The official AnyPS5 compatibility table lists Dreaming Sarah as playable on Windows. Community reports describe further progress, including patched Silksong gameplay, but they do not establish general support for PS5 games. Match each result to its exact environment before deciding what to try.
“Official” means the AnyPS5 project's own documentation. Community reports below are attributed experiments, often using development branches and local changes. We reviewed the original reports and later replies; we did not run the games.
The official compatibility table
The checked upstream table has one game: Dreaming Sarah, PPSA02929. Windows is marked “In game, playable”; Linux is marked ?. Its two performance entries are:
| Reported hardware | Recorded performance |
|---|---|
| GTX 1050 Ti / i5-7500 3.4GHz | 60 FPS |
| Intel(R) HD Graphic 620 / i5-7200 2.5GHz | 36 FPS |
The table does not record the AnyPS5 build, game version, Windows version, driver, resolution, test duration or completion coverage. Those fields remain unknown. These are observations for one title, not universal minimum specifications. An unlisted game is not automatically incompatible.
How to read a compatibility result
Use these stages to interpret the reports on this page; they are editorial descriptions, not a replacement for upstream labels.
- Converted: output was generated. It may still fail during loading.
- Started or reached menus: some runtime paths worked. Gameplay remains unproven.
- In game: a level ran. Rendering, performance or controls may prevent useful play.
- Playable in the report: the author describes usable gameplay under stated conditions. Full-game completion is a separate claim.
A high menu frame rate cannot substitute for a gameplay measurement. A successful test with extra patches cannot be assigned to an unmodified release.
Community reports and their limits
These selected records illustrate different stages. They are not an exhaustive supported-games list, and their results are not reproduced by this website.
| Game | Reported stage | Key limitation |
|---|---|---|
| Silksong | Playable opening areas | Patched Windows build and required switch |
| Astro Bot | Crash Site at 11.1 FPS | Local optimization stack; limited performance |
| Demon’s Souls | In-game experiments | Low performance; newer builds differ |
| Palworld | Startup after imports bind | Runtime blockers; no gameplay proof |
Hollow Knight: Silksong — patched gameplay
Environment: PPSA12544; Windows 11 Pro, Ryzen 7 9800X3D, RTX 5090, NVIDIA driver 617.42. Build: main d70b899, WinLibs GCC 15.2.0 r7 Release, seven listed PRs and APS5_STORE_AT_FLIP=1.
The October 9 report describes playable opening areas tut_01 and tut_02 at 60 FPS, measured as 600 presents per 10 seconds. Without the switch, backgrounds corrupt or loading crashes. One depth-surface failure occurred once and did not recur in two later runs. Other NVIDIA cards and full-game completion are not established by this report.
Read Discussion #1996 and its patch list · Checked October 11; no replies were present.
Astro Bot — gameplay with limited performance
Environment: PPSA21564; Windows 11, Core Ultra 9 285K, RTX 5080. Build: astrobot b58df66c plus PR #637 and a local optimization stack. The October 10 update reports 11.1 FPS at Crash Site in 150-second runs, with ±0.3 FPS noise.
This advances beyond the original October 2 startup failures. The local stack is material: its baseline was 3.2 FPS. The author's higher 12.2 figure changes shadow update frequency, so it is not the comparable result here. Driver version and full-game coverage are not recorded for this update; it does not establish normal-speed play or equivalent results on other hardware.
Read the October 10 update in Discussion #357 · Thread and later replies checked October 11.
Demon’s Souls — historical progress, newer caveats
Historical environment: PPSA01341 v1.07; Nobara 44, Linux 7.2, Ryzen 5 7600X, RTX 5070 Ti, driver 595.91.07, 32 GB RAM. The October 8 df16c4c2 build with PRs and local fixes reached the tutorial at 2.0–2.2 FPS with broken 3D rendering: in game, not playable.
Later evidence: an October 10 correction attributes a startup difference to ANYPS5_ENABLE_SPIRV_TOOLS. With it enabled, 8b393b6 plus PR #861 reached 87 seconds before a SPIR-V validation stop; the earlier 14-second failure was not evidence of a main-branch regression.
October 11 replies use CachyOS, kernel 7.3-rc, Ryzen 5 7600, RTX 3080 10 GB, driver 615.78 and 32 GB RAM. Their mixed main/PR/local build has no recorded combined commit. The Nexus becomes live but initially produces almost no new frames; later 1080p and memory changes improve measurements without establishing playable gameplay.
Read Discussion #1283, including the corrections · Checked through the October 11 follow-up.
Palworld — startup advances beyond missing imports
Environment: PPSA23126 v01.102.999. The original October 10 report uses Windows 10 build 19045, GCC 15.2.0 and main 1138fa96; CPU, GPU and RAM are not recorded. Subsequent patched runs clear successive missing imports but still stop before gameplay.
The October 11 contributor reports Windows 11 build 22621 and a title run on 292ecfb with added exports and other import changes. All imports bind and startup continues beyond ten seconds. An EOS socket-descriptor problem then blocks execution; a local workaround exposes a Discord SDK HTTP wait failure. The proposed explanation for that wait was not confirmed in the title.
This is startup progress, not a playable result. The same thread discusses overlapping exports, so its branches should not be treated as a ready-made setup recipe.
Read Issue #2648 and the October 11 replies · Checked October 11.
Choose your next step
Start with the game's source report, then compare its OS, hardware, title version, build and patches with your setup. Use the download guide for release files and getting started for the conversion workflow. If your result differs, record the first failing stage using the issue-report checklist.
Common compatibility questions
Does “playable” mean the whole game was completed?
No. Keep the label within the tested scope. Opening areas, one level and a completed game provide different coverage; unreported sections remain unknown.
Can I expect the same FPS on similar hardware?
No reliable estimate follows from these records alone. Compare the same scene, resolution, build, driver and settings. Menu FPS and repeated presentation of an old frame can misrepresent gameplay performance.
Why can an older patched build go further than newer main?
A private build may contain missing fixes or different settings. Keep both dated observations and compare their conditions before calling a difference a regression.
Sources and verification
The official table is pinned to commit 05c870cc. Linked GitHub discussions and issues were reviewed through their visible replies on October 11, 2026; they remain editable. This page distinguishes author-reported runs from our source review. No binary or game was executed for this article.