tomoaki/hakmem
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Update CURRENT_TASK.md: Phase 6 complete, next phase selection

Browse Source
This commit is contained in:
Moe Charm (CI)
2025-11-29 15:53:05 +09:00
parent 92cc187fa1
commit 1468efadd7
1 changed files with 22 additions and 24 deletions
Download Patch File Download Diff File Expand all files Collapse all files

46
CURRENT_TASK.md
View File

@ -1,25 +1,22 @@
# Current Task: Choose Next Phase # Current Task: Choose Next Phase
**Date**: 2025-11-29 **Date**: 2025-11-29
**Status**: Phase 5 ✅ COMPLETE → Next phase selection **Status**: Phase 6 ✅ COMPLETE → Next phase selection
**Achievement**: +28.9x improvement for Mid MT allocations (1KB-8KB) **Achievement**: Lock-free Mid MT (+2.65% improvement, code quality++)
--- ---
## Phase 5 Complete! ✅ ## Phase 6 Complete! ✅
**Result**: Mid/Large Allocation Optimization **COMPLETE** **Result**: Lock-free Mid MT Allocator **COMPLETE**
**Performance**: 1.49M → 41.0M ops/s (+28.9x for Mid MT, 1.53x faster than system malloc) **Performance**: 41.0M → 42.09M ops/s (+2.65% for Mid MT)
**Duration**: 1 day (focused execution) **Duration**: 1 day (quick improvement)
**Completed Steps**: **Completed Steps**:
- ✅ Step 1: Mid MT Verification (range bug identified) - ✅ Phase 6-A: Code readability (debug guard around SuperSlab lookup)
- ✅ Step 2: Mid Free Route Box (+28.9x improvement) - ✅ Phase 6-B: Header-based Mid MT free (lock-free, -127 lines)
- ✅ Step 3: Mid/Large Config Box (future workload infrastructure)
- ⏸️ Step 4: Mid Registry Pre-alloc (deferred, MT workload needed)
- ✅ Step 5: Documentation (PHASE5_COMPLETION_REPORT.md)
**See**: `PHASE5_COMPLETION_REPORT.md` for full details **See**: `PHASE6A_DISCREPANCY_INVESTIGATION.md` and `PHASE6B_DISCREPANCY_INVESTIGATION.md`
--- ---
@ -39,7 +36,7 @@
**Cons**: **Cons**:
- May be system noise (not real regression) - May be system noise (not real regression)
- Workload is Tiny-only (unaffected by Phase 5 changes) - Workload is Tiny-only (unaffected by Phase 5/6 changes)
- Could be time spent on noise instead of real gains - Could be time spent on noise instead of real gains
--- ---
@ -117,7 +114,7 @@
**Risk**: High (no MT benchmark exists yet) **Risk**: High (no MT benchmark exists yet)
**Pros**: **Pros**:
- Unlock Phase 5-Step4 (Mid registry pre-allocation) - Unlock Phase 5-Step4 (Mid registry pre-allocation, now obsolete with Phase 6-B)
- Real-world workloads are often MT - Real-world workloads are often MT
- Could show significant MT scalability gains - Could show significant MT scalability gains
@ -129,7 +126,7 @@
**Required Work**: **Required Work**:
1. Create MT benchmark (4+ threads, mixed sizes) 1. Create MT benchmark (4+ threads, mixed sizes)
2. Profile MT contention points 2. Profile MT contention points
3. Implement registry pre-allocation 3. Implement remote free (currently memory leak)
4. Add lock-free structures where needed 4. Add lock-free structures where needed
5. Validate MT correctness (TSAN, stress testing) 5. Validate MT correctness (TSAN, stress testing)
@ -196,21 +193,22 @@ Phase 4-Step3 (full): ~55-58 M ops/s (+5-8% expected)
``` ```
Phase 3 (mincore removal): 56.8 M ops/s Phase 3 (mincore removal): 56.8 M ops/s
Phase 4 (Hot/Cold Box): 57.2 M ops/s (+0.7%) Phase 4 (Hot/Cold Box): 57.2 M ops/s (+0.7%)
Phase 5 (current): 52.3 M ops/s (-8.6% regression) Phase 5/6 (current): 52.3 M ops/s (-8.6% regression)
``` ```
**Note**: Regression unrelated to Phase 5 (Tiny-only workload, doesn't touch Mid MT) **Note**: Regression unrelated to Phase 5/6 (Tiny-only workload, doesn't touch Mid MT)
### bench_mid_mt_gap (1KB-8KB, Mid MT workload) ### bench_mid_mt_gap (1KB-8KB, Mid MT workload)
``` ```
Before Phase 5 (broken): 1.49 M ops/s (mmap fallback) Before Phase 5 (broken): 1.49 M ops/s (mmap fallback)
After Phase 5 (fixed): 41.0 M ops/s (+28.9x) After Phase 5 (fixed): 41.0 M ops/s (+28.9x)
vs System malloc: 26.8 M ops/s (1.53x faster) After Phase 6-B (lock-free): 42.09 M ops/s (+2.65%)
vs System malloc: 26.8 M ops/s (1.57x faster)
``` ```
**Achievement**: ✅ Major success! **Achievement**: ✅ Major success! Lock-free, simpler code
### Overall Status ### Overall Status
- ✅ **Tiny allocations** (16B-1KB): 52-57 M ops/s (good, some regression) - ✅ **Tiny allocations** (16B-1KB): 52-57 M ops/s (good, some regression)
- ✅ **Mid MT allocations** (1KB-8KB): 41 M ops/s (excellent, 1.53x vs system) - ✅ **Mid MT allocations** (1KB-8KB): 42 M ops/s (excellent, 1.57x vs system, lock-free)
- ⏸️ **Large allocations** (32KB-2MB): Not benchmarked yet - ⏸️ **Large allocations** (32KB-2MB): Not benchmarked yet
- ⏸️ **MT workloads**: No MT benchmarks yet - ⏸️ **MT workloads**: No MT benchmarks yet
@ -225,11 +223,11 @@ vs System malloc: 26.8 M ops/s (1.53x faster)
- **Option D**: Production readiness & benchmarking - **Option D**: Production readiness & benchmarking
- **Option E**: Multi-threaded optimization - **Option E**: Multi-threaded optimization
**Or**: Take a break, Phase 5 is a big win! 🎉 **Or**: Take a break, Phase 5+6 are big wins! 🎉
--- ---
Updated: 2025-11-29 Updated: 2025-11-29
Phase: 5 COMPLETE → 6 PENDING Phase: 6 COMPLETE → 7 PENDING
Previous: Phase 4 (Tiny Front Optimization, +7.3%) Previous: Phase 5 (Mid/Large Optimization, +28.9x), Phase 6 (Lock-free Mid MT, +2.65%)
Achievement: +28.9x Mid MT improvement (1.49M → 41.0M ops/s) Achievement: Lock-free Mid MT allocator (42.09M ops/s, -127 lines code)