Implement detensorize-scf pass (for / if / while) #1075
Merged
Conversation
This file contains 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
vincentmr
changed the title
erick-xanadu
requested changes
Sep 5, 2024
There was a problem hiding this comment.
Just a few general comments:
- We can separate this into multiple functions
- If we can, I think we can separate this into two PatternOpRewriters, one for the
IfOp, and one for theForOp - The walkers will walk in post-order. I think you may be able to get access to the
YieldOpfor the current op by looking at the terminator of the basic block of the op you are looking at. This could clean the logic behindfirst_yield.
|
Hi Erick, thanks for the feedback.
|
dime10
reviewed
Sep 6, 2024
dime10
reviewed
Sep 6, 2024
dime10
reviewed
Sep 13, 2024
Co-authored-by: David Ittah <[email protected]>
Co-authored-by: David Ittah <[email protected]>
Codecov ReportAll modified and coverable lines are covered by tests ✅
Additional details and impacted files@@ Coverage Diff @@
## main #1075 +/- ##
=======================================
Coverage 97.87% 97.87%
=======================================
Files 76 76
Lines 10847 10847
Branches 1282 1282
=======================================
Hits 10616 10616
Misses 179 179
Partials 52 52 ☔ View full report in Codecov by Sentry. |
erick-xanadu
requested changes
Sep 20, 2024
rauletorresc
pushed a commit
that referenced
this pull request
Oct 9, 2024
**Context:**
After applying `func.func(linalg-detensorize{aggressive-mode})`,
operations in the SCF dialect still request scalar tensors for their
operands/arguments and yield scalar tensor results as well. This leads
to the introduction of several tensor.from_elements and tensor.extract
operations around the operation interface.
**Description of the Change:**
Introduce pass to detensorize `scf::for / if / while`. The while
detensorization is currently limited to
[cases](https://github.com/PennyLaneAI/catalyst/pull/1075/files#diff-f6ee410945b4f019017fb45560a6ae38d9568439e3933ff7aa1673b1094ecffbR217-R225)
where the before/after blocks have the same operands.
**Benefits:**
**Possible Drawbacks:**
**Related GitHub Issues:**
[sc-71505]
---------
Co-authored-by: David Ittah <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
Context:
After applying
func.func(linalg-detensorize{aggressive-mode}), operations in the SCF dialect still request scalar tensors for their operands/arguments and yield scalar tensor results as well. This leads to the introduction of several tensor.from_elements and tensor.extract operations around the operation interface.Description of the Change:
Introduce pass to detensorize
scf::for / if / while. The while detensorization is currently limited to cases where the before/after blocks have the same operands.Benefits:
Possible Drawbacks:
Related GitHub Issues:
[sc-71505]