Apotris (PPSA98513)

[VakzOs]

Existing thread

• I searched this category for the title ID and no thread exists for this title on any platform.

Title

Apotris (PPSA98513

Content version tested

01.000.000

AnyPS5 commit

main at cd3d2b4, no PRs or local changes

System tested

Windows 11 Pro build 26200, NVIDIA GeForce RTX 3080 (driver 617.42), i9-14900KF, WinLibs GCC 15.2.0, Release build

Furthest point

In game

What is shown

Menus and gameplay render correctly at a locked 60 fps, 3840×2160 output.
Audio: [music and sound effects OK].
Input: Xbox One Elite 2 controller, detected as Pad: connected game controller: Xbox One Elite 2 Controller, and keyboard work.
The game exits cleanly.
The log has no not implemented, exception or crash lines.
APS5_TRACE_SUMMARY_EVERY=10, 51 intervals over 510 s, digested with tools/trace_digest.py:

value
presents 30,596 (600 per 10 s in every interval after the first)
frame p50 / p90 / p99 16.7 / 16.9 / 16.9–17.0 ms in every interval
max frame 84.1 ms once in the first 10 s, then 17.0–17.3 ms
GPU busy per present 5.4–6.6 ms
largest phase VideoOut.Flip, 14.3–15.5 ms

Current blocker

Nothing stops it, but the title only converts with --trap-sdk-syscall-stubs. Without the flag, the relinker stops with FAIL: Forbidden syscall instruction at code offset 0xca968a: syscall number 191.
Analysis (AI-assisted, not measured): this is the only raw syscall in the executable. It is a ps5-payload-dev SDK stub (mov rax, 0xbf; mov r10, rcx; syscall; ret) for FreeBSD pathconf. Its only caller passes "." and 5 (_PC_PATH_MAX), next to the current_path string, which matches libc++ std::filesystem::current_path(). The trapped stub was never reached in either run.

How it was run

Unwrapped eboot.bin and sce_module/libc.prx with tools/unwrap_self.py.
Converted with relinker --windows --trap-sdk-syscall-stubs eboot.elf apotris.exe.
Laid out libs/ and app0/ as the relinker prints, with the converted libc.prx.guest.prx in app0/sce_module/.
Default game settings, empty shader cache on the first run.
Two runs on 2026-10-23:
about 4 min 20 s with no APS5_* / ANYPS5_* variables;
8 min 30 s with APS5_TRACE_SUMMARY_EVERY=10 only.