RAW BENCHMARK DATA
Raw Data. Every Number. Every Run.
All benchmark data is published on GitHub. SHA-256 sealed. Verifiable by any engineer with a terminal. No trust required.
V25 SEAL
Engine VersionV25
Fingerprint66c8f2d94b3e4725
Core SHAbd615f15
Sealed At2026-10-25T17:07:20+00:00
Twin Pairs17 sealed
Paired Reps97 per category
Monte Carlo10,000,000 runs
Hash AlgorithmSHA-256
RAW V25 VALUES
| Metric | Raw Value | Displayed As | Source |
|---|---|---|---|
| headline_pct (omni-v26) | 19.968 | +20.0% | OMNI_INDEX.json |
| service_headline_pct (omni-v25) | 93.97 | +94.0% (not shown on site — work+speed only, not a bill) | OMNI_INDEX.json |
| Kubernetes (omni-v26) | 22.229 | +22.2% | OMNI_INDEX.json |
| PostgreSQL | 4.1658 | +4.2% | ABC_TABLE.json |
| MySQL | 2.205 | +2.2% | ABC_TABLE.json |
| Redis | 0.112 | +0.1% | ABC_TABLE.json |
| Chip energy per request (A10, omni-v23) | −1.3% | GAIN, down 1.3% | SITE_NUMBERS.json → show[chip_energy_per_request] |
| ABC_TABLE web A mean_ms | 215.089 | 215ms | ABC_TABLE.json |
| ABC_TABLE web A work | 0.99999 | 99.999% | ABC_TABLE.json |
EXTERNAL REFERENCES
SPEC CPU benchmarks ↗TPC-C benchmark ↗TPC-H benchmark ↗Phoronix Test Suite ↗MLPerf Inference ↗MLPerf Training ↗Monte Carlo method ↗bootstrap resampling ↗paired t-test ↗Welch's t-test ↗Cohen's d ↗95% confidence interval ↗SHA-256 ↗reproducibility in CS benchmarks ↗Kubernetes ↗Redis ↗PostgreSQL ↗Apache Kafka ↗MongoDB ↗MySQL ↗NVIDIA H100 datasheet ↗Lyapunov stability ↗finite-time convergence ↗