AnyPS5.io
English
EnglishالعربيةPortuguêsDeutschFrançaisEspañol

Games · Compatibility · Home

Sword of the Sea on AnyPS5

Windows is in game — Sword of the Sea (the PS5 title PPSA25344, content 01.510.000), the opening cutscene after New Game at about 4 FPS on the first visit — on unmodified main d7a455a90 (Windows 11 Pro / AMD Ryzen 9 7900 / GeForce RTX 4070), with a cold shader cache killing the first New Game loads; Linux and macOS were not checked. Latest release v0.1.3 is 3fd1cc393b (20 Oct). Cited commit d7a455a90 (18 Oct) is in v0.1.3 (3fd1cc393b, 20 Oct). The cited furthest in-game used plain main, the parent of the #6112 merge, with no pull requests on top. Relinked with relinker --windows <decrypted>/eboot.bin app.exe. Those flags are relinker flags, not a game patch. The title's bundled prx/ and sce_module/ modules are converted alongside. The runtime layout puts the original eboot.bin, sce_sys/, engine/ and surfer/ in app0/. APS5_PROFILE_DRAW=1 and APS5_PROFILE_GPU=1 were set for the title-screen samples; APS5_THREAD_PRIORITY=high was set for the later runs. Those are host environment variables, not a game patch. Console output was redirected to a file: with logs going to the Windows console, the title screen was noticeably slower. A copy of uecommandline.txt had -dpcvars=g.TimeoutForBlockOnRenderFence=900000 appended for the third New Game attempt. That is a game command-line override, not a title patch file. The reporter cannot tell whether the shipping build honoured it or whether the warmer cache was enough. No extra title patch file is named as required. The same thread compared that main with official release v0.1.2 on the title screen. No cited run checked the current official release zip v0.1.3. Whether the current official release zip matches this report is unknown, so an official zip may stop earlier. This title is not in the official table. Build source is main — the cited furthest in-game used unmodified main d7a455a90, not a release zip. Title screen and menu render, audio plays, and input works (XInput pad and keyboard). New Game loads into the opening cutscene. On that day's title-screen samples (warm caches, [present] line per 10 s) v0.1.2 presented 36 to 37 times per 10 s (about 3.7 FPS) with GPU busy 111 to 115 ms and idle gaps 159 to 168 ms per frame; main d7a455a90 presented 128 to 129 times per 10 s (about 12.9 FPS) with GPU busy 55.2 ms and idle gaps 22.7 to 23.0 ms. v0.1.2 logs Windows direct host imports use comparisons, the #5692 path. d7a455a90 logs direct host imports of private memory are settled and stay watched and [guestmem] compare 151 calls 1 MiB 0.0 s, so #6060 recovers it here as well. #6060 merged (15 Oct) and is in v0.1.3. On d7a455a90 the title screen is GPU-bound: about 70 % of the frame is GPU busy time. The guest presents 3840x2160. The opening cutscene runs at about 4 FPS on the first visit (overlay reading), while shaders are still being compiled: 48, 14 and 11 shader-cache misses per 10 s. The reporter has no warm-cache figure for gameplay yet. Those FPS values are a report, not a benchmark. The form furthest point is In game (the opening cutscene after New Game). In game on that cited main is not a Playable claim. A single self-report is not an official-table row. Current blocker: a cold shader cache kills the first New Game loads. On the first two New Game attempts with a cold shader cache the title stopped with GameThread timed out waiting for RenderThread after 120.00 secs, then FATAL: unhandled exception 0xc0000005. During that load the shader cache took about 1,200 misses in 30 s (about 1.3 GB written), with two CPU threads at 100 % and the GPU at about 1 %. Each attempt keeps what it compiled, and the third attempt reached the cutscene. That matches the cold-load numbers in discussion #5436 (more than 700 s, mostly compute pipeline creation). Also seen, without visible effect: [gpu] queue 0x20 WAIT_REG_MEM ... timed out after 1000ms. Historical (18 Oct 2026): on then-current main 400baa252 the title aborted at start-up with Thread destructor callback is already registered on Windows too (#7636), so the reporter built the parent of the #6112 merge. #6112 merged (18 Oct, 18:33 UTC) and is in v0.1.3. #7646 reverted #6112 the same day (18 Oct, 21:10 UTC) and is in v0.1.3. The cited furthest in-game used d7a455a90 to avoid that abort. That abort was not re-checked on this title after the revert.

In gameCommunity reportWindows

PPSA25344

AnyPS5.io branded placeholder cover for Sword of the Sea. Not official box art.
Branded AnyPS5.io placeholder. Not official box art.

Snapshot

StageIn game
Platforms in this snapshotWindows
Build sourcemain · d7a455a90
Last verified in snapshot18 Oct 2026
Snapshot assembled14 Oct 2026 · checked 11 Oct 2026

Allowed values: release, main, fork, patched, unknown. Official table membership is not a release-zip claim.

Furthest public stage shown, including testing branches. Not an official playable verdict.

Page content updated: 23 Oct 2026. This date is when the page text changed. It is not a new hardware test.

Official zip may stop earlier

Playable here can come from later main or extra patches. The official release zip may stop before that furthest stage.

Verified, blocked, unknown

Verified

  • Windows 11 Pro (build 26200) / WinLibs GCC 15.2.0 (r7) / AMD Ryzen 9 7900 / 32 GB / NVIDIA GeForce RTX 4070 12 GB (NVIDIA 616.92) (discussion #7657, 18 Oct 2026): Sword of the Sea PPSA25344 content 01.510.000 on unmodified main d7a455a90, the parent of the #6112 merge, with no pull requests on top. Windows in game — title screen and menu render, audio plays, and input works (XInput pad and keyboard). New Game loads into the opening cutscene. Title-screen samples (warm caches): about 12.9 FPS on that main versus about 3.7 FPS on official v0.1.2. The opening cutscene runs at about 4 FPS on the first visit while shaders compile. d7a455a90 is in v0.1.3. Single community self-report. In game ≠ Playable.
  • Same Windows host and thread: Relinked with relinker --windows <decrypted>/eboot.bin app.exe. Those flags are relinker flags, not a game patch. The title's bundled prx/ and sce_module/ modules are converted alongside. The runtime layout puts the original eboot.bin, sce_sys/, engine/ and surfer/ in app0/. APS5_PROFILE_DRAW=1, APS5_PROFILE_GPU=1 and APS5_THREAD_PRIORITY=high are host environment variables, not a game patch. No extra title patch file is named as required. On d7a455a90 the title screen is GPU-bound (about 70 % of the frame is GPU busy time). The guest presents 3840x2160. #6060 merged (15 Oct) and is in v0.1.3.

Blocked

  • Current blocker named in discussion #7657: a cold shader cache kills the first New Game loads. On the first two New Game attempts with a cold shader cache the title stopped with GameThread timed out waiting for RenderThread after 120.00 secs, then FATAL: unhandled exception 0xc0000005. During that load the shader cache took about 1,200 misses in 30 s (about 1.3 GB written). Each attempt keeps what it compiled, and the third attempt reached the cutscene. The opening cutscene runs at about 4 FPS on the first visit. The reporter has no warm-cache figure for gameplay yet.
  • The form furthest point is In game (the opening cutscene after New Game). In game on that cited main is not a Playable claim. Historical (18 Oct 2026): then-current main 400baa252 aborted at start-up with Thread destructor callback is already registered (#7636). #7646 reverted #6112 the same day and is in v0.1.3. That abort was not re-checked on this title after the revert.

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 d7a455a90, not a release zip. The same thread compared the title screen with official v0.1.2 only. No cited run checked official v0.1.3.
  • Linux. No cited Linux session of this title in discussion #7657.
  • macOS. No cited macOS session of this title in discussion #7657.
  • Save or load in gameplay. Discussion #7657 does not name a save write or reload.
  • Network or online play. No cited report checks it.
  • Warm-cache gameplay FPS. The reporter has no warm-cache figure for gameplay yet. The about 4 FPS cutscene figure is a first-visit overlay reading while shaders compile.
  • How far play went past the opening cutscene, or completion. The cited session names the title screen, the menu and the opening cutscene after New Game.
  • AMD or Intel GPU in-game reports. The cited host is GeForce RTX 4070 only.
  • Whether the shipping build honoured -dpcvars=g.TimeoutForBlockOnRenderFence=900000 on the third New Game attempt, or whether the warmer cache was enough.
  • Whether the Historical 18 Oct 2026 Thread destructor abort still happens on this title after #7646. Discussion #7657 did not re-check it here.

Saves, patch, and network

Saves

Status: Unknown

Unknown. Discussion #7657 does not name a save write or reload in gameplay.

Game patch

Status: Known

Known. The cited report names Sword of the Sea PPSA25344 content 01.510.000. Relinked with relinker --windows <decrypted>/eboot.bin app.exe. Those flags are relinker flags, not a game patch. The title's bundled prx/ and sce_module/ modules are converted alongside. No extra title patch file is named as required. A copy of uecommandline.txt had -dpcvars=g.TimeoutForBlockOnRenderFence=900000 appended for the third New Game attempt; that is a game command-line override, not a title patch file. Furthest cited in-game used unmodified main d7a455a90 (the parent of the #6112 merge) with no pull requests on top. #6060 and the later #7646 revert of #6112 are in v0.1.3.

Network

Status: Unknown

Unknown. No cited report checks online or networked play.

Quoted public error strings

Verbatim from the linked report. One run, one build — not a promise that you will see the same line.

Windows

Windows 11 Pro (build 26200) / WinLibs GCC 15.2.0 (r7) / AMD Ryzen 9 7900 / 32 GB / NVIDIA GeForce RTX 4070 12 GB (NVIDIA 616.92) (discussion #7657, 18 Oct 2026): Sword of the Sea PPSA25344 content 01.510.000 on unmodified main d7a455a90, the parent of the #6112 merge, with no pull requests on top. Windows in game — title screen and menu render, audio plays, and input works (XInput pad and keyboard). New Game loads into the opening cutscene. Title-screen samples (warm caches): about 12.9 FPS on that main (128 to 129 presents per 10 s; GPU busy 55.2 ms; idle gaps 22.7 to 23.0 ms) versus about 3.7 FPS on official v0.1.2 (36 to 37 presents per 10 s; GPU busy 111 to 115 ms; idle gaps 159 to 168 ms). The opening cutscene runs at about 4 FPS on the first visit while shaders compile. Relinked with relinker --windows <decrypted>/eboot.bin app.exe. Those flags are relinker flags, not a game patch. The title's bundled prx/ and sce_module/ modules are converted alongside. APS5_PROFILE_DRAW=1, APS5_PROFILE_GPU=1 and APS5_THREAD_PRIORITY=high are host environment variables, not a game patch. No extra title patch file is named as required. Current blocker: a cold shader cache kills the first New Game loads (GameThread timed out waiting for RenderThread after 120.00 secs, then FATAL: unhandled exception 0xc0000005); the third attempt reached the cutscene. d7a455a90 (18 Oct, 18:33 UTC) is in v0.1.3 (3fd1cc393b, 20 Oct, 07:07 UTC). Historical (18 Oct 2026): then-current main 400baa252 aborted at start-up with Thread destructor callback is already registered (#7636). #7646 reverted #6112 the same day and is in v0.1.3. The cited furthest in-game used plain main. No cited run checked the official release zip v0.1.3. Linux and macOS were not checked. Single community self-report. In game ≠ Playable.

Linux

No report in the cited thread. Linux was not checked.

Sources

Live threads beat this snapshot: Game bring-up discussions.

Need the files? Download v0.1.3 from GitHub, then relink. This page is not a store and does not host games.