EXP-081EXPERIMENTAL
Canonical GPU engineering / SHA-256d

Long energy and thermal test of observable P8

P8-S1 preserves P8-S0 performance and meets paired energy noninferiority during long direct-counter runs.

Claim statusNO END-TO-END MINING ADVANTAGE DEMONSTRATED

Diagnostic work, exact algebra and local capabilities are not treated as end-to-end mining advantage.

Date2026-08-16
HardwareNVIDIA GeForce RTX 5080 · exact CUDA SHA-256d verifier
Run scopePreregistered paired campaign with sealed validation and audit
ReproducibilityRepeat the frozen S0/S1 long paired protocol on eight unseen headers with direct energy and the unchanged 1.003 bound.
01 / Question & hypothesis

What was tested?

P8-S1 preserves P8-S0 performance and meets paired energy noninferiority during long direct-counter runs.

02 / Scientific basis

Why the test is meaningful

A latency option can replace the default only if its synchronization overhead is bounded in throughput, energy and temperature under a sealed paired protocol.

R_E=(J/hash)_S1/(J/hash)_S0energy gate: CI95_upper≤1.003R_process=throughput_S1/throughput_S0
Long energy and thermal test of observable P8Visual reading of the published metrics and gates for EXP-081; it summarizes the registered result, not a mining advantage.EXP-081 / LONG ENERGY AND THERMAL TEST OF OBSERVABLE P8PROCESS RETENTION0.999965ENERGY RATIO0.998873CI95 UPPER1.003106
FIGURE / RESULT READINGVisual reading of the published metrics and gates for EXP-081; it summarizes the registered result, not a mining advantage.
03 / Method

How it was tested

Run 64 long paired processes on four new headers, two ABBA+BAAB cycles each, with frozen S0/S1, cumulative NVML energy and at least 507 thermal samples per process.

04 / Observed result

What happened

Process retention0.999965
Energy ratio0.998873
CI95 upper1.003106

S1 retained 0.999833 kernel and 0.999965 process throughput; observation was 0.020378 s. Energy/hash was 0.998873 and 3/4 headers passed, but CI95 upper was 1.003106—0.000106 above the fixed gate. Maximum temperature was 66°C.

05 / Validation

Exactness and statistical controls

SHA-256d/B32, compiled factors, pairing and thermal gates passed; the preregistered statistical boundary alone failed.

06 / Interpretation

What the result means

The valid campaign shows no meaningful point regression, but cannot promote S1. P8-S0 remains default pending an independent long replication.

Limitations

  • The CI missed the energy gate by 0.000106.
  • Only four headers were used.
  • A favorable point estimate is not confirmation.
07 / Reproduction

Evidence trail

Repeat the frozen S0/S1 long paired protocol on eight unseen headers with direct energy and the unchanged 1.003 bound.

Canonical variants

CANONICAL-EXP-157PREREGISTERED-CAMPAIGNSEALED-AUDIT

Source: internally audited canonical reports. Local filesystem structure, private headers and operational identifiers are excluded from publication.