Skip to main content
Print

Hardware Findings Registry

Registry ID: REG-HARDWARE-FINDINGS-001 · Purpose: accepted and bounded source-owned PSP findings, ordered by subsystem and stable ID.

This registry separates measured behavior from prior-art reconfirmation, methodology findings, and preliminary work. A single PSP-3001 result is not silently promoted to “all PSP models.”

Hardware findings

Finding ID Title Evidence / hardware scope Confidence Novelty posture Status
FIND-THREAD-NEGEXIT-001 WaitThreadEnd signed-negative boundary PSP-3001 / 6.61-ARK ONE-RUN PRELIMINARY N3 candidate MEASURED · REPLICATION WANTED
FIND-VFPU-TRANSCEND-001 VFPU transcendental corpus PSP-3001 / 6.61-ARK ONE-RUN PRELIMINARY N3/N4 candidate MEASURED · REPLICATION WANTED
FIND-VFPU-DENORM-001 VFPU denormal edge PSP-3001 / 6.61-ARK; one input ONE-RUN PRELIMINARY N3/N4 candidate MEASURED · REPLICATION WANTED
FIND-VFPU-ADDR-001 VFPU register addressing PSP-3001 / 6.61-ARK; 128 scalar + 14 wide REPEATED SAME DEVICE N1/N2 prior-art confirmation MEASURED · CROSS-MODEL NEEDED
FIND-DMAC-OVERLAP-001 DMAC overlap PSP-3001 / 6.61-ARK; bounded RAM cases REPEATED SAME DEVICE N2/N3 candidate MEASURED · CROSS-MODEL NEEDED
FIND-PSPLINK-TRANSPORT-001 PSPLink qualification same PSP A/B: Windows vs WSL2 REPEATED SAME DEVICE N2/N3 methodology MEASURED · ENVIRONMENT-SCOPED

Model / firmware matrix

NOT TESTED is intentional. Cells do not infer results.

Finding PSP-1000 PSP-2000 PSP-3000 PSP Go PSP Street
ThreadMan signed-negative NOT TESTED NOT TESTED MEASURED · 6.61 ARK NOT TESTED NOT TESTED
VFPU transcendental / denormal NOT TESTED NOT TESTED MEASURED · 6.61 ARK NOT TESTED NOT TESTED
VFPU addressing NOT TESTED NOT TESTED MEASURED · 6.61 ARK NOT TESTED NOT TESTED
DMAC overlap NOT TESTED NOT TESTED MEASURED · 6.61 ARK NOT TESTED NOT TESTED
PSPLink transport NOT TESTED NOT TESTED MEASURED · host A/B NOT TESTED NOT TESTED

Replication requests

Highest-value next devices are PSP-1000 and PSP-2000 for VFPU corpus/addressing and PSP-3000/another generation for ThreadMan and DMAC. Contributors should report model family, firmware/CFW, fixture/source identity, scalar output, and repeat count only. Do not submit retail game data, keys, saves, serial/MAC/account identifiers, or private captures.

Related: Experiments · Contradictions · Hardware Oracle Methodology · Last reviewed: 2026-08-11.

Replication path: Start at Hardware Replication, choose a request in Replication Wanted, follow the experiment runbook, and use Submitting Results Safely before posting a text-only report.

Table of Contents