[WIP] Refactor template matching #12351
Open
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
|
One or more of the the following people are requested to review this:
|
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Outdated suggestions cannot be applied.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.
Summary
Refactor the template matching transpiler pass achieving 4x speedup.
Details and comments
This PR refactors the template matching transpiler pass to incorporate the changes to
DAGDependencyV2and to inject a number of other efficiencies. This results in a 4x speedup with significantly reduced memory usage for large circuits.The PR incorporates the code from #12194 which requires
rustworkx0.15.0 to be released, so this should still be considered development code at this point.In spite of the speedup, the template matching algorithm which guarantees an optimal match is quadratic in the number of gates. I was able to run local tests on 3000 gates in about 2 min., but 10,000 gates would take 20 min or so. There are heuristics supplied in the original code, but these did not show significant improvement with the random clifford circuits I tested with.
There could be another round of optimizations to the code to try to develop better heuristics that would produce correct, but sometimes not optimal matches.