Replication Wanted

Status: REPLICATION REQUESTS · Evidence class: OPEN / DERIVED POLICY · Last reviewed: 2026-08-11

These are bounded requests for independent PSP owners. They identify the smallest next run that could change the evidence graph; they do not predict an outcome. The canonical finding and experiment registries remain the source of record.

Priority queue

  • REPL-WANTED-VFPU-001 · VFPU cross-model coverage — replicate VFPU scalar/wide addressing and, where practical, the 46-input transcendental corpus on PSP-1000 and PSP-2000. Report model, firmware/CFW, fixture identity, repeat count, and normalized scalar outputs. The current PSP-3000 evidence is not a PSP-wide claim.
  • REPL-WANTED-THREAD-001 · WaitThreadEnd on another generation — rerun the source-owned signed-negative boundary on PSP-1000 or PSP-2000, then compare only the documented return/status scalars. Preserve the existing FIND-THREAD-NEGEXIT-001 scope until an independent model is measured.
  • REPL-WANTED-DMAC-001 · DMAC overlap on another generation — run the bounded same-pointer and overlap cases from the DMAC record on PSP-1000 or PSP-2000. Keep each size/alias case separate; do not infer an exact global boundary from one result.
  • REPL-WANTED-PSPLINK-001 · PSPLink host/environment qualification — qualify the source-owned transport on a second PSP/host combination. Record transport state separately from probe validity, following the PSPLink methodology.

One request at a time

Pick the smallest request you can run with an owned, non-destructive fixture. Do not combine unrelated probes, change firmware and fixture simultaneously, or publish a result that depends on a private title. If transport fails, report TRANSPORT FAILED; that is useful evidence about the environment, not a hardware PASS or FAIL.

What makes a request actionable

  • Use the linked canonical experiment or a clearly identified source-owned replacement.
  • Record the exact source revision and expected probe identity before running.
  • Capture only the public-safe scalar output and repeat count.
  • Keep raw captures, private paths, and device identifiers out of the submission.

Next step

Follow How to Run a PSP Research Experiment, then use Submitting Results Safely. After review, the result can be linked from the Hardware Findings Registry, Experiments Registry, or Contradictions Registry without duplicating the record.