Batch long Ice Lake UTF-8 validation blocks - #2814
Open
perfloop-agent wants to merge 2 commits into
Open
Conversation
Member
|
Will need to be assessed manually. CI doesn't necessarily support avx-512. |
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
Batch long UTF-8 validation requests in the Ice Lake implementation. For inputs of at least 64 KiB, the validator loads groups of 64-byte blocks and checks each block against its physical predecessor instead of carrying validation results through every vector operation. It retains only a three-byte suffix and an ASCII summary between groups, preserving checks at block joins and in the padded final block. Shorter requests continue to use the existing stateful path to avoid batch setup overhead.
Performance
On workload
Ice Lake 64 KiB ASCII UTF-8 validation, median metriccpu_time_nschanged from 1376 to 1084; paired median delta -288.5 (-21% of baseline; at least 19/20 confidence interval -302.1 to -275 from 10 pairs).On workload
Ice Lake 1 MiB ASCII UTF-8 validation, median metriccpu_time_nschanged from 22664 to 17813; paired median delta -5016 (-22.1% of baseline; at least 19/20 confidence interval -5616 to -4324 from 10 pairs).On workload
Ice Lake 1 MiB multilingual UTF-8 validation, median metriccpu_time_nschanged from 68800 to 56998; paired median delta -11844 (-17.2% of baseline; at least 19/20 confidence interval -12787 to -10885 from 10 pairs).On workload
Ice Lake 64 KiB malformed UTF-8 at a 64-byte boundary, median metriccpu_time_nschanged from 1173 to 981.3; paired median delta -190.4 (-16.2% of baseline; at least 19/20 confidence interval -202.7 to -174.9 from 10 pairs).On workload
Ice Lake 1 MiB malformed UTF-8 at a 64-byte boundary, median metriccpu_time_nschanged from 19502 to 17171; paired median delta -2303 (-11.8% of baseline; at least 19/20 confidence interval -3343 to -1654 from 10 pairs).Testing
Added differential coverage against the fallback implementation for long inputs, block boundaries, padded tails, dense multibyte data, and invalid mutations. Ran the CMake CTest suite with the Ice Lake implementation forced.
The declared correctness check passed.
Authored and verified by Perfloop: every claim above was co-measured on both trees and independently re-verified before submission — the full record is public: case_pg5w1h5y4k. Replies from this account are human-approved, and a human operator is accountable for this contribution.