tomoaki/hakmem
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
552 Commits 2 Branches 0 Tags
b2724e6f5d848dc96dba4b4b75b332e891e95ca1
Commit Graph

440 Commits

Author SHA1 Message Date
Moe Charm (CI)
b2724e6f5d Phase ALLOC-TINY-FAST-DUALHOT-1: WIP (regression), FREE DUALHOT confirmed +13% 
**ALLOC-TINY-FAST-DUALHOT-1** (this phase):
- Implementation: malloc_tiny_fast() C0-C3 early-exit with policy snapshot skip
- ENV: HAKMEM_TINY_ALLOC_DUALHOT=0/1 (default OFF)
- A/B Result: -1.17% median regression (Mixed, 10-run)
- Root Cause: Branch prediction penalty on C4-C7 outweighs policy skip benefit
- Decision: Freeze as research box (default OFF)
- Difference from FREE: ALLOC requires structural changes (per-class paths)

**FREE-TINY-FAST-DUALHOT-1** (verified):
- A/B Confirmation: +13.00% improvement (42.08M → 47.81M ops/s, Mixed, 10-run)
- Success Criteria: +2% target ACHIEVED
- Health Check: PASS (verify_health_profiles.sh, ENV OFF/ON)
- Safety: HAKMEM_TINY_LARSON_FIX guard in place
- Decision: Promotion to MIXED_TINYV3_C7_SAFE profile candidate

**Next Steps**:
- Profile adoption of FREE DUALHOT for MIXED workload
- No further deep-dive on ALLOC optimization (deferred to future phases)

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <noreply@anthropic.com>
 2025-12-13 05:10:45 +09:00
Moe Charm (CI)
0a7400d7d3 Phase ALLOC-TINY-FAST-DUALHOT-1: C0-C3 alloc direct path (WIP, -2% regression) 
Add C0-C3 early-exit optimization to malloc_tiny_fast() similar to
FREE-TINY-FAST-DUALHOT-1. Skip policy snapshot for C0-C3 classes.

A/B Result (10-run, Mixed TINYV3_C7_SAFE):
- Baseline: 47.27M ops/s (median)
- Optimized: 46.10M ops/s (median)
- Result: -2.00% (regression, needs investigation)

ENV: HAKMEM_TINY_ALLOC_DUALHOT=0/1 (default OFF)

Implementation:
- core/front/malloc_tiny_fast.h: alloc_dualhot_enabled() + early-exit
- Design: docs/analysis/ALLOC_TINY_FAST_DUALHOT_1_DESIGN.md

Status: Research box (default OFF), needs root cause analysis

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <noreply@anthropic.com>
 2025-12-13 04:28:52 +09:00
Moe Charm (CI)
2b567ac070 Phase FREE-TINY-FAST-DUALHOT-1: Optimize C0-C3 direct free path 
Treat C0-C3 classes (48% of calls) as "second hot path" instead of
cold path. Skip expensive policy snapshot and route determination,
direct to tiny_legacy_fallback_free_base().

Measurements from FREE-TINY-FAST-HOTCOLD-OPT-1 revealed C0-C3 is NOT
rare (48.43% of all frees). Previous attempt to optimize via hot/cold
split failed (-13% regression) because noinline + function call on 48%
of workload hurt more than it helped.

This phase applies correct optimization: direct inline path for
frequent C0-C3 without policy snapshot overhead.

Implementation:
- Insert C0-C3 early-exit after C7 ULTRA check
- Skip tiny_front_v3_snapshot_get() for C0-C3 (saves 5-10 cycles)
- Skip route determination logic
- Safety: HAKMEM_TINY_LARSON_FIX=1 disables optimization

Benchmark Results (100M ops, 400 threads, MIXED_TINYV3_C7_SAFE):
- Baseline (optimization OFF): 44.50M ops/s (median)
- Optimized (DUALHOT ON):      48.74M ops/s (median)
- Improvement: +9.51% (+4.23M ops/s)

Perf Stats (optimized):
- Branch misses: 112.8M
- Cycles: 8.89B
- Instructions: 21.95B (2.47 IPC)
- Cache misses: 656K

Status: GO (significant improvement, no regression)

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
 2025-12-13 03:46:36 +09:00
Moe Charm (CI)
c503b212a3 Phase FREE-TINY-FAST-HOTCOLD-OPT-1: Hot/Cold split for free_tiny_fast [RESEARCH BOX - FREEZE] 
Split free_tiny_fast() into hot and cold paths to reduce I-cache pressure:
- free_tiny_fast_hot(): always_inline, fast-path validation + ULTRA/MID/V7
- free_tiny_fast_cold(): noinline,cold, cross-thread + TinyHeap + legacy

ENV: HAKMEM_FREE_TINY_FAST_HOTCOLD=0/1 (default 0)
Stats: HAKMEM_FREE_TINY_FAST_HOTCOLD_STATS=0/1 (TLS only, exit dump)

## Benchmark Results (random mixed, 100M ops)

HOTCOLD=0 (legacy): 49.35M, 50.18M, 50.25M ops/s (median: 50.18M)
HOTCOLD=1 (split):  43.54M, 43.59M, 43.62M ops/s (median: 43.59M)

**Regression: -13.1%** (NO-GO)

## Stats Analysis (10M ops, HOTCOLD_STATS=1)

Hot path:  50.11% (C7 ULTRA early-exit)
Cold path: 48.43% (legacy fallback)

## Root Cause

Design assumption FAILED: "Cold path is rare"
Reality: Cold path is 48% (almost as common as hot path)

The split introduces:
1. Extra dispatch overhead in hot path
2. Function call overhead to cold for ~48% of frees
3. "Cold" is NOT rare - it's the legacy fallback for non-ULTRA classes

## Conclusion

**FREEZE as research box (default OFF)**

Box Theory value:
- Validated hot/cold distribution via TLS stats
- Confirmed that legacy fallback is NOT rare (48%)
- Demonstrated that naive hot/cold split hurts when "cold" is common

Alternative approaches for future work:
1. Inline the legacy fallback in hot path (no split)
2. Route-specific specialization (C7 vs non-C7 separate paths)
3. Policy-based early routing (before header validation)

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
 2025-12-13 03:16:54 +09:00
Moe Charm (CI)
4e7870469c POOL-MID-DN-BATCH: Add hash-based TLS page map (O(1) lookup) 
Replace linear search (avg 16 iterations, -7.6% regression) with
open addressing hash table:
- Size: 64 slots (power-of-two)
- Collision: Linear probing, max 8 probes
- On probe limit: drain and retry (safe fallback)
- Hash function: Golden ratio with page-aligned shift

New ENV: HAKMEM_POOL_MID_INUSE_MAP_KIND=hash|linear (default: linear)

Implementation:
- Added hak_pool_mid_inuse_map_hash_enabled() ENV gate
- Extended MidInuseTlsPageMap with hash_pages[64], hash_counts[64], hash_used
- Added mid_inuse_hash_page() golden ratio hash function
- Added mid_inuse_dec_deferred_hash() O(1) insert with probing
- Updated mid_inuse_deferred_drain() to support hash mode
- Added decs_drained stats counter for batching metrics

Benchmark Results (10 runs each, bench_mid_large_mt_hakmem):
  Baseline (DEFERRED=0): median=9,250,340 ops/s
  Linear mode:           median=8,159,240 ops/s (-11.80%)
  Hash mode:             median=8,262,982 ops/s (-10.67%)

Hash vs Linear: +1.27% improvement (eliminates linear search overhead)

Note: Both deferred modes still show regression vs baseline due to
other factors (TLS access overhead, drain cost). Hash mode successfully
eliminates the linear search penalty as designed.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
 2025-12-13 00:28:03 +09:00
Moe Charm (CI)
6c849fd020 POOL-MID-DN-BATCH: Add last-match cache to reduce linear search overhead 
Root cause: Linear search in 32-entry TLS map averaged 16 iterations,
causing instruction overhead that exceeded mid_desc_lookup savings.

Fix implemented:
- Added last_idx field to MidInuseTlsPageMap for temporal locality
- Check last_idx before linear search (O(1) fast path)
- Update last_idx on hits and new entries
- Reset last_idx on drain

Changes:
1. pool_mid_inuse_tls_pagemap_box.h:
   - Added uint32_t last_idx field to struct

2. pool_mid_inuse_deferred_box.h:
   - Check last_idx before linear search (lines 90-94)
   - Update last_idx on linear search hit (line 101)
   - Set last_idx on new entry insert (line 117)
   - Reset last_idx on drain (line 166)

Benchmark results (bench_mid_large_mt_hakmem):
- Baseline (DEFERRED=0): median 9.08M ops/s, variance 300B
- Deferred with cache (DEFERRED=1): median 8.38M ops/s, variance 207B
- Performance: -7.6% regression (vs expected +2-4% gain)
- Stability: -31% variance (improvement as expected)

Analysis:
The last-match cache reduces variance but does not eliminate the
regression for this benchmark's random access pattern (2048 slots,
many pages). The temporal locality assumption (60-80% hit rate) is
not met by bench_mid_large_mt's allocation pattern.

Further optimization needed:
- Consider hash-based lookup for better than O(n) search
- OR reduce map size to decrease search iterations
- OR add drain triggers at better boundaries

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
 2025-12-13 00:04:41 +09:00
Moe Charm (CI)
16b415f5a2 Phase POOL-MID-DN-BATCH Step 5: Integrate deferred API into pool_free_v1 2025-12-12 23:00:06 +09:00
Moe Charm (CI)
cba444b943 Phase POOL-MID-DN-BATCH Step 4: Deferred API implementation with thread cleanup 2025-12-12 23:00:00 +09:00
Moe Charm (CI)
d45729f063 Phase POOL-MID-DN-BATCH Step 3: Statistics counters for deferred inuse_dec 2025-12-12 22:59:56 +09:00
Moe Charm (CI)
b381515b16 Phase POOL-MID-DN-BATCH Step 2: TLS page map for batched inuse_dec 2025-12-12 22:59:50 +09:00
Moe Charm (CI)
f5f03ef68c Phase POOL-MID-DN-BATCH Step 1: ENV gate for deferred inuse_dec 2025-12-12 22:59:45 +09:00
Moe Charm (CI)
506d8f2e5e Phase: Pool API Modularization - Step 8 (FINAL): Extract pool_alloc_v1_box.h 
Extract 288 lines: hak_pool_try_alloc_v1_impl() - LARGEST SIZE
- New box: core/box/pool_alloc_v1_box.h (v1 alloc baseline, no hotbox_v2)
- Updated: pool_api.inc.h (add include, remove extracted function)
- Build: OK, bench_mid_large_mt_hakmem: 8.01M ops/s (baseline ~8M, within ±2%)
- Risk: MEDIUM (simpler than v2 but large function, validated)
- Result: pool_api.inc.h reduced from 909 lines to ~40 lines (95% reduction)

ALL 5 STEPS COMPLETE (Steps 4-8):
- Step 4: pool_block_to_user_box.h (30 lines) - helpers
- Step 5: pool_free_v2_box.h (121 lines) - v2 free with hotbox
- Step 6: pool_alloc_v1_flat_box.h (103 lines) - v1 flatten TLS
- Step 7: pool_alloc_v2_box.h (277 lines) - v2 alloc with hotbox
- Step 8: pool_alloc_v1_box.h (288 lines) - v1 alloc baseline

Total extracted: 819 lines
Final pool_api.inc.h size: ~40 lines (public wrappers only)
Performance: MAINTAINED (8M ops/s baseline)

Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
 2025-12-12 22:28:13 +09:00
Moe Charm (CI)
76a5bb568a Phase: Pool API Modularization - Step 7: Extract pool_alloc_v2_box.h 
Extract 277 lines: hak_pool_try_alloc_v2_impl() - LARGEST COMPLEXITY
- New box: core/box/pool_alloc_v2_box.h (v2 alloc with hotbox, MF2, TC drain, TLS)
- Updated: pool_api.inc.h (add include, remove extracted function)
- Build: OK, bench_mid_large_mt_hakmem: 8.86M ops/s (baseline ~8M, within ±2%)
- Risk: MEDIUM (complex function with 30+ dependencies, validated)
- Note: Avoided forward declarations for types/macros already in compilation unit

Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
 2025-12-12 22:24:21 +09:00
Moe Charm (CI)
5f069e08bf Phase: Pool API Modularization - Step 6: Extract pool_alloc_v1_flat_box.h 
Extract 103 lines: hak_pool_try_alloc_v1_flat() + hak_pool_free_v1_flat()
- New box: core/box/pool_alloc_v1_flat_box.h (v1 flatten TLS-only fast path)
- Updated: pool_api.inc.h (add include, remove extracted functions)
- Build: OK, bench_mid_large_mt_hakmem: 9.17M ops/s (baseline ~8M, within ±2%)
- Risk: MINIMAL (TLS-only path, well-isolated)
- Note: Added forward declarations for v1_impl functions (defined later)

Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
 2025-12-12 22:20:19 +09:00
Moe Charm (CI)
0ad9c57aca Phase: Pool API Modularization - Step 5: Extract pool_free_v2_box.h 
Extract 121 lines: hak_pool_free_v2_impl() + hak_pool_mid_lookup_v2_impl() + hak_pool_free_fast_v2_impl()
- New box: core/box/pool_free_v2_box.h (v2 free with hotbox support)
- Updated: pool_api.inc.h (add include, remove extracted functions)
- Build: OK, bench_mid_large_mt_hakmem: 8.58M ops/s (baseline ~8M, within ±2%)
- Risk: LOW-MEDIUM (hotbox_v2 integration, well-isolated)

Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
 2025-12-12 22:17:53 +09:00
Moe Charm (CI)
0da8a63fa5 Phase: Pool API Modularization - Step 4: Extract pool_block_to_user_box.h 
Extract 30 lines: hak_pool_block_to_user() + hak_pool_block_to_user_legacy()
- New box: core/box/pool_block_to_user_box.h (helpers for block→user conversion)
- Updated: pool_api.inc.h (add include, remove extracted functions)
- Build: OK, bench_mid_large_mt_hakmem: 9.17M ops/s (baseline ~8M)
- Risk: MINIMAL (simple extraction, no dependencies)

Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
 2025-12-12 22:15:21 +09:00
Moe Charm (CI)
a92f3e52c3 Phase: Pool API Modularization - Step 3: Extract pool_free_v1_box.h 
Extracted pool v1 free implementation into separate box module:
- hak_pool_free_v1_fast_impl(): L1-FastBox (TLS-only path, no mid_desc_lookup)
- hak_pool_free_v1_slow_impl(): L1-SlowBox (full impl with lookup)
- hak_pool_free_v1_impl(): L0-SplitBox (fast predicate router)

Benefits:
- Reduced pool_api.inc.h from ~950 to ~840 lines
- Clear separation of concern (fast vs slow paths)
- Enables future phase extensions (e.g., POOL-MID-DN-BATCH)
- Maintains zero-cost abstraction (all inline)

Testing:
- Build: ✓ (no errors)
- Benchmark: ✓ (7.99M ops/s, consistent with baseline)

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
 2025-12-12 21:46:26 +09:00
Moe Charm (CI)
b01c99f209 Phase: Pool API Modularization - Steps 1-2 
Extract configuration, statistics, and caching boxes from pool_api.inc.h

Step 1: pool_config_box.h (60 lines)
  - All ENV gate predicates (hak_pool_v2_enabled, hak_pool_v1_flatten_enabled, etc)
  - Lazy static int cache pattern (matches tiny_heap_env_box.h style)
  - Zero dependencies (lowest-level box)

Step 2a: pool_stats_box.h (90 lines)
  - PoolV1FlattenStats structure with multi-phase support
  - pool_v1_flat_stats_dump() with phase-aware output
  - Destructor hook for automatic dumping on exit
  - Multi-phase design: supports future phases without refactoring

Step 2b: pool_mid_desc_cache_box.h (60 lines)
  - MidDescCache structure (TLS-local single-entry LRU)
  - mid_desc_lookup_cached() with fast TLS hit path
  - Minimal external dependency: mid_desc_lookup from pool_mid_desc.inc.h

Result: pool_api.inc.h reduced from 1050+ lines to ~950 lines
  Still contains: alloc/free implementations, helpers (next steps)

Build: ✅ Clean (no warnings)
Test: ✅ Benchmark passes (8.5M ops/s)

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
 2025-12-12 21:39:18 +09:00
Moe Charm (CI)
c86a59159b Phase POOL-FREE-V1-OPT Step 2: Fast/Slow split for v1 free 
Implement L0-SplitBox + L1-FastBox/SlowBox architecture for pool v1 free:

L0-SplitBox (hak_pool_free_v1_impl):
  - Fast predicate: header-based same-thread detection
  - Requires g_hdr_light_enabled == 0, tls_free_enabled
  - Routes to fast or slow box based on predicate

L1-FastBox (hak_pool_free_v1_fast_impl):
  - Same-thread TLS free path only (ring → lo_head → spill)
  - Skips mid_desc_lookup for validation (uses header)
  - Still calls mid_page_inuse_dec_and_maybe_dn at end

L1-SlowBox (hak_pool_free_v1_slow_impl):
  - Full v1 impl with mid_desc_lookup for validation
  - Handles cross-thread, TC lookup, etc.

ENV gate: HAKMEM_POOL_V1_FREE_FASTSPLIT (default OFF)

Stats tracking:
  - fastsplit_fast_hit: Fast path taken (>99% typically)
  - fastsplit_slow_hit: Slow path taken (predicate failed)

Benchmark result (FLATTEN OFF, Mixed profile):
  - Baseline: ~8.3M ops/s (high variance)
  - FASTSPLIT ON: ~8.1M ops/s (high variance)
  - Performance neutral (savings limited by inuse_dec still calling mid_desc_lookup)

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
 2025-12-12 19:52:36 +09:00
Moe Charm (CI)
dbdd2e0e0e Phase POOL-FREE-V1-OPT Step 1: Add v2 reject stats tracking 
Add reject reason counters for v2 free path to understand fallback patterns:
- v2_reject_total: Total v2 free rejects
- v2_reject_ptr_null: ptr == NULL
- v2_reject_not_init: pool not initialized
- v2_reject_desc_null: mid_desc_lookup returned NULL
- v2_reject_mf2_null: MF2 path but mf2_addr_to_page returned NULL

ENV gate: HAKMEM_POOL_FREE_V1_REJECT_STATS (default OFF)

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
 2025-12-12 19:43:03 +09:00
Moe Charm (CI)
fe70e3baf5 Phase MID-V35-HOTPATH-OPT-1 complete: +7.3% on C6-heavy 
Step 0: Geometry SSOT
  - New: core/box/smallobject_mid_v35_geom_box.h (L1/L2 consistency)
  - Fix: C6 slots/page 102→128 in L2 (smallobject_cold_iface_mid_v3.c)
  - Applied: smallobject_mid_v35.c, smallobject_segment_mid_v3.c

Step 1-3: ENV gates for hotpath optimizations
  - New: core/box/mid_v35_hotpath_env_box.h
    * HAKMEM_MID_V35_HEADER_PREFILL (default 0)
    * HAKMEM_MID_V35_HOT_COUNTS (default 1)
    * HAKMEM_MID_V35_C6_FASTPATH (default 0)
  - Implementation: smallobject_mid_v35.c
    * Header prefill at refill boundary (Step 1)
    * Gated alloc_count++ in hot path (Step 2)
    * C6 specialized fast path with constant slot_size (Step 3)

A/B Results:
  C6-heavy (257–768B): 8.75M→9.39M ops/s (+7.3%, 5-run mean) ✅
  Mixed (16–1024B): 9.98M→9.96M ops/s (-0.2%, within noise) ✓

Decision: FROZEN - defaults OFF, C6-heavy推奨ON, Mixed現状維持
Documentation: ENV_PROFILE_PRESETS.md updated

🤖 Generated with Claude Code

Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
 2025-12-12 19:19:25 +09:00
Moe Charm (CI)
e95e61f0ff Phase POLICY-FAST-PATH-V2 complete + MID-V35-HOTPATH-OPT-1 design 
## Phase POLICY-FAST-PATH-V2 (FROZEN)
- Implementation complete: free_policy_fast_v2_box.h + malloc_tiny_fast.h integration
- A/B Results:
  - Mixed (ws=400): -1.6% regression ❌ (branch cost > skip benefit)
  - C6-heavy (ws=200): +5.4% improvement ✅
- Decision: Default OFF, FROZEN (ws<300 / C6-heavy research only)
- Learning: Large WS causes branch misprediction to dominate

## Phase 3-GRADUATE + ENV probe fix
- 64-probe retry for getenv() stability during bench_profile putenv()
- C6 ULTRA intrusive freelist: FROZEN (research box)

## Phase MID-V35-HOTPATH-OPT-1-DESIGN
- Design doc for next optimization target
- Target: MID v3.5 alloc/free hot path (C5-C6)
- Boxes: Stats Gate, TLS Layout, Boundary Check elimination
- Expected: +3-9% on Mixed mainline

Files:
- core/box/free_policy_fast_v2_box.h (new)
- core/box/free_path_stats_box.h/c (policy_fast_v2_skip counter)
- core/front/malloc_tiny_fast.h (fast-path integration)
- docs/analysis/MID_V35_HOTPATH_OPT_1_DESIGN.md (new)
- docs/analysis/PHASE_3_GRADUATE_*.md (new)
- CURRENT_TASK.md (phase status update)

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <noreply@anthropic.com>
 2025-12-12 18:40:08 +09:00
Moe Charm (CI)
0c8583f91e Phase TLS-UNIFY-3+: Refactoring - Unified ENV gates for C6 ULTRA 
Consolidate C6 ULTRA ENV gate functions:
- tiny_c6_ultra_intrusive_env_box.h now contains both:
  - tiny_c6_ultra_free_enabled() - C6 ULTRA routing (policy gate)
  - tiny_c6_ultra_intrusive_enabled() - intrusive LIFO mode (TLS optimization)
- Simplified ENV gate management with clear separation of concerns

Removes code duplication by centralizing environment checks in single header.
Performance verified: ENV_OFF=56.4 Mop/s, ENV_ON=57.6 Mop/s (parity maintained)

Note: Avoided macro-based segment learning consolidation (C4/C5/C6) as it
would hinder compiler optimizations. Current inline approach is optimal.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
 2025-12-12 16:31:14 +09:00
Moe Charm (CI)
1a8652a91a Phase TLS-UNIFY-3: C6 intrusive freelist implementation (完成) 
Implement C6 ULTRA intrusive LIFO freelist with ENV gating:
- Single-linked LIFO using next pointer at USER+1 offset
- tiny_next_store/tiny_next_load for pointer access (single source of truth)
- Segment learning via ss_fast_lookup (per-class seg_base/seg_end)
- ENV gate: HAKMEM_TINY_C6_ULTRA_INTRUSIVE_FL (default OFF)
- Counters: c6_ifl_push/pop/fallback in FREE_PATH_STATS

Files:
- core/box/tiny_ultra_tls_box.h: Added c6_head field for intrusive LIFO
- core/box/tiny_ultra_tls_box.c: Pop/push with intrusive branching (case 6)
- core/box/tiny_c6_ultra_intrusive_env_box.h: ENV gate (new)
- core/box/tiny_c6_intrusive_freelist_box.h: L1 pure LIFO (new)
- core/tiny_debug_ring.h: C6_IFL events
- core/box/free_path_stats_box.h/c: c6_ifl_* counters

A/B Test Results (1M iterations, ws=200, 257-512B):
- ENV_OFF (array): 56.6 Mop/s avg
- ENV_ON (intrusive): 57.6 Mop/s avg (+1.8%, within noise)
- Counters verified: c6_ifl_push=265890, c6_ifl_pop=265815, fallback=0

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
 2025-12-12 16:26:42 +09:00
Moe Charm (CI)
d5ffb3eeb2 Fix MID v3.5 activation bugs: policy loop + malloc recursion 
Two critical bugs fixed:

1. Policy snapshot infinite loop (smallobject_policy_v7.c):
   - Condition `g_policy_v7_version == 0` caused reinit on every call
   - Fixed via CAS to set global version to 1 after first init

2. Malloc recursion (smallobject_segment_mid_v3.c):
   - Internal malloc() routed back through hakmem → MID v3.5 → segment
     creation → malloc → infinite recursion / stack overflow
   - Fixed by using mmap() directly for internal allocations:
     - Segment struct, pages array, page metadata block

Performance results (bench_random_mixed 257-512B):
- Baseline (LEGACY): 34.0M ops/s
- MID_V35 ON (C6):   35.8M ops/s
- Improvement:       +5.1% ✓

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
 2025-12-12 07:12:24 +09:00
Moe Charm (CI)
212739607a Phase v11a-3: MID v3.5 Activation (Build Complete) 
Integrated MID v3.5 into active code path, making it available for C5/C6/C7 routing.

Key Changes:
- Policy Box: Added SMALL_ROUTE_MID_V35 with ENV gates (HAKMEM_MID_V35_ENABLED, HAKMEM_MID_V35_CLASSES)
- HotBox: Implemented small_mid_v35_alloc/free with TLS-cached page allocation
- Front Gate: Wired MID_V35 routing into malloc_tiny_fast.h (priority: ULTRA > MID_V35 > V7)
- Build: Added core/smallobject_mid_v35.o to all object lists

Architecture:
- Slot sizes: C5=384B, C6=512B, C7=1024B
- Page size: 64KB (170/128/64 slots)
- Integration: ColdIface v2 (refill/retire), Stats v2 (observation), Learner v2 (dormant)

Status: Build successful, ready for A/B benchmarking
Next: Performance validation (C6-heavy, C5+C6-only, Mixed benchmarks)

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
 2025-12-12 06:52:14 +09:00
Moe Charm (CI)
0dba67ba9d Phase v11a-2: Core MID v3.5 implementation - segment, cold iface, stats, learner 
Implement 5-layer infrastructure for multi-class MID v3.5 (C5-C7, 257-1KiB):

1. SegmentBox_mid_v3 (L2 Physical)
   - core/smallobject_segment_mid_v3.c (9.5 KB)
   - 2MiB segments, 64KiB pages (32 per segment)
   - Per-class free page stacks (LIFO)
   - RegionIdBox registration
   - Slots: C5→170, C6→102, C7→64

2. ColdIface_mid_v3 (L2→L1)
   - core/box/smallobject_cold_iface_mid_v3_box.h (NEW)
   - core/smallobject_cold_iface_mid_v3.c (3.5 KB)
   - refill: get page from free stack or new segment
   - retire: calculate free_hit_ratio, publish stats, return to stack
   - Clean separation: TLS cache for hot path, ColdIface for cold path

3. StatsBox_mid_v3 (L2→L3)
   - core/smallobject_stats_mid_v3.c (7.2 KB)
   - Circular buffer history (1000 events)
   - Per-page metrics: class_idx, allocs, frees, free_hit_ratio_bps
   - Periodic aggregation (every 100 retires)
   - Learner notification callback

4. Learner v2 (L3)
   - core/smallobject_learner_v2.c (11 KB)
   - Multi-class aggregation: allocs[8], retire_count[8], avg_free_hit_bps[8]
   - Exponential smoothing (90% history + 10% new)
   - Per-class efficiency tracking
   - Stats snapshot API
   - Route decision disabled for v11a-2 (v11b feature)

5. Build Integration
   - Modified Makefile: added 4 new .o files (segment, cold_iface, stats, learner)
   - Updated box header prototypes
   - Clean compilation, all dependencies resolved

Architecture Decision Implementation:
- v7 remains frozen (C5/C6 research preset)
- MID v3.5 becomes unified 257-1KiB main path
- Multi-class isolation: per-class free stacks
- Dormant infrastructure: linked but not active (zero overhead)

Performance:
- Build: clean compilation
- Sanity benchmark: 27.3M ops/s (no regression vs v10)
- Memory: ~30MB RSS (baseline maintained)

Design Compliance:
✅ Layer separation: L2 (segment) → L2 (cold iface) → L3 (stats) → L3 (learner)
✅ Hot path clean: alloc/free never touch stats/learner
✅ Backward compatible: existing MID v3 routes unchanged
✅ Transparent: v11a-2 is dormant (no behavior change)

Next Phase (v11a-3):
- Activate C5/C6/C7 routing through MID v3.5
- Connect TLS cache to segment refill
- Verify performance under load
- Then Phase v11a-4: dynamic C5 ratio routing

🤖 Generated with Claude Code

Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
 2025-12-12 06:37:06 +09:00
Moe Charm (CI)
babd884b96 Phase v11a-1: Infrastructure - Multi-class segment and learner v2 box definitions 
Create core box definitions for MID v3.5 consolidation (Phase v11a):

1. smallobject_segment_mid_v3_box.h
   - Multi-class unified segment (2MiB, C5-C7)
   - Per-class free page stacks
   - SmallHeapCtx_MID_v3 for TLS caching
   - Refill/retire/validation APIs

2. smallobject_stats_mid_v3_box.h
   - SmallPageStatsMID_v3: per-page lifetime stats
   - Aggregation for Learner input
   - Free hit ratio tracking (basis points)

3. smallobject_learner_v2_box.h
   - SmallLearnerStatsV2: multi-class and global metrics
   - Extended from v7 (C5-only ratio) to full workload analysis
   - Per-class retire efficiency, global free hit ratio
   - Decision API for route optimization

4. smallobject_policy_v2_box.h
   - SmallPolicyV2: routing with Learner integration
   - Version-based TLS cache invalidation
   - Route update from Learner stats
   - Backward compatible with v1 interface

Dependency graph:
  segment → stats → learner → policy → malloc routing

Architecture Decision: Option A (MID v3.5 consolidation)
- v7 frozen as C5/C6-only research preset
- MID v3.5 becomes 257-1KiB main implementation
- Learner scope: multi-class tracking (C5 ratio primary, Phase v11a)
- Future (v11b): multi-dimensional optimization

🤖 Generated with Claude Code

Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
 2025-12-12 06:20:01 +09:00
Moe Charm (CI)
bbc4b66a22 Phase v10: Enable Learner v7 by default 
Change: Learner now defaults to ON (when v7 is enabled)
- Old behavior: Learner only enabled if explicitly requested
- New behavior: Learner always ON (can disable with ENV=0)
- Learner is optional dependency of v7 (not intrusive)

Configuration:
- HAKMEM_SMALL_HEAP_V7_ENABLED=1: enables v7 + Learner
- HAKMEM_SMALL_LEARNER_V7_ENABLED=0: disable Learner only (keeps v7)

Benefits:
- Automatic workload detection without user configuration
- C5 allocation ratio monitored by default
- Route optimization happens transparently

Performance: v7+Learner C5/C6 workload = 39M ops/s (maintained)

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
 2025-12-12 06:09:53 +09:00
Moe Charm (CI)
79674c9390 Phase v10: Remove legacy v3/v4/v5 implementations 
Removal strategy: Deprecate routes by disabling ENV-based routing
- v3/v4/v5 enum types kept for binary compatibility
- small_heap_v3/v4/v5_enabled() always return 0
- small_heap_v3/v4/v5_class_enabled() always return 0
- Any v3/v4/v5 ENVs are silently ignored, routes to LEGACY

Changes:
- core/box/smallobject_hotbox_v3_env_box.h: stub functions
- core/box/smallobject_hotbox_v4_env_box.h: stub functions
- core/box/smallobject_v5_env_box.h: stub functions
- core/front/malloc_tiny_fast.h: remove alloc/free cases (20+ lines)

Benefits:
- Cleaner routing logic (v6/v7 only for SmallObject)
- 20+ lines deleted from hot path validation
- No behavioral change (routes were rarely used)

Performance: No regression expected (v3/v4/v5 already disabled by default)

Next: Set Learner v7 default ON, production testing

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
 2025-12-12 06:09:12 +09:00
Moe Charm (CI)
540230c301 v7-7: Modularize Learner into separate box 
Refactoring: Separate Learner API and types from Policy Box
- New: core/box/smallobject_learner_v7_box.h
  - SmallLearnerStatsV7 type definition
  - Learner recording API (record_refill, record_retire)
  - Learner evaluation and stats snapshot
  - Learner configuration constants
- Updated: core/box/smallobject_policy_v7_box.h
  - Removed Learner API (moved to Learner Box)
  - Removed SmallLearnerStatsV7 type (moved to Learner Box)
  - Added include of smallobject_learner_v7_box.h
  - Kept small_policy_v7_update_from_learner() (L3 integration)
- Updated: core/smallobject_policy_v7.c
  - Added include of smallobject_learner_v7_box.h

Benefits:
- Clearer module boundaries (Policy vs Learner)
- Easier testing and debugging (stats isolation)
- Reduced coupling between components

Performance: No regression (v7+Learner: 41M ops/s on C5/C6)

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
 2025-12-12 06:06:44 +09:00
Moe Charm (CI)
6c8c7b7f6c v7-5b/v7-7: Fix free path for C5 and Learner route switching 
Bug fixes:
- Free path now handles C5 (not just C6) for v7 routing
- After Learner route switch, old V7 pointers are correctly freed
  via V7 (instead of being misrouted to legacy)

Change: Always try V7 free for SMALL_V7_CLASS_SUPPORTED classes
(C5/C6). V7 returns false if ptr is not in V7 segment, allowing
proper fallback to legacy for non-V7 pointers.

This fix is essential because Learner may dynamically switch
C5 from V7→MID_V3, but pointers allocated before the switch
still reside in V7 segments and must be freed via V7.

Performance (C5/C6 workload 200-500B):
- v7 OFF: ~19M ops/s
- v7+Learner: ~43M ops/s (+126%)

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
 2025-12-12 06:02:13 +09:00
Moe Charm (CI)
6f559e1a1d v7-7: Implement Learner for dynamic C5 route switching 
- Add SmallLearnerStatsV7 type + API to policy box
- Hook ColdIface refill/retire to collect stats (capacity-based)
- Implement C5 route switching: if C5 ratio < 30%, switch to MID_V3
- Version-based TLS cache invalidation for policy updates
- Evaluation interval: every 100 refills

Tested with c6heavy scenario: C5 ratio=12% triggers V7 → MID_V3 switch

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
 2025-12-12 05:51:27 +09:00
Moe Charm (CI)
d5aa3110c6 Phase v7-5b: C5+C6 multi-class expansion (+4.3% improvement) 
- Add C5 (256B blocks) support alongside C6 (512B blocks)
- Same segment shared between C5/C6 (page_meta.class_idx distinguishes)
- SMALL_V7_CLASS_SUPPORTED() macro for class validation
- Extend small_v7_block_size() for C5 (switch statement)

A/B Result: C6-only v7 avg 7.64M ops/s → C5+C6 v7 avg 7.97M ops/s (+4.3%)
Criteria: C6 protected ✅, C5 net positive ✅, TLS bloat none ✅

ENV: HAKMEM_SMALL_HEAP_V7_CLASSES=0x60 (bit5+bit6)

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
 2025-12-12 05:11:02 +09:00
Moe Charm (CI)
17ceed619c Phase v7-5a: Hot path stats removal (C6 v7 極限最適化) 
- Remove per-page stats from hot path (alloc_count, free_count, live_current)
- Add ENV-gated global atomic stats (HAKMEM_V7_HOT_STATS)
- Stats now collected only at retire time (cold path)
- Header write kept at alloc time (freelist overlaps block[0])

A/B Result: -4.3% overhead → ±0% (target: legacy ±2%)
v7 OFF avg: 9.26M ops/s, v7 ON avg: 9.27M ops/s (+0.15%)

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
 2025-12-12 04:51:17 +09:00
Moe Charm (CI)
8143e8b797 Phase v7-4: Policy Box 導入 (L3 層の明確化とフロント芯の作り直し) 
- SmallPolicyV7 Box: L3 Policy layer に配置、route 決定を一元化
- Route kind enum: SMALL_ROUTE_ULTRA / V7 / MID_V3 / LEGACY
- ENV priority (fixed): ULTRA > v7 > MID_v3 > LEGACY
- Frontend integration: v7 routing を Policy Box 経由に変更 (段階移行)
- Legacy compatibility: 既存の tiny_route_env_box.h は併用維持

Box Theory layer structure:
- L0: ULTRA (C4-C7, FROZEN)
- L1: SmallObject v7 (research box)
- L1': MID_v3 / LEGACY (fallback)
- L2: Segment / RegionId
- L3: Policy / Stats / Learner ← Policy Box added here

Frontend now follows clean "size→class→route_kind→switch" pattern.
ENV variables read once at Policy init, not scattered across frontend.

Future: ULTRA/MID_v3/LEGACY consolidation, Learner integration, flexible priority.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
 2025-12-12 03:50:58 +09:00
Moe Charm (CI)
2bdf29a9ed Phase v7-3: TLS segment fast path optimization (RegionIdBox overhead reduction) 
- SmallHeapCtx_v7: Add TLS segment hints (tls_seg_base/end) for fast bounds check
- free fast path: TLS segment hit → skip RegionIdBox binary search
- Simplified control flow: removed same-page cache (negligible benefit vs branch cost)
- Optimization: O(1) page_idx calculation via bit shift vs O(log N) RegionIdBox lookup

Performance improvement:
- Phase v7-2: 54.5M ops/s (-7.0% vs 58.6M legacy)
- Phase v7-3: 56.3M ops/s (-4.3% vs legacy)
- Overhead reduction: 38% (from -7.0% to -4.3%)

TLS segment hit path bypasses RegionIdBox for most C6 frees.
Remaining -4.3% overhead acceptable for modular v7 architecture.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
 2025-12-12 03:38:39 +09:00
Moe Charm (CI)
39a3c53dbc Phase v7-2: SmallObject v7 C6-only implementation with RegionIdBox integration 
- SmallSegment_v7: 2MiB segment with TLS slot and free page stack
- ColdIface_v7: Page refill/retire between HotBox and SegmentBox
- HotBox_v7: Full C6-only alloc/free with header writing (HEADER_MAGIC|class_idx)
- Free path early-exit: Check v7 route BEFORE ss_fast_lookup (separate mmap segment)
- RegionIdBox: Register v7 segment for ptr->region lookup
- Benchmark: v7 ON ~54.5M ops/s (-7% overhead vs 58.6M legacy baseline)

v7 correctly balances alloc/free counts and page lifecycle.
RegionIdBox overhead identified as primary cost driver.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
 2025-12-12 03:12:28 +09:00
Moe Charm (CI)
a8d0ab06fc MID-V3: Specialize to 257-768B, exclude C7 (ULTRA handles 1KB) 
Role separation based on ultrathink analysis:
- MID v3: 257-768B専用 (C6 only, HAKMEM_MID_V3_CLASSES=0x40)
- C7 ULTRA: 769-1024B専用 (existing optimized path)

Changes:
- core/box/hak_alloc_api.inc.h: Remove C7 route, restrict to 257-768B
- core/box/mid_hotbox_v3_env_box.h: Update ENV comments
- docs/analysis/MID_POOL_V3_DESIGN.md: Add performance results & role
- CURRENT_TASK.md: Document MID-V3 completion & role separation

Verified:
- 257-768B with v3 ON: 1,199,526 ops/s (+1.7% vs baseline)
- 769-1024B with v3 ON: 1,181,254 ops/s (same as baseline, C7 excluded)
- C7 correctly routes to ULTRA instead of MID v3

Rationale: C7-only showed -11% regression, but C6/mixed showed +11-19%
improvement. Specializing to mid-range (257-768B) leverages v3 strengths
while keeping C7 on the proven ULTRA path.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
 2025-12-12 01:14:13 +09:00
Moe Charm (CI)
510cf338f3 MID-V3-6: hakmem.c integration (box modularization) 
Integrate MID/Pool v3 into hakmem.c main allocation path using
box modularization pattern.

Changes:
- core/hakmem.c: Include MID v3 headers
- core/box/hak_alloc_api.inc.h: Add v3 allocation gate
  - C6 (145-256B) and C7 (769-1024B) size classes
  - ENV opt-in via HAKMEM_MID_V3_ENABLED + HAKMEM_MID_V3_CLASSES
  - Priority: v6 > v3 > v4 > pool
- core/box/hak_free_api.inc.h: Add v3 free path
  - RegionIdBox lookup based ownership check
- Makefile: Add core/mid_hotbox_v3.o to TINY_BENCH_OBJS_BASE

ENV controls (default OFF):
  HAKMEM_MID_V3_ENABLED=1
  HAKMEM_MID_V3_CLASSES=0x40  (C6)
  HAKMEM_MID_V3_CLASSES=0x80  (C7)
  HAKMEM_MID_V3_DEBUG=1

Verified with bench_mid_large_mt_hakmem (7-9M ops/s, no crashes)

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
 2025-12-12 01:04:55 +09:00
Moe Charm (CI)
710541b69e MID-V3 Phase 3-5: RegionId integration, alloc/free implementation 
- MID-V3-3: RegionId integration (page registration at carve)
  - mid_segment_v3_carve_page(): Register with RegionIdBox
  - mid_segment_v3_return_page(): Unregister from RegionIdBox
  - Uses REGION_KIND_MID_V3 for region identification

- MID-V3-4: Allocation fast path implementation
  - mid_hot_v3_alloc_slow(): Slow path for lane miss
  - mid_cold_v3_refill_page(): Segment-based page allocation
  - mid_lane_refill_from_page(): Batch transfer (16 items default)
  - mid_page_build_freelist(): Initial freelist construction

- MID-V3-5: Free/cold path implementation
  - mid_hot_v3_free(): RegionIdBox lookup based free
  - mid_page_push_free(): Page freelist push
  - Local/remote page detection via lane ownership

ENV controls (default OFF):
  HAKMEM_MID_V3_ENABLED=1
  HAKMEM_MID_V3_CLASSES=0xC0 (C6+C7)
  HAKMEM_MID_V3_DEBUG=1

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
 2025-12-12 00:53:42 +09:00
Moe Charm (CI)
2b35de2123 MID-V3 Phase 0-2: Design doc, type skeleton, and RegionIdBox API 
- MID-V3-0: Create design doc (docs/analysis/MID_POOL_V3_DESIGN.md)
  - Lane vs Page role clarification
  - Phase plan and checklist

- MID-V3-1: Type skeleton + ENV
  - MidHotBoxV3, MidLaneV3, MidPageDescV3 structures
  - ENV controls (HAKMEM_MID_V3_ENABLED, HAKMEM_MID_V3_CLASSES)
  - Cold interface declarations

- MID-V3-2 (V6-HDR-2): RegionIdBox Registration API completion
  - RegionEntry structure with sorted array storage
  - Binary search lookup implementation
  - region_id_register_v6() / region_id_unregister_v6()
  - REGION_KIND_MID_V3 added to enum

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
 2025-12-12 00:46:25 +09:00
Moe Charm (CI)
ce372cfc7e Phase V6-HDR-4: Headerless 最適化 (P0 + P1) 
## P0: Double validation 排除
- region_id_lookup_v6() で TLS segment 登録済み + 範囲内なら
  small_page_meta_v6_of() を呼ばずに直接 page_meta を計算
- 削除された重複チェック:
  - slot->in_use (TLS登録で保証)
  - small_ptr_in_segment_v6() (addr範囲で既にチェック済み)
  - 関数呼び出しオーバーヘッド
- 推定効果: +1-2% (6-8 instructions 削減)

## P1: TLS cache に page_meta キャッシュ追加
- RegionIdTlsCache に追加:
  - last_page_base / last_page_end (ページ範囲)
  - last_page (SmallPageMetaV6* 直接ポインタ)
- region_id_lookup_cached_v6() で same-page hit 時は
  page_meta lookup を完全スキップ
- 推定効果: +1.5-2.5% (10-12 instructions 削減)

## ベンチマーク結果 (揺れあり)
- V6-HDR-3 (P0/P1 前): -3.5% ~ -8.3% 回帰
- V6-HDR-4 (P0+P1 後): +2.7% ~ +12% 改善 (一部の run で)

設計原則:
- RegionIdBox は薄く保つ (分類のみ)
- キャッシュは TLS 側に寄せる
- same-page 判定で last_page_base/end を使用

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
 2025-12-12 00:16:32 +09:00
Moe Charm (CI)
df216b6901 Phase V6-HDR-3: SmallSegmentV6 実割り当て & RegionIdBox Registration 
実装内容:
1. SmallSegmentV6のmmap割り当ては既に v6-0で実装済み
2. small_heap_ctx_v6() で segment 取得時に region_id_register_v6_segment() 呼び出し
3. region_id_v6.c に TLS スコープのセグメント登録ロジック実装:
   - 4つの static __thread 変数でセグメント情報をキャッシュ
   - region_id_register_v6_segment(): セグメント base/end を TLS に記録
   - region_id_lookup_v6(): TLS segment の range check を最初に実行
   - TLS cache 更新で O(1) lookup 実現
4. region_id_v6_box.h に SmallSegmentV6 type include & function 宣言追加
5. small_v6_region_observe_validate() に region_id_observe_lookup() 呼び出し追加

効果:
- HeaderlessデザインでRegionIdBoxが正式にSMALL_V6分類を返せるように
- TLS-scopedな簡潔な登録メカニズム (マルチスレッド対応)
- Fast path: TLS segment range check -> page_meta lookup
- Fall back path: 従来の small_page_meta_v6_of() による動的検出
- Latency: O(1) TLS cache hit rate がv6 alloc/free の大部分をカバー

🤖 Generated with Claude Code

Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
 2025-12-11 23:51:48 +09:00
Moe Charm (CI)
406835feb3 Phase V6-HDR-0: C6-only headerless core 設計確定 
- CURRENT_TASK.md: V6-HDR-0 セクション追加（4層 Box Theory）
- SMALLOBJECT_CORE_V6_DESIGN.md: V6-HDR-0 設計方針追加
- REGIONID_V6_DESIGN.md: RegionIdBox 設計書新規作成
- smallobject_core_v6_box.h: SmallTlsLaneV6 型＋TLS API 追加
- smallobject_core_v6.c: OBSERVE モード追加
- region_id_v6_box.h: RegionIdBox 型スケルトン
- page_stats_v6_box.h: PageStatsV6 箱スケルトン
- AGENTS.md: v6 研究箱ルールセクション追加

サニティベンチ: Mixed 42.1M, C6-heavy 25.0M（挙動不変確認）

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
 2025-12-11 23:07:26 +09:00
Moe Charm (CI)
2d684ffd25 Phase SO-BACKEND-OPT-1: v3 backend 分解＆Tiny/ULTRA 完成世代宣言 
=== 実装内容 ===

1. v3 backend 詳細計測
   - ENV: HAKMEM_SO_V3_STATS で alloc/free パス内訳計測
   - 追加 stats: alloc_current_hit, alloc_partial_hit, free_current, free_partial, free_retire
   - so_alloc_fast / so_free_fast に埋め込み
   - デストラクタで [ALLOC_DETAIL] / [FREE_DETAIL] 出力

2. v3 backend ボトルネック分析完了
   - C7-only: alloc_current_hit=99.99%, alloc_refill=0.9%, free_retire=0.1%, page_of_fail=0
   - Mixed: alloc_current_hit=100%, alloc_refill=0.85%, free_retire=0.07%, page_of_fail=0
   - 結論: v3 ロジック部分（ページ選択・retire）は完全最適化済み
   - 残り 5% overhead は内部コスト（header write, memcpy, 分岐）

3. Tiny/ULTRA 層「完成世代」宣言
   - 総括ドキュメント作成: docs/analysis/PERF_EXEC_SUMMARY_ULTRA_PHASE_20251211.md
   - CURRENT_TASK.md に Phase ULTRA 総括セクション追加
   - AGENTS.md に Tiny/ULTRA 完成世代宣言追加
   - 最終成果: Mixed 16–1024B = 43.9M ops/s (baseline 30.6M → +43.5%)

=== ボトルネック地図 ===

| 層 | 関数 | overhead |
|-----|------|----------|
| Front | malloc/free dispatcher | ~40–45% |
| ULTRA | C4–C7 alloc/free/refill | ~12% |
| v3 backend | so_alloc/so_free | ~5% |
| mid/pool | hak_super_lookup | 3–5% |

=== フェーズ履歴（Phase ULTRA cycle） ===

- Phase PERF-ULTRA-FREE-OPT-1: C4–C7 ULTRA統合 → +9.3%
- Phase REFACTOR: Code quality (60行削減)
- Phase PERF-ULTRA-REFILL-OPT-1a/1b: C7 ULTRA refill最適化 → +11.1%
- Phase SO-BACKEND-OPT-1: v3 backend分解 → 設計限界確認

=== 次フェーズ（独立ライン） ===

1. Phase SO-BACKEND-OPT-2: v3 header write削減 (1-2%)
2. Headerless/v6系: out-of-band header (1-2%)
3. mid/pool v3新設計: C6-heavy 10M → 20–25M

本フェーズでTiny/ULTRA層は「完成世代」として基盤固定。
今後の大きい変更はHeaderless/mid系の独立ラインで検討。

🤖 Generated with Claude Code

Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
 2025-12-11 22:45:14 +09:00
Moe Charm (CI)
fc1c47043c Phase PERF-ULTRA-REFILL-OPT-1a/1b: C7 ULTRA refill パス最適化 
実装内容:
- Phase 1a: Page size macro化
  - TINY_C7_ULTRA_PAGE_SHIFT (16) を定義
  - tiny_c7_ultra_page_of で division → bit shift に変更
  - refill/free での seg_end 計算を multiplication → bit shift に最適化

- Phase 1b: Segment learning を移動
  - segment learning を free初回 → alloc refill時に移動
  - free側での unlikely segment_from_ptr call を削除
  - normal pattern (alloc → free) での segment既学習を前提

ベンチマーク結果（Mixed 16-1024B, 1M iter, ws=400）:
  - Baseline: 39.5M ops/s
  - Phase 1a: 39.5M ops/s (誤差範囲)
  - Phase 1b: 42.3M ops/s
  - 最終平均: 43.9M ops/s (+11.1% = +4.4M ops/s)

tiny_c7_ultra_page_of は計測では同じ値だが、実際には以下が改善:
- division コスト削減（数cycle/call）
- free時のsegment learning削除（per-thread 1回削減）
- refill での計算簡素化

これにより全体の refill パス最適化が達成できました。
 2025-12-11 22:16:07 +09:00
Moe Charm (CI)
0f15adae4e Phase ALLOC-GATE-OPT-1: tiny_alloc_gate_fast 統計計測 
- AllocGateStats 構造体追加（size2class/route/env/class分布）
- malloc_tiny_fast にカウンタ埋め込み
- ENV: HAKMEM_ALLOC_GATE_STATS (default 0)
- 挙動変更なし（計測のみ）

計測結果:
- Mixed: total=542k, size2class=0, route_calls=0, env_checks=275k, C4-C7=95.2%
  - size_to_class/route_for_class は完全削減済み（LUT 効果）
  - C4-C7 が 95% → ULTRA fast path が有効
  - env_checks ≈ c7_calls → C7 ULTRA の ENV gate が毎回呼ばれる
- C6-heavy: total=11 → malloc_tiny_fast はほぼ通らない（mid/pool 主体）

結論:
- alloc gate は既に十分最適化済み（LUT + ULTRA で削減済み）
- さらなる最適化余地は小さい（env_checks は軽量化済み、数%以下の効果）
- 次フェーズでは free dispatcher (29%) や C7 ULTRA refill (7%) など、他のボトルネックを狙う

詳細: docs/analysis/ALLOC_GATE_ANALYSIS.md

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
 2025-12-11 21:32:40 +09:00
Moe Charm (CI)
118c0e4857 Phase FREE-DISPATCHER-OPT-1: free dispatcher 統計計測 
**目的**: free dispatcher（29%）の内訳を細分化して計測。

**実装内容**:
- FreeDispatchStats 構造体追加（ENV: HAKMEM_FREE_DISPATCH_STATS, default 0）
- カウンタ: total_calls / domain (tiny/mid/large) / route (ultra/legacy/pool/v6) / env_checks / route_for_class_calls
- hak_free_at / tiny_route_for_class / tiny_route_snapshot_init にカウンタ埋め込み
- 挙動変更なし（計測のみ、ENV OFF 時は overhead ゼロ）

**計測結果**:

Mixed 16-1024B (1M iter, ws=400):
- total=8,081, route_calls=267,967, env_checks=9
- BENCH_FAST_FRONT により大半は早期リターン
- route_for_class は主に alloc 側で呼ばれる（267k calls vs 8k frees）
- ENV check は初期化時の 9回のみ（snapshot 効果）

C6-heavy (257-768B, 1M iter, ws=400):
- total=500,099, route_calls=1,034, env_checks=9
- fg_classify_domain に到達する free が多い
- route_for_class 呼び出しは極小（snapshot 効果）

**結論**:
- ENV check は既に十分最適化されている（初期化時のみ）
- route_for_class は alloc 側での呼び出しが主で、free 側は snapshot で O(1)
- 次フェーズ（OPT-2）では別のアプローチを検討

**ドキュメント追加**:
- docs/analysis/FREE_DISPATCHER_ANALYSIS.md（新規）
- CURRENT_TASK.md に Phase FREE-DISPATCHER-OPT-1 セクション追加

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
 2025-12-11 21:21:40 +09:00
Moe Charm (CI)
11dc9d390a Phase PERF-ULTRA-FREE-OPT-1: C4-C7 ULTRA free 薄型化 
- C4-C7 ULTRA free を pure TLS push + cold segment learning に統一
- C7 ULTRA free を同じパターンに整列（likely/unlikely + FREE_PATH_STAT_INC）
- C4/C5/C6 ULTRA は既に最適化済み（統一 legacy fallback 経由）
- base/user 変換を tiny_ptr_convert_box.h マクロで統一

実測値 (Mixed 16-1024B, 1M iter, ws=400):
- Baseline (C7 のみ): 42.0M ops/s, legacy=266,943 (49.2%)
- Optimized (C4-C7): 46.5M ops/s, legacy=26,025 (4.8%)
- 改善: +9.3% (+4M ops/s)

FREE_PATH_STATS:
- C6 ULTRA: 137,319 free + 137,241 alloc (100% カバー)
- C5 ULTRA: 68,871 free + 68,827 alloc (100% カバー)
- C4 ULTRA: 34,727 free + 34,696 alloc (100% カバー)
- Legacy: 266,943 → 26,025 (−90.2%, C2/C3 のみ)

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
 2025-12-11 20:49:39 +09:00