JIT: fix count reconstruction when a natural loop contains an improper loop #100449
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
…r loop If a natural loop contains an improper loop, the cyclic probability computation for the natural loop will be an understimate, as the cyclic probability computation assumes one pass convergence. In such cases count reconstruction may report convergence when it has not in fact converged, as any natural loop header ignores flow from its backedges, assuming their impact has been accounted for by the cyclic probability. So when a loop contains improper loops, fall back to normal iterative computation for the loop. We could use the cyclic probability initially as a convergence accelerator, but would need to switch over to not using it to guarantee full convergence. But that complicates the logic and these cases are rare.
dotnet-issue-labeler
bot
added
the
area-CodeGen-coreclr
|
Tagging subscribers to this area: @JulieLeeMSFT, @jakobbotsch |
|
@jakobbotsch @BruceForstall PTAL Locally one perf score diff (no asm diffs). Found while scouting changes for block count consistency after profile incorporation, with more widespread use of count repair than we do now. |
51 tasks
BruceForstall
approved these changes
Mar 29, 2024
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
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Outdated suggestions cannot be applied.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.
If a natural loop contains an improper loop, the cyclic probability computation for the natural loop will be an underestimate, as the cyclic probability computation assumes one pass convergence.
In such cases count reconstruction may report convergence when it has not in fact converged, as any natural loop header ignores flow from its back edges, assuming their impact has been accounted for by the cyclic probability.
So, when a loop contains improper loops, fall back to normal iterative computation for the loop. We could use the cyclic probability initially as a convergence accelerator but would need to switch over to not using it to guarantee full convergence. But that complicates the logic, and these cases are rare.