[PASS] StorageRewrite Fold Inplace op storage when possible #759
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
Member
Author
tqchen
added a commit
to tqchen/tvm
that referenced
this pull request
Jul 6, 2018
* [PASS] StorageRewrite Fold Inplace op storage when possible * update comment to fix typos
sergei-mironov
pushed a commit
to sergei-mironov/tvm
that referenced
this pull request
Aug 8, 2018
* [PASS] StorageRewrite Fold Inplace op storage when possible * update comment to fix typos
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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Outdated suggestions cannot be applied.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.
Resolves #746
Detect pattern
dst[index] = f(src[index])and fold src with dst when possibleKnown limitation:
the current detection algorithm cannot handle the case when a location in an array is written multiple times. For example, the following program will pass the check, but we cannot make A and B to be the same array.
The high-level code generator needs to ensure that the generated code only write each location of the target once. This is the case with the current schedule.
This limitation is documented in the comment
So this pass works fine for now. To enhance the safety, we might be able to use another pass to detect some sufficient condition when an array is written once in each location, which requires some work but can be useful to make it work for an arbitrary segment of IR.