Files
hakmem/benchmarks/results/benchmark_summary.md
Moe Charm (CI) 52386401b3 Debug Counters Implementation - Clean History
Major Features:
- Debug counter infrastructure for Refill Stage tracking
- Free Pipeline counters (ss_local, ss_remote, tls_sll)
- Diagnostic counters for early return analysis
- Unified larson.sh benchmark runner with profiles
- Phase 6-3 regression analysis documentation

Bug Fixes:
- Fix SuperSlab disabled by default (HAKMEM_TINY_USE_SUPERSLAB)
- Fix profile variable naming consistency
- Add .gitignore patterns for large files

Performance:
- Phase 6-3: 4.79 M ops/s (has OOM risk)
- With SuperSlab: 3.13 M ops/s (+19% improvement)

This is a clean repository without large log files.

🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
2025-11-05 12:31:14 +09:00

1.5 KiB

HAKMEM 包括的ベンチマーク結果サマリー

実測データ (Baseline)

1. Tiny サイズ (ST)

Test HAKMEM System vs System
tiny_hot 32B 63.85 M/s 175.27 M/s -63.6%
tiny_hot 64B 54.56 M/s 169.49 M/s -67.8%

2. Mixed サイズ

Test HAKMEM System vs System
random_mixed (8B-1KB) 16.39 M/s 22.41 M/s -26.9%

3. Mid-Large サイズ

Test HAKMEM System vs System
mid_large (8-32KB ST) 28.30 M/s 13.56 M/s +108.7% STRONG
mid_large_mt (8-32KB 2T) 7.27 M/s 11.67 M/s -37.7%

4. HAKX Specialty (8-32KB専用最適化版)

Test HAKX mimalloc System vs mimalloc vs System
8-32KB optimized 167.75 M/s 136.17 M/s 61.81 M/s +23.2% +171%

Comprehensive Benchmark (進行中)

ワークロードパターン:

  1. Sequential LIFO (LIFO順 free)
  2. Sequential FIFO (FIFO順 free)
  3. Random Free (ランダム順 free)
  4. Interleaved Alloc/Free (交互パターン)
  5. Long-lived vs Short-lived (長期/短期混在)
  6. Mixed Sizes (8B, 16B, 32B, 64B)

サイズクラス: 16B, 32B, 64B, 128B

HAKMEM 結果 (21 tests)

平均スループット: 52.59 M ops/sec

詳細分布:

  • 最高: 66.38 M/s
  • 最低: 35.49 M/s
  • 中央値: ~51 M/s

待機中: System, mimalloc の結果