VALIDATION
Validation results and comparison layers.
This page collects the measured outputs, comparison logic, and archive-bound interpretation that sit behind the public benchmark claims.
Wall-clock runtime, clean-binary timing, and cross-system comparison remain separate methodological objects and are reported with their own context.
Measured logic
The validation layer keeps the measured runtime, the comparison method, and the release boundary separated so each result can be interpreted on its own terms.
- Deterministic execution under ARM64 benchmark conditions
- Comparison against independent computational references
- QA summaries and archive manifests tied to the release layer
Cross-system comparison
MAYAN_ALFA measurements are read together with MR and primesieve references when the archive evidence needs a stricter comparison path.
- MR comparison layers
- primesieve validation references
- QA mismatch control
Publication discipline
Public releases remain bounded at 10B, while larger runs are archived separately and prepared as controlled evidence for deeper review.
- Public presentation stops at 10B
- 100B+ lives in controlled archive evidence
- Later instrumentation will refine binary timing
Comparison overview
| Limit | MAYAN_ALFA | MR only | PRIMESIEVE | pi(N) | Status |
|---|---|---|---|---|---|
| 10M | 1.684354 s | 3.604905 s | 1.000000 s | 664579 | OK |
| 100M | 0.360000 s | 23.874431 s | 1.000000 s | 5761455 | OK |
| 1B | 1.793206 s | 244.145627 s | 9.000000 s | 50847534 | OK |
| 10B | 18.806417 s | NOT RUN | 94.000000 s | 455052511 | OK |
Interpretive note
Measured values represent internal benchmark execution time under specific ARM64 testing conditions and should not be interpreted as universal performance metrics. Results depend on hardware configuration, compiler optimization, execution methodology, and dataset structure.
All benchmark outputs were cross-validated against independent computational methods and archived within the MAYAN_ALFA validation framework.
