Graphics Cards

[Naustil]

I bought a 7800 xt card a few weeks ago, upgrading from a GTX 1080 and have had nothing but issues since upgrading.  Previously I had absolutely no issues with my system with my GTX 1080 and have had all kinds of black screen/reboot issues since the swap. Issue seems to be random at best (between 5mins-1hr)

Initially I ran the uninstall for my nVidia card and used DDU in safe mode to clear up all the old drivers.

Installed new 7800 XT card with latest drivers 23.11.01 (Yes I tried most recent 23.12.1 drivers)

My system is Mobo: Aorus X570S Pro, CPU: AMD Ryzen 7 5800X3D, PSU: Corsair RM850x, RAM: Kingston DDR4-3200 (KHX3200C16D4/8GX) in slots 2 & 4, Single Monitor Aorus FI27Q-P 

Temps and PSU are not an issue as temps have been reasonable and the PSU is new. Things I have tried based off of notes I have read off various websites:

1) Disabling PBO & CPB

2) Disabling XMP profiles in BIOS

3) Updating BIOS (New BIOS had new features supporting my CPU specifically)

4) Undervolting

5) Lowering clocks

Etc.  I've also tried clearing steam cache and various other minor fixes to no avail.

As someone who has worked previously at ATI/AMD for ~9 years this has really been a terrible experience and is most likely related to unstable drivers in my opinion (though it's been impossible to pinpoint the exact issue).  I'm really hoping AMD can figure out the cause and address the issue.

I can only imagine how many people are experiencing this issue since not everyone is joining a forum and giving their feedback,  I wanted to add my experience since I've been struggling to fix this issue for awhile and have seen no specific solutions that have worked for me.

Good luck everyone, hope we all get the fixes we need

Hopefully AMD is aware of the issue (even though they have not acknowledged it as being a significant issue) and is working towards a solution that works for everyone experiencing similar problems.

[learnerspermit]

Assuming a PSU is good because it's new is just bad practice. Test it along with system memory before jumping to conclusions as we all know what we get when we assume.

[sudomywoodo]

Hundreds of people are reporting the same issue. I highly doubt the PSU is to blame

[Naustil]

Yeah, I wasn't assuming that the PSU was not a problem, to be clear there are a couple scenarios in which it could be a problem i.e. the additional cable and connector from the PSU since the new card uses 2 connectors and the old one only used one.  My comment regarding the PSU being new was to suggest that I don't believe swapping to the new video card caused any damage due to different load levels.  There should also be more than enough headroom in terms of power.

I wasn't trying to solicit any advice for further testing, I know there are some things I haven't tried but with regards to power, given that many other users have tried swapping it to no avail and the fact that it had been running stable in my system for 1.5+ years it is not at the top of the list of suspects for the root cause.

I mainly just wanted to share my story so I could commiserate with my fellow AMD'ers, I'm keeping the card because I do believe the issue can be resolved and if I limit my FPS to < 75 the issue doesn't occur as frequently so I'll tolerate it in the interim, while I try to find solutions when I have the time!

[Naustil]

Update:  Since my initial post I've tried a few more things and I think the system is stable now.  I need to do further testing to confirm however I wanted to share my findings.

1) When I initially swapped video cards I had forgotten to enable Freesync in my monitor settings (By default it is in G-sync mode). However, Freesync wasn't enabled in the drivers either so honestly it should not have caused any major issues, at least or so I thought. So I changed both of these to be on.

2) I rechecked all the power cables leading from the PSU to the GPU.  They all seemed snug altho I think I may have heard a slight click from one of them so possibly it was not in 100%? They all looked flush but can't be sure.

3) There was a patch for Cyberpunk 2.0 > 2.1, not sure if there was anything specific for my config even tho I read thru the patch notes. This may not be a factor.

These changes allowed me to run Cyberpunk with medium ray tracing on for over 1.5 hours and no issues at all (before this used to crash usually between 5min-10mins).  To try and test which setting was causing my black screen/restarts I put the monitor back into G-sync mode and disabled Freesync on the drivers but Cyberpunk still seemed stable (didn't test it long tho)... however when I changed the V-sync from Off to 144Hz in Cyberpunk settings it definitely introduced the instabilities I was seeing before that eventually lead to the black screen/restarts. 

I will have to do some more testing since the problem was not consistent before but maybe the above notes may help someone!

Cheers

[Naustil]

Update 2:

Was able to run Cyberpunk for 2+ hours with absolutely no issues, followed by Everspace 2 last weekend and I was thinking the Freesync/V-sync settings might be the solution, however when I played Wingspan (a less graphically intensive game) after my second game I got a black screen/restart.

Further testing with Wingspan, desktop running 144Hz refresh, Wingspan will run at ~72FPS with V-sync enabled and 30FPS with V-sync disabled.  I definitely had more flickering/unstable graphics when using the Adrenaline overlay and V-sync enabled. I tried using Adrenalin's Enhanced Sync and I was able to get the variable refresh but again it was flickering within the game and with the Adrenalin overlay on. With V-sync off the game had no flickering in game or within the overlay.

Considering that I also had more instability running Cyberpunk with V-sync enabled previously (mentioned above). My best guess for the black screen/reset issue seems to be a conflict between V-sync and Freesync settings.  Why or how these settings working together is causing so much instability is a mystery. When both are enabled I will eventually get a 1-2 second switch to black screen and after a while a black screen with looping audio followed by a restart.

I think I've done about all the testing I can without a lab full of similar setups and unlimited time. I hope AMD or anyone else reading finds my analysis to be helpful/useful to reducing the issue until a real solution can be found.

Cheers!

[Naustil · Accepted solution]

Final Update:

Well after trying a handful of other fixes with no success I decided to RMA my card. Within a few days they received and tested my card an issued me a new one with new serial number. After getting the new card back a few days after that I've had absolutely no issues (other than the pause when loading some games at the beginning). The card has been running several games for several hours over the past 1.5 weeks and I can report no more hangs, black screens or restarts. Turns out it was the GPU all along.  RMA didn't specify what exactly the problem was i.e. cold solder, bad memory chip... who knows?

Anyhow, happy to finally have a card that's working!