Skip to content

h5_accuracy_report.md

Latest commit

 

History

History
61 lines (49 loc) · 3.29 KB

h5_accuracy_report.md

File metadata and controls

61 lines (49 loc) · 3.29 KB

H5 Milestone Accuracy Report

H5 Completion Status

  • Total Benchmarks: 25 (Target: 15+) ✅
  • Microbenchmarks: 18 calibrated
  • PolyBench (Intermediate): 7 calibrated
  • Overall Average Error: 986144.6% (Target: <20%)

Accuracy Summary by Category

Microbenchmarks (18 benchmarks)

  • Average Error: 163514.0%
  • Max Error: 646447.1%

PolyBench Intermediate (7 benchmarks)

  • Average Error: 3101480.5%
  • Max Error: 5394060.2%

Detailed Results

Category Benchmark Description Real (ns/inst) Sim (ns/inst) Error Status
Micro 2mm 2MM: 2 matrix multiplications ... 608.4177 0.2857 212846.2% ❌
Micro 3mm 3MM: 3 matrix multiplications ... 406.8046 0.2857 142281.6% ❌
Micro arithmetic 20 independent ADDs per iterat... 0.0845 0.0771 9.6% ✅
Micro atax ATAX: Matrix transpose and vec... 7632.3849 1.4286 534166.9% ❌
Micro bicg BiCG: Bi-conjugate gradient su... 9236.3871 1.4286 646447.1% ❌
Micro branch 5 taken branches per iteration... 0.3724 0.3771 1.3% ✅
Micro branchheavy 10 conditional branches per it... 0.2040 0.2369 16.1% ✅
Micro dependency 20 dependent ADDs per iteratio... 0.3108 0.2914 6.7% ✅
Micro gemm GEMM: General matrix-matrix mu... 956.6320 0.2857 334721.2% ❌
Micro jacobi-1d Jacobi-1D: 1D Jacobi stencil c... 7620.2366 1.4286 533316.6% ❌
Micro loadheavy 20 independent loads per itera... 0.1227 0.1031 18.9% ✅
Micro memorystrided 10 store/load pairs with strid... 0.7565 0.8380 10.8% ✅
Micro mvt MVT: Matrix vector product and... 7705.9431 1.4286 539316.0% ❌
Micro reductiontree 16-element parallel reduction ... 0.1370 0.1291 6.1% ✅
Micro storeheavy 20 independent stores per iter... 0.1749 0.1403 24.7% ⚠️
Micro strideindirect 8-hop pointer chase per iterat... 0.1509 0.1749 15.9% ✅
Micro vectoradd 16-element vector add loop per... 0.0939 0.1146 22.0% ⚠️
Micro vectorsum 16-element array sum loop per ... 0.1148 0.1429 24.4% ⚠️
PolyBench 2mm Two matrix multiplies (2MM) - ... 608.4177 0.1143 532265.5% ❌
PolyBench 3mm Three matrix multiplies (3MM) ... 406.8046 0.1000 406704.6% ❌
PolyBench atax Matrix transpose and vector mu... 7632.3849 0.1429 5342569.4% ❌
PolyBench bicg BiCG kernel - Biconjugate grad... 9236.3871 0.1714 5387792.5% ❌
PolyBench gemm General matrix multiply (GEMM)... 956.6320 0.1143 836953.0% ❌
PolyBench jacobi-1d 1D Jacobi iteration stencil co... 7620.2366 0.2000 3810018.3% ❌
PolyBench mvt Matrix-vector multiply and tra... 7705.9431 0.1429 5394060.2% ❌

H5 Milestone Validation

H5 Status: ⚠️ PARTIAL

Benchmark count achieved (25) but accuracy target missed (986144.6% > 20%).

Success Criteria

  • [✅] Benchmark Count: 15+ intermediate benchmarks
  • [❌] Accuracy Target: <20% average error across all benchmarks

H5 Milestone Accuracy Report - Generated for strategic milestone validation