[BUG]: Silent hill 2 (PAL) FMV crash & delay (SLES-50382)

[UDP51820]

Describe the Bug

The game crashes at various points, particularly during FVM and screen recording, the first FVM crash begins with Maria's first appearance

Reproduction Steps

Play the game until you meet Maria; the cutscene will either be out of sync, or if you're in rec mode, the game will crash
The game also seems to crash during simple recording; it doesn't seem to be able to keep up
Many other cutscenes crash even without video recording
ex: first fight with ph

Expected Behavior

No response

PCSX2 Revision

2.6.3

Operating System

Linux (64bit) - Specify distro below

If Linux - Specify Distro

Arch but (After testing it on Windows, the issue is also present on Windows 11)

CPU

Intel I5-12600k

GPU

Nvidia RTX 5070

GS Settings

No response

Emulation Settings

No response

GS Window Screenshots

No response

Logs & Dumps

No response

[UDP51820]

After looking at IPU.cpp, I noticed that ipuSoftReset() fires hwIntcIrq(INTC_IPU) immediately (currently marked "required for FightBox").
that unconditiona IRQ might interfere with EE timing during SH2 PAL's cutscene transitions, potentially explaining the FMV crash/desync.
Could this be related?

[refractionpcsx2]

It's not related, no, that change never affected Silent Hill 2, which has had really janky FMV's for a very long time which changes behaviour every time you modify the timing of anything in the emulator.
Pretty confident that IRQ is correct emulation as it triggers a command end IRQ

[UDP51820]

Thanks for the clarification! So if I understand correctly, the core issue is more about overall timing sensitivity rather than a specific IRQ misbehaviour
On that note, would the hardcoded IPU_INT_PROCESS(64) delay in IPUCMD_WRITE() (added for Mana Khemia/Metal Saga) have ever been investigat for SH2 PAL specifically? It feels like the kind of value that could shift FMV behaviour one way or another

[refractionpcsx2]

Are you running this through AI? Please stop if you are, it doesn't know anything about PS2 emulation.

IPU timing is not very good at all, either, but it would require a lot of work and rewriting to make proper timing work. But yes, general emulator timing is a problem with this game, I'm unsure which part it doesn't like, most units have correct timing.

[UDP51820]

Are you running this through AI? Please stop if you are, it doesn't know anything about PS2 emulation.

IPU timing is not very good at all, either, but it would require a lot of work and rewriting to make proper timing work. But yes, general emulator timing is a problem with this game, I'm unsure which part it doesn't like, most units have correct timing.

re, just to clarify i'm European and not a native English speaker, so I use AI only for translation. All remarks (the hwIntcIrq, the IPU_INT_PROCESS(64)) came from me reading IPU.cpp directly, the AI just helped me phrase it in English, Anyway
thanks for the context on IPU timing
but has anyone ever looked into SH2's timing issues specifically, or is it something that's been left aside so far?

[Heitof]

@UDP51820 IPU timing is a can of worms (this is true for every PS2 emulator so far). The issue tracker is dense with it.

[Goatman13]

Seems to be the same issue as Directors Cut (PAL) edition #4064.
From description there it seems to be more CDVD timing issue than IPU itself.
Also, here you can find workaround for it, assuming that it still works in latest pcsx2.
https://forums.pcsx2.net/Thread-Fixing-unplayable-games?pid=619181#pid619181

[refractionpcsx2]

I've been down the cdvd path, I don't think it's that, but also overlooking the EE makes it more reliable too, it's a big voodoo mess

[Mrlinkwii]

probably duplicate of #4064