Octane Pixel Determinism Versus Runtime Instability

Frozen Octane workload on one hardware configuration.

Engine: OctaneRender 2026.3
Contract: 1.0.2
Hardware: Two NVIDIA GeForce RTX 3090 GPUs
Measured passes: 6

Workload SHA-256: 32771d4753270bfdf81500e69bd2ea33267c82081fad0062c3c32e663a99be37

FARPY tested output determinism separately from runtime stability. Matching rendered pixels are evidence of output consistency. They do not prove that execution time is stable enough to publish a performance score.

Six observed runtimes.

Required runtime CV: ≤ 3%
Best consecutive three-pass CV: 15.579055%

The measured variance exceeded the frozen acceptance gate. FARPY therefore rejected the benchmark result.

Decoded pixel payloads matched: YES
Complete PNG hashes matched: NO
Recorded cause of the full-file hash difference: changing PNG metadata.

REJECTED benchmark. Useful finding.

Output determinism was observed under the tested workload, but runtime stability was insufficient for a verified performance score.

Performance result published: NO
Verified Octane results published: 0

Limitations of this record.

• The record covers one known hardware configuration: Two NVIDIA GeForce RTX 3090 GPUs.
• The record covers one workload and contract version: OctaneRender 2026.3 1.0.2.
• The execution count is limited to 6 measured passes.
• These observations should not be generalized to all Octane versions, GPUs, drivers, or scenes.
• This report does not claim Octane is generally nondeterministic or unsuitable for benchmarking.
• No verified Octane performance score is published from this record.

Trace the record.

OCTANE-BENCHMARK-V1.0.2

octane-benchmark-v1.0.2-contract-20260717T235242Z

Published: 2026-07-20