Just a bit more information since its been a few years. The below does not include video codec issues which still occur. Only referencing actual game play.

NVIDIA GPU: no issue during gameplay (Played over an hour and a half)
AMD GPU: playable for roughly 10 to 30 minutes.. eventually leads to invisible units/buildings, Low texture terrain then crash to desktop.

Tested two difference systems, both on Archlinux, both AMD CPU's, but one has Nvidia GTX 1660, the other has an AMD RX 5600 XT.

As requested. I didn't care to let the game fully CTD, but I did let it get to the point it was unplayable. This didn't affect my girlfriend's computer which has a GTX 1660, and I was able to remote stream it just fine.. My system (AMD RX 5600 XT) is unplayable.

Gist in the original post was updated.

steam-757210.log

Same issue in 2024. Codec issues seem to have gotten worse on Experimental, though Proton-GE seems to have dealt with that somehow. Both still have either have the issue in the above screenshots, or outright black screen and CTD out of nowhere, taking anywhere from 10 to 30 minutes to do so. These issues are not present on Windows.

I have an RX 5600 XT as well, currently running Mesa 23.3.6. Running kernel 6.7.4-200 under Fedora.

Update: I have been able to confirm this issue on a laptop with an AMD Ryzen 7 5700U (iGPU). It does not seem to be specific to the RX 5600 XT.

2025, still a massive issue. iGPU in a Ryzen 9 7900X3D, RX 5600 XT, RX 6700 XT, and RX 7900 XTX. All suffer the same issue.

Intel Arc A770 appears to be working fine.

I didn't see any elements disappear in free play nor weird graphical corruptions on my rx6600 with mesa 24.3.3. This was leaving the game idling the game for about an hour twice, just letting the place burn down by itself. The one issue I noticed is that when you start the game focuses in some sort of warehouse looking building that has 4 lights, said lights turn off based on the camera position, this is something I can see from the start of the match. Perhaps specific actions within the game are needed for the issue to occur (or make it happen faster otherwise)

I believe it has something to do with some sort of limitation. I'd been inferring it has something to do with VRAM, and I often noticed that the game tended to start exhibiting the above behaviors once I'd gone through a number of events/callouts and spawned a number of units/vehicles. I haven't bothered with the game on Linux for a few months now, so I'll do my own testing on Mesa 24.3.3 and latest Proton and see what it hands me.

I didn't see any elements disappear in free play nor weird graphical corruptions on my rx6600 with mesa 24.3.3. This was leaving the game idling the game for about an hour twice, just letting the place burn down by itself. The one issue I noticed is that when you start the game focuses in some sort of warehouse looking building that has 4 lights, said lights turn off based on the camera position, this is something I can see from the start of the match. Perhaps specific actions within the game are needed for the issue to occur (or make it happen faster otherwise)

The original images (modless gameplay) I posted showing the issue with ground textures/models is 100% of what I see after a few minutes in endless freemode with or without mods with actual gameplay (not idle).

Over the past 5 years (since issue creation) and prior to me doing an RMA for my 7900 XTX recently, I've gone through multiple GPU's all AMD and all affected. I would honestly be surprised if your RX 6600 doesn't have this issue.

Hello @Sora012, it may be worthwhile to also mention the misrendering to your video driver vendor if it hasn't been already.

I'll do that sometime next week when I get a replacement card and able to provide additional information when I create the issue.

I believe it has something to do with some sort of limitation. I'd been inferring it has something to do with VRAM, and I often noticed that the game tended to start exhibiting the above behaviors once I'd gone through a number of events/callouts and spawned a number of units/vehicles. I haven't bothered with the game on Linux for a few months now, so I'll do my own testing on Mesa 24.3.3 and latest Proton and see what it hands me.

I noticed that higher quality mods usually trigger the issue much quicker. (LA, Miami) Its very likely related to graphics/memory in some capacity.

I own the following game 911: First Responders which is no longer available but is EM4 without the deluxe content. When I receive my replacement card I'll try and see if it suffers the same fate as EM4. EM4 contains a few additional libraries shipped with it in comparison to 911:FR (shown below)

• dbghelp.dll
• Foundation.dll
• HAPI.win32.dll
• LC_API.dll
• optgraph.dll
• XCPTHLR.dll

Hi I'm a mesa developer I was never able to reproduce these issues when I tried a couple of years ago. Are you still having the the problems with the most recent release of mesa? If so maybe you could try capturing an apitrace of the problem?

Hi, I do still face these issues with latest Mesa. I'll see if I can capture a trace.

Hi, I do still face these issues with latest Mesa. I'll see if I can capture a trace.

Great thanks! Please note I believe this is a 32bit game so you might need to build 32bit apitrace from source. The 64bit package apitrace that comes with most distros wont work. Happy to give more specific help if needed.

I just saw the updates, I'm using Proton Experimental (Beta: Bleeding Edge) + Mesa 25.0.5-1 on Archlinux, at some point after I got my RMA for the 7900 XTX I did end up retesting the game to have noticed game stopped crashing. I thought perhaps all of my cards were faulty, despite only the 7900 XTX needing an RMA due to actual hard crashing. With @tarceri saying they couldn't reproduce it in the past, and it seemingly have vanished now. I assume its a DXVK issue that had a regression and then fixed again. I am having artifacting near water with large black boxes stretching across the area (on Miami mod) but that seems to be the only issue and is otherwise stable now. I was able to play previously on Habor city mod for ~2 hours in multiplayer without issues after I noticed it stopped crashing.

@m2jest1c Are you running Proton Experimental or its beta? We could try to downgrade to proton 9, or 8 to test against and see if it still has issues there because I run everything (including this game) against experimental bleeding edge. I think in a new prefix downgraded (if you run experimental) that if it crashes again it would indicate a DXVK issue and not directly Mesa?

I tested against experimental bleeding edge again for ~40 minutes or so on Miami mod without any CTD. I'm planning on retesting against Proton 8 and 9 later to check if they have the CTD issue still to determine DXVK vs Mesa

As for the black box issues I was mentioning. It might be related to camera zoom as I increase maximum scroll out as seen below.

If you mean the stuff going on at the top and the top right corner, that's because of zoom, yes. You'll see the same on Windows.

I tested against experimental bleeding edge again for ~40 minutes or so on Miami mod without any CTD. I'm planning on retesting against Proton 8 and 9 later to check if they have the CTD issue still to determine DXVK vs Mesa

Good to hear it might be fixed! Also I don't think this should matter, I believe the game uses OpenGL for graphics not Directx

Unfortunately, I can report that this problem still exists. Running on a Steam Deck, after ~15-20 minutes of gameplay, all models unload, leaving only the map and emergency lights floating in the air. Notably, sometimes, only some models will unload while others will stay half rendered until the camera moves away from them.

Running on GE-Proton10-26.

I've uploaded the trace here; https://drive.google.com/file/d/1LSatUeWpAe-kLHOz3joWjiq_SEbzxwT2/view?usp=sharing

Seems that trace link no longer works so I can't look at what the game is doing but I came across this today MR today which made me wonder if the game is running into similar memory issues that are improved by this fix: https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28688 Is anyone able to built that MR and test if it helps?