Hi everyone,
I'm creating my own post about the DXGI error because, although I've seen others mentioning it, I want to share my specific case and findings. After (maybe) 100 crashes since launch — and even back in the Beta — I think I’ve gathered enough consistent data that might help identify what’s happening.
While this issue may vary slightly from player to player, my crashes are 100% reproducible under certain conditions. I believe it’s worth documenting here, hoping it might reach the devs at EA/DICE for further investigation.
Short summary
It seems there’s a potential bug in the FSR2 implementation, where the game passes a null or invalid texture/buffer to FSR2. When FSR2 tries to access that invalid resource, DXGI triggers a crash.
Why I believe this:
1. Crashes usually occur when swapping sights quickly or when there’s a zoom event — e.g., 3x scopes, RPGs, vehicle weapons, etc. Not always, but ~99% of the time.
2. Every crash dump shows the same call stack reference:
bf6!ffxFsr2ResourceIsNull+0x22ad1a0 bf6!ffxFsr2ResourceIsNull+0x22ad834
The crash is always a Device Hung (0x887A0006), suggesting the GPU is processing invalid data coming from FSR2.
So in summary:
Things I’ve tested (none fixed it)
- Disabled FSR as much as possible → tried TAA, no upscaling, no AMD frame generation → same crash.
- Launch options: -dx12 -noasync → no effect.
- Re-tested FSR on/off and various AA modes → crashes happen faster with FSR enabled.
- Cleared shader/cache files between tests.
- Disabled all overlays (Radeon, Discord, Steam, etc.).
- Edited TDR registry values (Delay and DdiDelay) → no stable improvement.
- Underclocked GPU core frequency → no change.
- Many more smaller tests with same results.
Key debug extract (WinDbg !analyze -v)
EXCEPTION_CODE: 0x887A0006 (DXGI_ERROR_DEVICE_HUNG) PROCESS_NAME: bf6.exe MODULE_NAME: bf6 FAULTING_FUNCTION: bf6!ffxFsr2ResourceIsNull STACK (top frames): bf6!ffxFsr2ResourceIsNull+... KERNELBASE!RaiseException+...
Failure indicates FSR2 attempted to access an invalid/null resource during frame reconstruction.
System
GPU: AMD RX Vega 64
Driver: Adrenalin 25.8.1 (clean install via DDU, last driver available for the Vega)
CPU: Ryzen 5 3600X
OS: Windows 10 22H2
API: DirectX 12
If anyone else is facing the same issue, especially with AMD GPUs, please share your crash stack or reproduction pattern — the more data we gather, the better the chances EA/DICE can identify it.
Summary
“This is a consistent, reproducible crash seemingly caused by an invalid FSR2 resource being passed to DirectX during zoom/scope transitions. Requires dev-side fix.”
I am running into the same issue consistently.
My setup:
Basically the bare minimum...
Additionally I was fighting performance problems on large maps (mostly conquest): infantry combat was like walking through honey. Rush and the other modes on smaller maps were running stable though, probably hinting at the CPU being a bottleneck.
I was able to increase performance to somewhat usable levels by doing a combination of the following:
That left me with the "DXGI_ERROR_DEVICE_HUNG" crashes, which can very reliably be reproduced on my setup:
My crash dumps look pretty much identical to the other ones posted here. I doubt that the crash is related to FSR2 though. Addresses like "bf6!ffxFsr2ResourceIsNull+0x22ad1a0" and similar are given relative to the "ffxFsr2ResourceIsNull" symbol, but the offset "0x22ad1a0" is huge. My guess is that they just stripped all / most other symbols from the executable, so "ffxFsr2ResourceIsNull" just happens to be the best WinDbg can get.
You can also just look at the different thread callstacks in WinDbg, which for me mostly look like this:
[0x0] win32u!NtUserGetMessage+0x14
[0x1] user32!GetMessageW+0x22
[0x2] gameoverlayrenderer64!OverlayHookD3D3+0x284f7
[0x3] bf6!ffxFsr2ResourceIsNull+0x274cbfe
[0x4] bf6!ffxFsr2ResourceIsNull+0x274b87c
[0x5] bf6!ffxFsr2ResourceIsNull+0xc4dc3b
[0x6] bf6!ffxFsr2ResourceIsNull+0xc4bf0b
[0x7] bf6!ffxFsr2ResourceIsNull+0xc4dc81
[0x8] bf6!ffxFsr2ResourceIsNull+0x274d0ad
[0x9] bf6!isGlimpseEnabled+0xb976
[0xa] bf6!ffxFsr2ContextGenerateReactiveMask+0xa03f02
[0xb] kernel32!BaseThreadInitThunk+0x17
[0xc] ntdll!RtlUserThreadStart+0x2cI highly doubt that "ffxFsr2ResourceIsNull" is a crazy large function that is called recursively.
I also experimented with the settings in the "user.cfg" file (needs to be created in the same directory as "bf6.exe"), with some success at mitigating the problem at the cost of performance and audio issues.
DISCLAIMER: again I have no clue what this does and there is no official documentation, use at your own risk.
The following seemed to suppress the issue on my machine:
Thread.ProcessorCount 6
Thread.MaxProcessorCount 6
Thread.MinFreeProcessorCount 0
Thread.JobThreadPriority 0
GstRender.Thread.MaxProcessorCount 12Most sources claim to set "Thread.ProcessorCount" and "Thread.MaxProcessorCount" to your number of physical CPU cores and "GstRender.Thread.MaxProcessorCount" to the number of logical ones.
I also played around with "Thread.MinFreeProcessorCount" which seemed to have some effect, though I did not find anything conclusive on what it actually controls. Seemed to kill my game audio when set to the wrong value though.
With above settings the performance dropped considerably on my machine and frame times were a lot more spikey, but in my short test session I was unable to reproduce the crash.
