Repository navigation
Stage: Sonic Unleashed Recompiled (PPSA99902) misreads CPU tests and a gameplay negation as main progress #60
Description
Title ID / game
PPSA99902 — Sonic Unleashed Recompiled, 01.003.000.
Stage on the page now
Game entry: Presents on main (Windows / AMD), plus an unverified Ingame-on-main claim. Observed snapshot: 2026-10-13T16:52:31Z.
Correction requested
I authored the cited reports. Neither cited sentence establishes the stage assigned to it:
- Suppress the Presents inference from AnyPS5 #3475. Its exact sentence begins: "This is a source-model consistency defect; PS5 hardware/render equivalence and a title hitting these particular cases have not been measured." It reports CPU offset comparisons against AMD PAL under a synthetic configuration, not presented game frames on AMD hardware. The live data names
renderas the matched keyword and classifies it as Windows/AMD/main. - Suppress the Ingame inference from #3371 comment6108079462. Its exact sentence says: "Normal before/changed twice/restored PPSA99902 controls reach60s without a fatal diagnostic; they do not by themselves establish late startup or gameplay." The same comment identifies a pinned-main integration with additional changes and Windows11/RTX5080, not an unchanged-main/Linux run. The live data matches
gameplay, classifies it as Linux/main and exposes the unverified claim.
Please review the remaining inferred main reports before choosing a replacement main stage. I am not requesting a promotion of main to Menus, Ingame or Playable.
What our game runs actually establish
Historical local patched integration reached PRESS START, verified menu interaction and opening movie on Windows11 / NVIDIA RTX5080: canonical #419 report. It explicitly leaves interactive gameplay and unchanged upstream main unverified. This can support an off-main Menus record, not a ranked main-stage claim; stage_on_branches is currently null.
The current, narrower a353/main021 integration stops at normal-render 8x DCC in new AnyPS5 #5336, confirmed with a diagnostic-only replay. Bring-up discussion #4443 maintains the pins, next stop and historical scope. #4285's alpha stop is historical on a different integration; it is not the current integration's first stop. #5336 was published after the observed snapshot, so its absence may only be normal collection/rebuild delay.
Build check
This is a correction of mined source interpretation, not a new unchanged-main stage report. Local integration progress includes unmerged/local work, so I cannot truthfully check the form's main-only-stage checkbox. Suggested data action: suppress the two false stage matches; retain correctly scoped fork progress and refresh the conditional blocker ordering. No game assets attached.
AI-assisted investigation/reporting: Codex.
Activity
Do not read negated claims, CPU-only checks or render nouns as stage …
Do not read negated claims, CPU-only checks or render nouns as stage …
AlxcBTW commentedon Oct 13, 2026
Thank you @Amoenus, both reports were right, and they were generic misreads rather than one-offs. Fixed in #61 (merged).
What changed in the stage extraction:
renderas a noun (render target, render thread, hardware/render equivalence) no longer counts as "renders".Over the cached data this removes 6 mined stage rows (Ingame 92 -> 85, Presents 161 -> 154) and moves a few Ingame claims to the Menus they actually support (Demon's Souls follow-ups, Blasphemous 2). For PPSA99902 the main stage is now Starts (stops at the AGC multisample-target error, PR #3644 report); the off-main Menus record from #419 stays as the branch stage. #5336 should appear with the next collection. The live page updates on the next rebuild.
Amoenus commentedon Oct 15, 2026
Snapshot follow-up: 2026-10-24 UTC
The original fixes above are appreciated. The current game entry still needs a build-scope review; this is not a request to promote unchanged main to Ingame.
stage_on_branchesis null even though the entry's own evidence includes patched-fork gameplay. Our completed integration60c0cc1845, incorporating main0db47b263d1019043f7d28fad4e7ee0a1ea5ec6a, reached daytime movement, 32 rings, pots and the Homing Attack instruction, then aborted at a mesh-stage TYPE refusal. Pinned completed-run evidence and limits; maintained bring-up discussion. This supports Ingame on a patched branch, not Playable or a ranked main milestone. No minimal prerequisite set has been established.stagerow labels itsb9b8c5ab7052summarybuild_kind=main. The issue explicitly identifies baseline/change as composed forks on main83462cc57ddecac5e5f3cdafafbe38d406e55953, not unchanged main. A clean working tree and apresentscounter do not identify the build as upstream main. Please reclassify that evidence and choose the main stage from independently qualifying reports.0e7fba38b5. That composition also carried other changes; merging the profiling PR does not establish that all its gameplay prerequisites merged. Please keep that observation scoped to the fork.11e8ad293422aad5e30e6c8f5c85048d4dc56649, entered the menu, skipped the opening movie and stopped on NOW LOADING before gameplay, at the existing #3036 image-descriptor refusal after 412.270 seconds. The earlier a9ba30003d replay reached daytime gameplay, 32 rings, pots and the cliff, then stopped at a mesh-stage buffer TYPE refusal recorded under #5879. That conditional Sonic stop is absent from the current game entry's open-blocker list. Please retain it with its fork scope and review the mined #7260-first ordering: these reports do not establish a single blocker order on unchanged main, nor prove that all TYPE refusals share one root cause.Windows 11 Enterprise 26200, RTX 5080 / NVIDIA 610.62, PPSA99902 /01.003.000. The two newer completed runs above have different composed source pins; neither establishes unchanged-main gameplay or isolates a regression. Historical blockers are conditional on their source pins; this follow-up does not claim a fresh main blocker order. I have edited this existing report to avoid another duplicate issue/comment. AI-assisted investigation/reporting: yes (Codex); author responsible.
Previous follow-up, 2026-10-14: render-decoder noun inference
Follow-up to #61: another noun context is still counted as stage evidence.
In the live snapshot generated 2026-10-14T22:34:57Z, PPSA99902 is First draw / main / Windows / NVIDIA, sourced from AnyPS5 #5939, with matched keyword
renderand this exact quote:I authored that sentence. It names a descriptor-decoding refusal; it does not report a completed draw or rendered frame. I have clarified the source to “stops during descriptor decoding on main”. Please also cover
render decoder/ hyphenatednormal-render decoderin the generic noun rule, so similar reports do not create stage evidence. The exact old sentence above can serve as a regression fixture; choose the remaining main stage from independently qualifying evidence.The off-main Menus classification remains appropriate. Our newer maintained bring-up report and #5851 validation reach the title screen, opening cutscene and level loading, then stop at resident-MSAA revalidation. That run uses main4808 + open #5851 + retained local patches. It does not establish gameplay or a higher main stage.
AI-assisted: yes.