Anger Foot on AnyPS5

Verified

• Windows 11 / WinLibs GCC 15.2.0 / RTX 5090 (NVIDIA 617.42) / AMD Ryzen 9 9950X3D (discussion #7379, 18 Oct 2026): Anger Foot PPSA29043 content 01.000.000 on unmodified main f1af6c591, with no pull requests applied. Windows in game — intro at about 60 FPS (594 presents in a 10 s window at the start, about 59 FPS), then the main menu at about 6.8 FPS, the first cutscene at 4.4 FPS, and walking in the first playable part at 3 to 9 FPS. The log's own count agrees: 274 presents in one window, then 24, 61, 67 and 69 in the next four, and 84 to 89 in the steady part that follows. The player walks around. No fatal error, exception or NotImplemented line appears in the log. The run lasted about 3 minutes. Audio and controller behaviour were not checked. f1af6c591 predates v0.1.3 and is the merge of #7325, which is in v0.1.3. Single community self-report. In game ≠ Playable.
• Same Windows host and thread: Unity (IL2CPP) title. The dump's decrypted folder already held ELF copies of eboot.bin and the modules (libc.prx, libSceNpCppWebApi.prx, Media/Modules/Il2CppUserAssemblies.prx and PS5Util.prx, Media/Plugins/PSN.prx), so no unwrap step was needed. Relinked with relinker --windows --windows-diagnostics --trace-imports --registry --sce-module-path decrypted --module-dir Media/Modules --module-dir Media/Plugins decrypted/eboot.bin app.exe. Those flags are relinker flags, not a game patch. app0 holds the dump's sce_sys as a junction, a copy of eboot.bin, and the Media folders as junctions and hard links. The dump has PlayGo metadata, so no PlayGo setting was used. Run with ANYPS5_GPU=5090, VK_LOADER_DRIVERS_DISABLE=*dzn*, APS5_LOG_TIMESTAMPS=1 and APS5_PROFILE_DRAW=1. No extra title patch file is named as required.

Blocked

• Current blocker named in discussion #7379: the frame rate after the intro. APS5_PROFILE_DRAW=1 in the steady slow part shows about 88 presents per 10 s, GPU busy 29 ms per frame and idle gaps of 81 ms per frame, so the GPU is waiting for the emulator side; about 440,000 DRAW_INDEX_2 packets per 10 s, roughly 5,000 draws per frame, and the draw path costs 14 us per draw on average (13 us for the 440,000 template hits; 1,400 builds at 337 us each, so it is not shader compilation). The 10 s total of about 6.3 s splits across vertex 1.0 s, prepare 1.0 s, lookup 1.7 s and record 1.6 s, with pipeline 0.3 s. The intro window had 2,136 DRAW_INDEX_2 packets per 10 s at 594 presents. Most draws continue an existing render pass (432,603 of 444,658) and only 612 of them hit a draw recipe.
• A follow-up second run in discussion #7379, with a temporary counter (not part of any PR), counted about 650,000 draws in the slow part: 613,272 (94%) have a read-only guest buffer binding whose address differs from the cached template (MovedReadOnlyBuffers); 6,692 were not cacheable; 1,852 were not reusable. The title issues each draw with its own constant buffer address. Those draws take the 14 us template-hit path instead of the 6 us recipe-hit path. The player noted the game speeds up when no smoke is on screen, which matches the draw count being the cost. Making a recipe accept moved bindings was not tried.
• Not checked in the cited report: audio, controller behaviour, Linux, macOS, save or save reload, AMD or Intel GPU, how far play went past the first playable part, or completion. In game on that cited main is not a Playable claim.

Unknown

• Official-table membership. This title is a community report (official: false), not a tested-compatibility table row.
• Whether the current official release zip matches this report. The cited play used unmodified main f1af6c591, which predates v0.1.3 and is the merge of #7325 (already in v0.1.3), not a release zip.
• Linux. No cited Linux session of this title in discussion #7379.
• macOS. No cited macOS session of this title in discussion #7379.
• Audio. Discussion #7379 says audio was not checked.
• Controller behaviour. Discussion #7379 says controller behaviour was not checked.
• Save or save reload. Discussion #7379 does not mention a save slot or a save round trip.
• AMD or Intel GPU in-game reports. The cited Windows host is RTX 5090 only.
• How far play went past the first playable part, or completion. The report names walking in the first playable part for about 3 minutes.
• Whether making a recipe accept moved bindings would change the frame rate. Discussion #7379 says that was not tried.

Saves

Status: Unknown

Unknown. Discussion #7379 does not mention save or save reload.

Game patch

Status: Known

Known. The cited report names Anger Foot PPSA29043 content 01.000.000. Relinked with relinker --windows --windows-diagnostics --trace-imports --registry --sce-module-path decrypted --module-dir Media/Modules --module-dir Media/Plugins decrypted/eboot.bin app.exe. Those flags are relinker flags, not a game patch. The dump's decrypted folder already held ELF copies, so no unwrap step was needed. No extra title patch file is named as required. Furthest cited in-game used unmodified main f1af6c591 (the merge of #7325) with no pull requests applied. #7325 is in v0.1.3. The cited run used ANYPS5_GPU=5090, VK_LOADER_DRIVERS_DISABLE=*dzn*, APS5_LOG_TIMESTAMPS=1 and APS5_PROFILE_DRAW=1.

Network

Status: Unknown

Unknown. Media/Plugins/PSN.prx is converted. No cited report tests online or networked play.