AnyPS5 game tracker
Hades II key art
Hades II icon

Hades II on AnyPS5

PPSA36082 ยท last report 11 Oct 2026

In game Runs into gameplay on Linux on a testing build (about 24 fps in the hub), but only after the shader cache is warm, and plain main still crashes when loading a level.

  1. Nothing
  2. Boots
  3. Menus
  4. In game
  5. Playable

Reached gameplay (standing in the hub, ~24 fps) on Linux using current main plus several open, unmerged PRs and an environment-variable memory setting; the tester calls it 'playable, with warm shader caches'. Plain main still crashes loading into a level.

Running it and it stopped on an error? Check the error against every report and fix for this game.

Screenshots

No screenshot of this game running has been posted in the project's threads yet. Testers who reach a new screen usually post one in the discussion linked on the right.

Where it stands

Hades II is one of the furthest-along titles in the project. On Linux, a tester (Raghav314, RTX 4090) reports that it boots, loads a game and plays, with no skipped draws, at about 24 fps while standing in the hub (v01.006.000). An older test stack with some local speed work reached 40-45 fps, but that work is not in the current stack. The tester labels it 'playable, with warm shader caches'. That result is on current main plus a list of open pull requests, not on plain main.

There are two big catches. First, it only starts when the shader caches are warm. On a first launch, or the first launch after any code change to the shader translator, the game's loader runs out of its 512 MiB upload memory about 3.2-3.5 seconds in, because the graphics queue is busy compiling pipelines and the loader cannot free memory in time. The cache is only saved when the game exits cleanly, so retrying after a crash does not help. Second, on current main the game crashes when you load into a level: a thread runs out of stack space since a recent change (#1988). A fix is tested locally but no pull request is open yet. The game also needs a large memory-import budget (at least 6784 MiB); the tester sets APS5_HOST_IMPORT_MIB=8192.

Other problems: loud sounds crackle and sound 'boomy' (the audio decoding is correct, but the game's sound buffer is read at the wrong moment; a fix is planned). Correct rendering in gameplay needs several still-open pull requests for multisampled (anti-aliased) graphics, and speed needs another open one. Those all need updating to the latest main before they can merge.

How it got here: between October 5 and 10, mostly one contributor merged fixes for the game's FMOD sound engine, thread waiting, CPU-core queries, display-size detection, a black main menu (a missing shader input), and slow module loading on Linux. The maintainers also merged a new way of building graphics pipelines from reusable parts, which is the first step toward fixing the cold-start problem; the next step is still to come. Only Linux has been reported; no Windows or macOS results appear.

WindowsNo test report yet
LinuxPlays in the hub at ~24 fps on a testing stack with warm shader caches (RTX 4090); cold start and level loading on plain main still crash
macOSNo test report yet

What is stopping it

  • First launch (cold shader cache) crashes about 3.2-3.5 s in: the loader overflows its 512 MiB upload pool while pipelines compile. Fix is being built by compiling shaders earlier (phase B of the pipeline-library work). Thread
  • On current main the game crashes when loading into a level (stack overflow on the loading thread since #1988). A fix is tested locally but no PR is open yet. Thread
  • Crackle and 'boomy' distortion on loud sounds; root cause found (audio buffer read at the wrong moment), fix planned. Thread
  • Correct gameplay rendering needs open multisampling (anti-aliasing) PRs, including a clear-on-fill fix and resolve fix, which need a rebase on current main. Thread
  • Needs a very large memory-import budget (at least 6784 MiB); currently set with an environment variable because the PR for it no longer merges cleanly. Thread

How it got here

  1. Fixes for the PS5 audio-decoder calls that Hades II's FMOD sound engine relies on are merged. Source
  2. A missing CPU-core query is added; the PR says Hades II now runs into gameplay on the author's local branch. Source
  3. A shader input the game never sets is now read as zero, which fixes a main menu that stayed black. Source
  4. Linux symbol lookup is made much faster, one of the delays that made the game's loader run out of upload memory at startup. Source
  5. Bring-up thread opened: boots, loads a game and plays on a testing stack, about 24 fps in the hub with warm caches. Source
  6. Memory-mapping and import fixes for GPU-visible memory are merged, which the game's 6 GiB heap needs to boot. Source
  7. Graphics pipelines are now linked from cached per-stage parts, cutting compile stalls (first step toward fixing the cold-start crash). Source
  8. A speed fix is opened: tracking of GPU-written pages took 7.4% of the graphics worker's time while standing in the hub. Source

Tested hardware

PCResult
Linux (CachyOS, kernel 7.2.9) / Ryzen 7 7800X3D, 32 GB / RTX 4090 24 GB, NVIDIA 615.71.09About 24 fps standing in the hub on current main plus open PRs, warm caches (MangoHud); an older stack with local speed work gave 40-45 fps.Report
Linux (CachyOS) / Ryzen 7 7800X3D / RTX 4090Used to test the CPU-core query fix; the PR says the game runs into gameplay on a local branch.Report