Radeon RX6800xt Issues

So finally got a hold of a 6800xt and got super stoked. Uninstalled my Nvidia driver with DDU, removed it, installed the AMD card and the latest drivers. Immidiately i notice that the core clocks are all over the place (compared to what I’m used to with nvidia), but I think nothing of it.

Played some wow and today i wanted to play some Warzone. Clock speeds again all over the place, almost no utilization, CPU pinned and I’m hitting 10-20 fps in the menus, and all of a sudden the game freezes and the screens turn of.
I try to reset my via the reset button, but nothing happens. I hold the power button until the systems turns off completely and turn it back on, and the screens are back.

My system is a z370 board with a 8600k, and 16gb of ram. 850w seassonic prime px psu (with two cables).

I play at 1440@165hz
Temps were never over 70c.

Swapped back to a 3060ti that I also have, and everything is back to normal with solid speeds etc.

I also had Fortnite installed since my nephew was here recently and i thought to try that since that’s known to be fairly optimized (at least what I can recall), and 9fps while dropping and 20 roaming around the woods.

Tried Bfv as well, 10-20 fps.

Have also tried to format my system drive and install AMD drivers completely clean (zero trace of nvidia) but no luck at all.

Is my GPU dead?

Oh no doesn’t sound good. Do you not get the bios screen with the 6800xt at all?

Hi Wendell, no and pretty dissapointing. Nope, when the screens went black, they were completely off even if I pressed the reset switch on my case multiple times. Can also confirm that the reset switch works as intended. The only things that brought the screens back to life was hard powering off the system and powering it on again.

Edit to clarify:
Normally I get the BIOS screen, but not the times it «Blacked out»

it might not be the gpu, it might be the over current protection in your seasonic psu.

several folks reported and I can confirm that the seasonic psu overcurrent would trip with a 3090. maybe it’s also happening for you?

1 Like

I see. How can I test that reliably? I connected a powermeter to my wall plug and I never exceeded 500w, but I guess it would be hard to catch a spike.

I found it weird that i feel like i bought the best possible PSU available, but that isn’t good enough. What else can I do?

Put the 6800xt back in and did some testing.

Did a couple of Heaven runs and Time spu runs, and it seems to run those fine.

In Timespy I got a graphics score of 17 897, which to me seems good, comparing to guru3d’s benchmark.

Also BfV runs better now, but not as well as I’d hope - 130fps ish which seems low compared to TPU’s 190fps albeit with a better cpu, but should cpu have that much to say on performance?

CoD is running like sh*t, but I notice that it bogs down my whole cpu, but only when I have a AMD gpu installed, so I’m thinking that has to be a driver issue :thinking:

Depends if the CPU is the limiting factor for that game. If the CPU is pegged at 100% and the GPU not then the CPU is the limit.

Try turn the resolution up, might not get any higher FPS but you should hopefully see it switch from 100% CPU to 100% GPU load and the CPU drop down.

When switching cards around in windows make sure to run DDU to clean up other drivers. There have been cases where they can cause performance issues when they don’t have something to grab on to. It’s generally a rare thing, and the most recent case I recall was AMD drivers eating cpu in some weird edge case that was fixed, but it’s possible:

Thanks for the heads up, and I have done that every time. Started in safe mode, DDU and shut down, swap cards. But to me it looks like a driver issue. Going to test with a 4k monitor later and see if it helps put some more strain in the gpu

I believe there is a Virtual Super Resolution option or something to that effect as both nVidia and AMD have it under different names. You can we the card to render in 4k (2160p) even if you only have a 1080p monitor. So without the hassle of swapping monitors you can try it.

Okay, cool good to know. Had the 4k monitor handy so tried that. All of a sudden the screen went black again, but pc did not turn off.
I restarted the pc and the screen lit up again, and worked normally until I logged into windows, were it went black. It continued to do so, so I decided to format. And twice now, as soon as I install a driver, the screen just turns off as soon as windows loads.
I’m really as a loss now :man_shrugging:

See if you can use RDP from another computer to access the machine. Will let you know if windows is still working or not, and make it easier to try different things.

Other things to try:

  • Disable PCIe power management in BIOS (ASPM)

  • Disable PCIe link managament in windows settings in power settings

  • Try earlier versions of drivers.

  • Disable any overclocking of ram/cpu

  • Manually set the PCIe state to version 3 or 4.

  • Disable smart access memory if that’s an option.

Thanks for the advice Log. Seems like i fixed it, i noticed the “something is really hot smell” and noticed that the gpu ran really hot. Let it cool down for a while, then made sure it had more cooling (mITX), and now it’s seems to run normally. So seems like this issue was just my own fault :man_shrugging:

Curious what your pc case is

Just out of interest did you catch the Temps at all?

Oh wow, I geniunely can’t believe you overlooked cooling in 2021, especially a high tdp card.:joy:

1 Like

@ regulareel the case is a Corsair NR200 without tempered glass and with two Noctua NF-A12x25 fans In the top and in the bottom.

@ Zibob did not catch the temps since I never got into windows unfortunately, but felt damn toasty

@ CybeastRaystriker the issue was that the fans are connected to a Corsair Commander Pro, and for some reason it didn’t even start the fans to the minimum speed it’s supposed to. First time that has happened since I got it a couple of years ago. So Wouldn’t say I overlooked it, but rather that it failed. So I genuinely can’t believe you jumped to that conclusion based on one comment alone.

@wendell I tried installing the 20.11.2 WHQL driver from AMD and everything seems normal now. Warzone is stable and running nice, same with BfV.

I’ve gotten so used to having and older card (980 ti), so I’ve just downloaded from TPU, and first when I checked with AMD directly I saw that they had the lastest and the WHQL driver.

1 Like

hwinfo64 is your friend for temps :slight_smile:

3 Likes

With all due respect, you did indeed overlook something as simple as some fans not spinning. You changed gpus, formatted the system, did a few benchmark runs but didn’t notice fans not spinning? Didn’t look at gpu temps? C’mon dude. Not flaming you or anything, just saying that we all need to remember the basics and have a good presence of mind.

The fans are below the gpu I can’t really see that they are not spinning, unless I flip the whole case around.

I did not do any benchmark runs when the issue with overheating was, since the screens turned of right after logged into my user with my pin. Therefore I was also not able to check temps, since again the screens turned off right when I entered windows.

I also find it weird for a GPU to get that hot during idle, since the most cards now have idle fan stop I don’t expect any fans to run during startup of the system.

1 Like