Troubleshooting advice and when to RMA?

[flumph]

Hi All,

PC Specs

AMD Ryzen 5 5600X
Gigabyte RX 6900 XT GAMING OC 16G
MSI MPG B550 GAMING EDGE WIFI
32GB Corsair RAM
1TB Sabrent Rocket 4.0

Background

I purchased a 6900XT around February of this year (Gigabyte RX 6900 XT GAMING OC 16G) at a pretty penny.

Since the start I have had some problems with games randomly crashing, but it was rare enough that I put it down to the game only.

Tweaking the graphic settings of the games would reduce the crashes substantially (disabling FSR, RT etc.).

Turning on "RT" in a game will instantly crash it.

I was able to put a few hundred hours into any of my games (including the ones listed below).

In the last 4 or so months it has become progressively worse however. I'm now at a point where any relatively modern game nearly-instantly crashes:

RE Village, RE7, RE2, RE3, Back4Blood, Doom Eternal, Halo Infinite, Metro Exodus: Enhanced Edition, etc.

A few games I have tried that are over a decade old will run without instantly crashing.

The behaviour is usually that I can make it past any logos/title screens but as soon as I get into the main menu the game hangs and I get an AMD driver timeout error.

Chrome will also randomly freeze when playing videos and I will get an AMD driver timeout error.

I can run Furmark (OpenGL only?), but I cannot run 3DMark DX12 benchmarks at all, I get "an error occurred".

Troubleshooting

The steps I have taken to remedy this:

• Turning graphics quality down in games - this helped for the first few months.
• Installing the latest drivers
• Installing beta drivers
• Installing as listed on the manufacturers website (2020).
• Installing drivers from the start of the year.
• Installing minimal drivers
• Installing drivers only
• Turning On/Off "smart access memory"
• Upgrading motherboard BIOS
• Installing motherboard chipset drivers
• Buying a high quality / brand new power supply (1000w)
• Installing Windows 11
• Installing Windows 10

Edit: I forgot I also have played with these:

• Switching GPU bios between the "OC" and "Silent" settings
• Disabling Steam overlay
• Disabling Radeon overlay

Next steps?

I'm not sure what I should do now?

I've gone to every length that I can think of and after months of trouble shooting I'm at my wits end. I'm tired of having a blank PC all the time from nuking it to install the next combo of drivers and settings.

Does anyone have any other trouble shooting tips?

I think it is a hardware issue, which I have been really pushing back as without a GPU I cannot use my PC so any time without a card is an annoyance at best. Although now I wish I just sent it back months ago.

Do people have any experience with the RMA process?

Thanks for your time,

 

[flumph]

I'm using a Seasonic Prime PX-1000. All three 8pin PCIe connectors have their own cable back to the PSU.

 

[flumph]

The game freezes and the game screen goes black both in Windowed/Fullscreen modes. Some games will not hang and will crash to desktop.

Windows itself remains OK, and no BSOD.

I will then receive a pop-up:

"AMD Bug Report Tool"
"AMD software detected that a driver timeout has occurred on your system."

I get this even when trying to run the "stress test" built into the "AMD Software Adrenaline Edition" drivers... lol

Sometimes I will also get a crash tool pop-up for the RE games.

The entire window for Chrome will flicker and the video will black out, then a few seconds later I will get the above AMD driver pop-up.

From the stress tests that do run (3DMark Firestrike) I got a max CPU of 65C and 75C for GPU.

It's a pretty cold day though, and I'm sure I've seen those edge towards 90C in games when they were working.

Edit: I say all that and Windows itself just froze up completely - even the keyboard wasn't registering changes to CAPS LOCK - then the screen went black and then returned and there was an AMD driver timeout pop-up waiting for me...

 

[flumph]

I'm using 4 sticks of Crucial Ballistix 8GB DDR4-3600 for a total of 32GB.

I have tried the system with XMP enabled (3600MHz) and disabled (2666MHz).

I have tried disabling PBO.

I have also tried down clocking the GPU (by setting the maximum MHz to 80% in the Radeon drivers).

I just completed a run of "Windows Memory Diagnostics" and checking the Event Viewer it found no problems. Are there any other ways to check system memory? Years ago the advice was to run Prime95 overnight.



Edit: I just tried some tests again with XMP off and PBO disabled since going back to Windows 10 and some of the near-instant crashes have stopped, but I will still get a crash (and sometimes an AMD Driver timeout warning) after a few minutes running a stress test. Back4Blood and RE village now at least gets to the main menu. Enabling RT gives me an instant crash. TimeSpy (DX12) now runs for 30s before crashing rather than crashing instantly.

I'm away for a few days but I can try swapping my GPU out and putting it in the Wife's PC next week (and the hassle of DDU-ing the nvidia drivers etc.). Although I think her PSU is only 500w...