PC Drivers & Software

[hd1]

Describe your system:

• AMD Graphics Card
  • Sapphire RX 5700 XT 8GB Radeon, Stock Blower Fan Variant, bought right after Radeon RX 5700 XT was released
• Desktop
• Operating System
  • Windows 10 64bit Version 1909
    
• Driver version installed
  • 20.4.2
• Display Devices
  • Single monitor 3840x2160 benq pd3200u on displayport
• Motherboard + Bios Revision
  • Gigabyte H170N-WIFI, BIOS AMI F22e
• CPU/APU
  • Intel Core i5-6600 @ 3.30GHz  4 Cores, 4 Logic Cores
• Power Supply Unit  Make, Model & Wattage
  • Corsair RMi Series RM750i 750W
• RAM
  • 16 GB RAM

Describe your issue:

Squad, Digital Combat Simulator, Unigine Superposition Benchmark experience flickering of parts of the screen (mostly white). Sometimes short black screens, then back to the game. Sometimes artifacts all over the screen. Sometimes the game crashes to Windows desktop. Sometimes the desktop has artifacts all over after the crash. Sometimes after the crash to desktop with artifacts the computer will crash completely. Sometimes the game crashes the whole computer without warning. And everything inbetween.

Before gaming I start Unigine Superposition Benchmark in game mode where you get a static screen if you don't move and let it run for ~5 minutes to see if I get flickering and artifacts. If I do I just reboot and try again until it "appears" as if I don't get flickering. Then I can usually game for hours without problems (sometimes it would still crash). How often I need to reboot to try this varies approximately between 2-10 times.

Furmark did not crash in the past even though games did (did not retry with the recent drivers).

I already switched from a 650W PSU to a 750W one. The next things I am doing is getting a powered USB hub and 32GB RAM (both unrelated to the issue, but maybe it helps with it too).

This issues has been ongoing in all driver versions since I bought the GPU shortly after it was released.

The weird thing is that sometimes there are no flickers, artifacts, etc. and games would run for hours without issues. But if there are the slightest issues in my Unigine Superposition Benchmark test before gaming, I can bet on it crashing the game and/or computer.

EDIT: I am aware of most tips and tricks that have been posted in the mega thread and elsewhere. I repeatedly filed the problem with AMDs support system. I just explicitly mention it in a new thread because AMD requested us to do so in the now defunct mega thread. Still thanks, for any suggestions you have. But I believe it is a driver issues that needs to be fixed. After all the situation slightly improved over the months with newer drivers.

[morelock]

How are your temperatures?

Any overclocking/custom settings?

Pigtail connection from the GPU to PSU or two cables?

Tried an HDMI cable instead of Displayport?
Have you eliminated ram issues (memtest86 passed) - probably not it but worth crossing off the list.

[hd1]

Thank you for the suggestions!

Temperatures are normal.

No overclocking/custom settings.

Two cables from PSU.

Never tried HDMI. Might try it but I think it's a driver issue as behaviour slightly improved over the updates.

New RAM is on its way (replacement modules because of general upgrade to 32GB). But previous GPU had no problems with same PC.

[hd1]

The issue still occurs with two new RAM modules that replace the old ones.

[trek]

What you describe looks rather like defective card. I personally would reseat card in PCIE slot and then reseat all PSU cables, make sure everything is plugged.

If that does not work I would RMA card, because PSU problem in my opinion do not cause artifacts (and you have nice new PSU). Everything just points to defective card.

[ademk28]

I have everything he described but with a Vega 64. I've never had an issue like this, and it's a brand new card straight from manufacturer.

[hd1]

Thank you for the suggestions!

I believe it's a driver issues. Because over the months the situation improved slightly and sometimes it runs without problems. And because of the likely change of behaviour on reboot and the way the drivers improved things for everyone over the month. I guess the drivers are tuning something dynamically. Maybe on reboot based on temperature or voltage or something. I don't know much about GPU drivers. But I guess it sometimes makes a wrong setting on some minor voltage or temperature fluctuation on reboot.

After all there are standards for OpenGL, Direct3D etc. and still GPU manufacturers update their drivers with problem fixes for specific games. Also users (like ademk28) with other kinds of GPU are affected, this makes me think it's a driver problem.

[mickeekung]

Please try this

rawintellectMay 9, 2020 11:55 PM

Access violation errors (Exception code: 0xc0000005) can happen because of different reasons. Corrupted files being one of them. I was getting this error daily. After doing a fresh install of the drivers (DDU and factory reset), I'm currently testing whether or not giving RadeonSoftware.exe admin access will resolve it if it is indeed running into access issues.

I'm on day 3 with no black screen yet. At this point I'd do anything to resolve it. I also gave full access in the security tab and disabled windows "full screen optimizations". I have no idea whether this will work or not but it can't hurt. I'll do what I can until AMD finally fixes this issue.

[hd1]

Thank you for the suggestion!

But I'd rather not try more Windows tweaks to fix AMD GPU driver issues. They should just work. There should be no need to even use a special uninstaller before installing a new driver version (I tried several of these methods, none helped). Anything like that is ridiculous. That AMD themselves suggests to do that is very weird from a software development perspective. But maybe there is something specific about drivers in Windows that I don't know about, that could mysteriously cause files or settings to be used that are not part of the installation process and were leftovers from a previous install. Very weird.

Still, thanks for trying to help.

[ademk28]

Install AIDA 64 and check for a stability test. If you have any stability issues your PSU could be the culprit for random system crashes. Still I wouldn’t go off that until you test all of what I’m about to tell you. Underclock your GPU and see if that fixes the artifacts. If it does you may need to either get an RMA on GPU, reflash the vbios of the GPU or it may be a driver issue that’s affecting only a certain amount of cards. artifacts  and system crashes are all linked to your GPU. You PSU may not be enough for the card (highly unlikely unless psu is defective or low rated psu.) GPU frequencies are too high not stable, or temp may be too hot to occupy the given frequencies. Check temps when GPU is underload, if temps go beyond safe temps for your card then decrease power limit. If still no avail, try underclocking the GPU. If that doesn’t work, you unfortunately have two options left. It’s either a defective card and you need to get an RMA or it could be a possible driver issue although it’s highly unknown until you test it, and certainly you must try a bios reflash for your card as a FINAL resort ONLY. Don’t reflash bios unless everything is hopeless and you can’t get an RMA.

[hd1]

I'm seriously thinking about RMAing the card. More than eight months of this driver mess is enough.

[hd1]

Still crashes on Driver Version 20.5.1. Filed another report through the bug report system...

[mickeekung]

Can you update window to 20H1? I think it will fix.

[hd1]

This has been ongoing since I bought the card in July 2019, that's almost a year! It took AMD about half a year to even publicly confirm that there is an issue. And the issue still exists.

I got myself a card from a different manufacturer. That manufacturer sucks too. The card is more expensive for about the same performance. But at least their drivers seem to work. Will try to RMA this junk.