Hey everyone,
at the end of this year I switched to my first ever AMD setup. Since then however, I noticed that a lot of Games I used to play just keep on crashing.
For Example: CoD MW2 just crashes on boot.
Error Message: dxgi_error_device_hung 0x887a0006 or 0x887a0005
Civilization VI crashes when the loading of creating a new game is done.
When i loaded my older rounds it worked fine for some reason and I could then create a new round... really bizarre.
Error Message: Tells me it's because of a DirectX driver missing. Checking with dxdiag I have everything installed.
The Hunter: Call of the Wild crashes once I reach the main Menu.
Unreal Engine 5.0.3 and 5.1 crash once I finish creating a new Project.
However loading in an existing Project seems to work sometimes.
Error Message: DXGI_ERROR_DEVICE_REMOVED
Switching to Unreal 4 Projects. Everything works as normal.
My PC components are as follows.
ASUS ROG Strix 550-F
1000 Watt Corsair HX 80+ Platinum
AMD Ryzen 7 5800X3D
G.SKILL DDR4 3600 2x16 GB 16-16-16-36
Samsung 980 PRO 2TB PCIe 4.0 NVMe
ASUS TUF GAMING RX 6900 XT OC
I updated my BIOS to the newest version yesterday
and have the D.O.C.P profile enabled for my RAM. (also tried disabling it)
OS: Windows 11 Home ( I also tried Windows 10, same results )
AMD Driver: 22.5.1 ( I also tried the newest driver, same results )
I added Tdr entries in my registry and used several setups with lower clockspeed and lower voltage.
Sometimes my GPU seems to spike to 100% which causes a crash, but sometimes it doesn't.
The only thing thats maybe a bit special for me is, that I use a 49 inch Monitor from Samsung.
However theres so many Games that just work. Maybe someone can help me.
are you using 2 separate cables from psu to gpu ?
I'm using 2x 6+2 Connectors from PSU to GPU
to be sure, not speaking about connectors ,you have to use 2 separate cables , not just one splitting in two connectors
Sorry, I am indeed using 2 separate cables to connect the GPU to the PSU
okay, mmh... saying you lowered voltage and clockspeed , did you see problematic temps ?
did you check windows integrity ? (sorry i don't have the command line in memory)
speaking about spikes and then crash, i would still investigate with some power weakness..
and other thing you can test is putting your gpu in another computer (friend computer ?) to check if the behaviour is the same
Temps were mostly in the 60°C range.
I actually havent tried the sfc and dism commands yet. However I had to hardreset my OS a few times so i figured it shouldnt be a problem.
Will test that tomorrow anyway.
I might also just grab my 1070 from my old PC and check if it works with that.
Thanks for the Input for now. I also noticed that I opened this thread in the wrong category...
I have a feeling that I might repost this question in the driver & software category once i'm done checking my options
thing is 1070 and 6900xt OC not using same wattage, that's to be remembered regarding results than , in the case it would "work" with the 1070
in my view i don't think your problem is a driver problem, not directly at least , until driver files are corrupted , or corrupted windows installation.. .. coming from a msi 6900xt gaming x trio OC , i never had crashes etc whatever drivers I installed , i always try to be up to date just waiting sometimes to see report when beta drivers , but never had problems , so sticking to 22.5.1 doesn't feels relevant to me either as you don't seem to have an exotic setup
You need to put each pcie cable on its own 12v rail if available. Many modular ps's have individual rails for each pcie connector on the ps.
Look at your model's power distribution chart to ensure that its not power sharing its pcie rail with any peripherals and if it does don't plug into that port.
My aib 6900xt would trigger the pcp if the 12v rails had any peripherals on them too. I ended up replacing that 1350w enermax platimax with an evga supernova 1300w where each pcie connector has its own 12v rail.
Sorry for replying so late.
In the meantime I brought my GPU to the local hardware shop where I bought the GPU. They ran some benchmarks and the GPU ran withouts any flaws. However, they also tested Unreal Engine 5 and ran into the same driver errors as the ones I mentioned at the start of my Post.
They didn't test any games that I mentioned as thats not something they do, but I assume it would have been the same result.
I also checked the Manual of my PSU and the Corsair HX1000 has a single and multimode for 12v rails. I tried different positions and settings, but same results.
If my GPU creates the same errors in another Setup, it's probably related to driver issues, or not?
At least I'm running out of ideas.