Commits on Mar 18, 2026

1. fix: FatTeddy AVX2 + prefilter acceleration + AC v0.2.1 (#143) …

   * fix: FatTeddy AVX2 ANDL→ORL — correct lane combining for 33-64 patterns
   
   teddyFatAVX2_2 used ANDL to combine low/high 128-bit lane candidate masks.
   This required a match in BOTH lanes, but patterns assigned to a single lane
   (e.g., GET variants in buckets 0-7 = low lane only) were zeroed out.
   Fix: ORL — candidate exists if EITHER lane has a match.
   
   Also adds DECQ SI at handle_tail to cover prev0 carry-over position.
   
   * test: FatTeddy AVX2 regression tests — lane combining + case-fold
   
   - TestFatTeddyAVX2_SingleLaneBuckets: verifies GET/POST/PUT found
     with patterns distributed across specific lanes
   - TestFatTeddyCaseFoldRegression: validates match count vs stdlib
     for (?i)get|post|put via meta-engine (750/750 on 1000 log lines)
   
   * refactor: remove AC replacement and small-haystack fallback
   
   FatTeddy AVX2 is now correct (ORL fix). Remove:
   - buildACPrefilter: replaced ALL FatTeddy with AC (workaround)
   - ac_prefilter.go: AC wrapper type (unused)
   - fatTeddyFallback path in findIndicesTeddyAt (FatTeddy handles all sizes)
   
   LangArena methods (?i)get|post|put: AC 41ms → FatTeddy 11ms
   
   * perf: prefilter candidate loop in isMatchDFA with anchored verification
   
   isMatchDFA now loops through prefilter candidates and verifies each with
   anchored DFA instead of single unanchored PikeVM scan from first candidate.
   
   Issue #137 match case: 176us → 27us (6.5x faster, 3.7x faster than stdlib)
   
   * perf: prefilter candidate loop in findIndicesDFA for large NFA (>100 states)
   
   For patterns with >100 NFA states (e.g., (?i) case-fold), bidirectional DFA
   cache-thrashes. Prefilter candidate loop with anchored DFA verification
   avoids building full DFA state table.
   
   Guard: nfaStateCount > 100 — small NFAs (34-57 states) use bidirectional
   DFA directly (no cache thrashing, lower overhead than candidate loop).
   
   Also adds nfaStateCount field to Engine for the guard check.
   
   * perf: Rust-style cascading prefix trim for >64 literals
   
   When >64 literals extracted, try cascading trim: keep 4 bytes → dedup,
   if still >64 → keep 3 bytes → dedup, if still >64 → keep 2 bytes → dedup.
   Falls back to original if trimming can't reduce below 64.
   
   Matches Rust's optimize_for_prefix_by_preference ATTEMPTS table.
   auth_attempts (?i)/login|/signin: 128 literals → 18 four-byte → Teddy.
   34ms → 10ms.
   
   * perf: VPTEST in FatTeddy hot loop — replace 8-instruction candidate detection
   
   VPTEST Y6,Y6 (1 instruction, sets ZF if all zero) replaces:
   VPXOR+VPCMPEQB+VPMOVMSKB+NOTL+MOVL+ANDL+SHRL+ORL (8 instructions).
   
   ORL lane combining moved to found_candidate cold path (only on actual
   candidates, not every 16-byte chunk).
   
   FatTeddy 35 patterns on 6MB: 55ms → 41ms (24% faster).
   
   * wip: FatTeddy batch ASM + FindAllPositions (intermediate save)
   
   Batch ASM: fatTeddyAVX2_2_batch scans entire haystack in one call,
   writes all (pos, bucketMask) candidates to pre-allocated buffer.
   Eliminates Go→ASM round trip overhead (11.6 GB/s SIMD vs 283 MB/s with round trips).
   
   FindAllPositions: batch-based FindAll using single ASM call.
   Find: uses original per-candidate path (batch too slow for first-match).
   
   * deps: upgrade ahocorasick v0.1.0 → v0.2.1 (DFA + SIMD prefilter, 11-22x)
   
   Flat DFA with premultiplied state IDs, match flag in high bit,
   SIMD skip-ahead prefilter via bytes.IndexByte.
   
   Throughput: Find 3.4 GB/s, IsMatch 5.9-7.0 GB/s (was 260-545 MB/s).
   
   * docs: release v0.12.13 — changelog, roadmap, lint fix
   
   * fix: add hasAVX2 and batch stubs for non-amd64 platforms (macOS ARM64)

   

   kolkov authored Mar 18, 2026
   Configuration menu

   • View commit details
   • Copy full SHA for 7a29fab
   • Browse repository at this point

   Copy the full SHA

   7a29fab View commit details

   Browse the repository at this point in the history