
junrushao
left a comment
junrushao
left a comment
There was a problem hiding this comment.
LGTM! Perhaps we should send both PRs to mainline.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 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.
PR #14182 introduces the default GPU schedule pass, which makes uses
of the BlockCollector in Meta-Schedule. When requested, the
BlockCollector goes through every function in the input IRModule,
while in the case of applying default GPU schedule where we apply
schedule to one PrimFunc at a time, going through every PrimFunc is
wasteful.
Therefore, this PR enhances the BlockCollector so that it only focuses
on the function being worked on.