Initial P8 energy and thermal confirmation
P8 retains its throughput gain while meeting paired energy/thermal noninferiority.
Diagnostic work, exact algebra and local capabilities are not treated as end-to-end mining advantage.
What was tested?
P8 retains its throughput gain while meeting paired energy/thermal noninferiority.
Why the test is meaningful
Even an independently replicated scheduler needs energy confirmation; an odd number of order cycles can leave residual external-position confounding.
Eratio=(J/hash)_P8/(J/hash)_P1noninferiority requires header consistency and bounded CIthree cycles create 2:1 external-order imbalanceHow it was tested
Measure four unseen headers, three ABBA+BAAB cycles per header, 96 processes and direct 100-Hz NVML telemetry.
What happened
Speed persisted at 1.003089× kernel and 1.003261× process. Aggregate energy ratio was 0.995164, but only 2/4 headers passed and paired 95% interval was 0.984555–1.005888. Header parity perfectly tracked the residual order imbalance.
Exactness and statistical controls
The energy endpoint is kept inconclusive rather than averaged into a favorable claim. Exactness, depth and thermal checks passed.
What the result means
P8 throughput is confirmed, while energy remains unresolved until four balanced cycles give each external order equal representation.
Limitations
- Three cycles are externally unbalanced.
- Energy is neither confirmed nor rejected.
- Maximum observed temperature was 65°C.
Evidence trail
Repeat with four cycles per header, two of each external sequence, preserving sealed P1/P8 binaries.
Canonical variants
CANONICAL-EXP-151PREREGISTERED-CAMPAIGNSEALED-AUDITSource: internally audited canonical reports. Local filesystem structure, private headers and operational identifiers are excluded from publication.