Repository navigation
Conversation
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…irror column sets The last phase of an r2c row's half-length grid runs over pairs of column sets (columns c hb + 1 .. c hb + hb and their mirrors L - l), both gathered into one block, transformed, paired in registers and written; column 0 in a stage of its own. No separate pairing pass over the half spectrum. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…X30b) Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…ing pass Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…st with sets of 2) Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
… sets wider than half) Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
…gs::r2c_pair_last for tests Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
… pow2_chain_bytes, as c2c rows do (X32) Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Experiment X33 (PLAN lead (b), long real rows):
dev/experiments/X33-long-real.md.plan_gridno longer excludes real rows). EPYC long real rows 0.58–0.76 of main (FFTN ÷ FFTW 1.14–1.41 → 0.82–0.93), Mac 0.69–0.97; tracking tiers 0.972 (EPYC) and 0.970 (Mac), nothing slower with evidence. The Xeon has no power-of-two chains, so nothing changes there.avx512and NEON, not onavx2): an r2c grid pairs its bins in its last phase. The phase runs over pairs of mirror column sets (columns c·hb + 1 … c·hb + hb and L − l, with hb = block/2 and each set at least a cache line wide), gathered into one block, transformed, paired in registers and written back. Column 0 gets its own stage, and there is no separate pairing pass. The kernels find the iteration from the loop's twiddle cursor, asc2r_pair_gatherdoes. Xeon r2c 2¹⁸–2²⁰ 0.95–0.96; EPYC no gain;build_settings::r2c_pair_lastforces it in tests.🤖 Generated with Claude Code