Calibrate the CS-vs-FA precision tolerances from their measured spread - #1291
Merged
Merged
Conversation
Each UTCKKSRNSCSvsFA test compares one COMPOSITESCALINGAUTO run against one FLEXIBLEAUTO run with unseeded randomness, so the precision gap is a random variable. The shared 3.0-bit tolerance sat only ~2.4 standard deviations above the mean gap for the three scenarios with a positive mean, which failed about one run in 200; the failures were not CS precision losses but FA reference runs that happened to land high. Measure the gap distribution per test (200 runs each, 52 for the two expensive scenarios) and set each tolerance at least 5 standard deviations above its own mean and at most half the pre-fix gap it guards against, so every test stays both stable and sensitive. Hold each tolerance in a named kGapTol constant and build the failure message from it, so the value can no longer desync between the assertion and its message.
yspolyakov
approved these changes
Sep 8, 2026
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Each UTCKKSRNSCSvsFA test compares one COMPOSITESCALINGAUTO run against one FLEXIBLEAUTO run with unseeded randomness, so the precision gap is a random variable. The shared 3.0-bit tolerance sat only ~2.4 standard deviations above the mean gap for the three scenarios with a positive mean, which failed about one run in 200; the failures were not CS precision losses but FA reference runs that happened to land high.
Measure the gap distribution per test (200 runs each, 52 for the two expensive scenarios) and set each tolerance at least 5 standard deviations above its own mean and at most half the pre-fix gap it guards against, so every test stays both stable and sensitive. Hold each tolerance in a named kGapTol constant and build the failure message from it, so the value can no longer desync between the assertion and its message.