Can you please try this with just plain Proton Experimental or something? It's rather dauting when everyone uses different unofficial Proton builds that I then have to manually set up and fish the d3d11.dll out of.

I'm also not sure how to really test this game because 237 GB is just complete insanity, would take literal days to download and I kind of need the disk space for other things too.

That said, there's a) roughly a million d3dcompiler errors in the log and b) a known issue with the game crashing because it simply requires extreme amounts of memory (see [#5799](https://github.com/doitsujin/dxvk/issues/5799)).

Reproduced on plain Proton Experimental with the same launch options, logs attached below.

If you need anything extra i'll gladly provide, thanks for taking the time to look at this.

steam-15499925930595319808.zip

dcs.log-20260823-205815.zip

Let the thing download overnight, how do I reproduce the bug in game?

You need to do some explosion effects in game, i tried to spectate in multiplayer but the crash did not happen, only when i did the explosions

Ideally I'll need a way to repro this locally without having to do much/anything on my own, don't really have the time or patience to learn the controls here. Tried to do the bombing tutorial but it always inevitably ends with me crashing the plane before anything happens.

You don't need to reproduce it yourself. Just download the track file below and place it in:
(your Steam directory for the game)/pfx/drive_c/users/steamuser/Saved Games/DCS/Tracks/

Then open the game, go to Replay, and the track should show up so you can play it back.

One important note: the crash is not guaranteed. The track I recorded crashed before the actual mission ended, while my own crash during gameplay happened after the track file had already ended. It also could not be reproduced on the first attempt. Still, this track crashes almost every time for me.

track.zip

I tried the mission above playing without an account that has the F16 bought, it says its corrupted and the track file is not doing what it is supposed to do, i'll try to replicate it and send another track file

I couldnt reliably crash the game using the tracks, but opening the mission and letting the bot do its thing is almost guaranted to crash all the times.

Directory for custom missions
(your steam directory)/pfx/drive_c/users/steamuser/Saved Games/DCS/Missions/

PS: Use CTRL + Z to speed up and SHIFT + Z to get back to normal speed

I recorded a video with the crash (like in the video, sometimes it crashes with the first bomb)

https://www.youtube.com/watch?v=nwLqvncDGcM

crash3.zip

track2.zip

mission.zip

Unfortunately this is a use-after-free bug in the game, it binds a resource view that is already destroyed and we end up crashing when trying to apply that.

Gonna see if there's a sane way to work around this problem, but it might just not really be viable.

Ok, thank you so much

Quick update. I put together a patch that seems to fix this on my end. It adds an opt-in check that validates a resource view is still alive before binding it, treating an already-destroyed view like a null bind instead of dereferencing it. It's gated behind a new d3d11.viewUAFGuard option that defaults to off and is only auto-enabled for DCS.exe. Since it's a general opt-in guard and not DCS-specific under the hood, it could also be enabled for other games if a similar use-after-free pattern turns up down the line. Ran multiplayer gameplay and several missions that reliably crashed before, including explosions, and got no crashes so far, with no noticeable performance impact in-game.

Full disclosure: I did not write this code myself. It was written entirely by an AI (Claude Code), based on your diagnosis above. I'm not a C++/graphics developer, so I can't personally vouch for the implementation. If you have any interest in it, i can send it so you can review it, since I have no way to know if it could break something else or introduce a regression elsewhere.

Happy to test further or share the diff/branch if useful.

dcs.log

Adding another data point on this: same crash signature (C0000005 ACCESS_VIOLATION
in d3d11.dll → dx11backend → effects2 → visualizer → edcore) on AMD/RADV, not just
NVIDIA, so this isn't vendor-specific.

• OS: Bazzite Linux (Fedora Atomic-based)
• Proton version: Proton Experimental
• GPU: AMD Radeon RX 7800 XT (RADV NAVI32), 16GB VRAM
• CPU: Ryzen 5 7500F
• DCS: 2.9.29.27468 (build 559)
• Aircraft: FA-18C Hornet

Crash happened ~8.5 minutes into a flight, no scripted mission events I'm aware of.
The dcs.log shows two EFFECTS2 warnings shortly before the crash:

19:58:56 WARNING EFFECTS2: OParticleSystem: preset 'missile' not found in
'volumetricPointLight.lua', using default
20:01:24 WARNING EFFECTS2: OParticleSystem: preset 'smoke' not found in
'speedSmoke.lua', using default

...then the crash at 20:02:12, going through the exact same effects2 → visualizer
path already reported here. Happy to attach my full dcs.log if useful.

Also happens when G smoke on the cockpit's side appears

Adding another data point on this: same crash signature (C0000005 ACCESS_VIOLATION in d3d11.dll → dx11backend → effects2 → visualizer → edcore) on AMD/RADV, not just NVIDIA, so this isn't vendor-specific.

* OS: Bazzite Linux (Fedora Atomic-based)

* Proton version: Proton Experimental

* GPU: AMD Radeon RX 7800 XT (RADV NAVI32), 16GB VRAM

* CPU: Ryzen 5 7500F

* DCS: 2.9.29.27468 (build 559)

* Aircraft: FA-18C Hornet

Crash happened ~8.5 minutes into a flight, no scripted mission events I'm aware of. The dcs.log shows two EFFECTS2 warnings shortly before the crash:

19:58:56 WARNING EFFECTS2: OParticleSystem: preset 'missile' not found in 'volumetricPointLight.lua', using default 20:01:24 WARNING EFFECTS2: OParticleSystem: preset 'smoke' not found in 'speedSmoke.lua', using default

...then the crash at 20:02:12, going through the exact same effects2 → visualizer path already reported here. Happy to attach my full dcs.log if useful.

Also happens when G smoke on the cockpit's side appears

Im currently playing the game for about a month with no issues, see below

https://github.com/doitsujin/dxvk/pull/5862

https://github.com/ggabriel537/dxvk

I have the exact same issue. I've just come back to DCS after many years away but this time I have a Linux box.

Mint 22.3 / Kernel 7.0.0-31
Proton Experimental
CPU: Ryzen 9 9950X3D, 32GB RAM
GPU: Radeon 9070 XT 16GB VRAM
DCS 2.9.29.27468

Aircraft: A-10C

Crash happens within only 2-3 minutes of free flight. Which is to say there are no threats so there is nothing exploding and no scripted events to occur.

Graphics were set to Medium to see if that made any difference.

This is actually the first game I have found that does not play reliably under Proton!

dcs.log-20260923-090531.zip

I have the same issue. Game starts and after few minutes crashes. The fix that @ggabriel537 provided helps but only if I set DXVK_HUD=1 in launch arguments. If I set DXVK_HUD=0 it still crashes.

I tried switching to GE-Proton Latest and no crashes when I set DXVK_HUD=fps. If I set DXVK_HUD=0 then it crashes after few minutes.

Same for me

• Fedora 44 workstation
• RTX 4060
• DCS 2.9.30.28718
• F-16 Viper

The DXVK_HUD=fps doesn't help

The crash varies, sometimes its 3 minutes into flight and sometimes its 14+

I tried switching to GE-Proton Latest and no crashes when I set DXVK_HUD=fps. If I set DXVK_HUD=0 then it crashes after few minutes.

Do I just have to type DXVK_HUD=fps into the command text bar in Steam to try?

I wonder what's the connection with the smoke animation..

@parasite6 . @krickovicd @SimonGreen-NZ @MByte27

My custom DXVK build has been running without any issues so far. I recently played a mission for about 1.5 hours with no crashes or other problems

To use it, you’ll need to compile DXVK with the patches ( see https://github.com/ggabriel537/dxvk ) yourself using the package-release.sh build script, then create a custom Proton installation and move the compiled DLLs into the new proton installation, then put it inside:

~/.steam/steam/compatibilitytools.d/

This allows Steam to detect it and list it as a custom Proton version under the compatibility tools.

These are my current launch options:

WINE_CPU_TOPOLOGY=12:0,1,2,3,4,5,6,7,8,9,10,11 NVPRESENT_ENABLE_SMOOTH_MOTION=1 DXVK_NVAPI_DRS_NGX_DLSS_SR_OVERRIDE_RENDER_PRESET_SELECTION=render_preset_latest WINE_SIMULATE_WRITECOPY=1 MANGOHUD=1 STEAM_COMPAT_MOUNTS=/mnt/nvme1tb,/opt/opentrack %command%

If you don't want to compile DXVK yourself, I've uploaded my pre-compiled Proton build here: https://drive.google.com/file/d/1umX-K1Z6N0GUjdOnz-x5Tqm7w7ASXdcq/view?usp=sharing -> unzip it and move it to ~/.steam/steam/compatibilitytools.d/ and restart steam, should work out of the box

If you guys run into any issues or have any questions, I’ll gladly try to help you with any step of the process

@parasite6 . @krickovicd @SimonGreen-NZ @MByte27

My custom DXVK build has been running without any issues so far. I recently played a mission for about 1.5 hours with no crashes or other problems

To use it, you’ll need to compile DXVK with the patches ( see ggabriel537/dxvk ) yourself using the package-release.sh build script, then create a custom Proton installation and move the compiled DLLs into the new proton installation, then put it inside:

~/.steam/steam/compatibilitytools.d/

This allows Steam to detect it and list it as a custom Proton version under the compatibility tools.

These are my current launch options:

WINE_CPU_TOPOLOGY=12:0,1,2,3,4,5,6,7,8,9,10,11 NVPRESENT_ENABLE_SMOOTH_MOTION=1 DXVK_NVAPI_DRS_NGX_DLSS_SR_OVERRIDE_RENDER_PRESET_SELECTION=render_preset_latest WINE_SIMULATE_WRITECOPY=1 MANGOHUD=1 STEAM_COMPAT_MOUNTS=/mnt/nvme1tb,/opt/opentrack %command%

If you don't want to compile DXVK yourself, I've uploaded my pre-compiled Proton build here: drive.google.com/file/d/1umX-K1Z6N0GUjdOnz-x5Tqm7w7ASXdcq/view?usp=sharing -> unzip it and move it to ~/.steam/steam/compatibilitytools.d/ and restart steam, should work out of the box

If you guys run into any issues or have any questions, I’ll gladly try to help you with any step of the process

Thanks, I actually did verify and then compile it. DCS Seems to be fine now