Drivers & Software

[CitizenVeen]

Hi all,

My AMD driver keeps crashing while playing games. Sometimes it crashes the game, I get an error message from AMD and I need to restart my system or sometimes disable/enable the driver. On other occasions, I just get a black screen. Fans on my GPU start blowing like crazy, sound still works, and I need to switch of my system by keeping the power button pressed 5seconds.

This is my setup:

GPU: ASRock Phantom Gaming X Radeon RX 590 8G OC
CPU: AMD Ryzen 5 3600 Boxed
Motherboard: MSI B450-A PRO MAX
RAM: Corsair Vengeance LPX CMK16GX4M2D3000C16 2x 8gb = 16gb
SSD: Intel 660p 512GB

System is 1year 9 months old.

Crashes happen consistently while playing Valheim, so I've been using that to check. I checked my stats while playing and while it crashes:
GPU temp is arround 80-83 celsius.
Fanspeed arround 80%.
Clock speed is around 1300 or so I believe.
Power was around 60% of max in the AMD program, but forgot how much it was. I can retry

I completely formatted my drive, and reinstalled windows.
I've installed and updated all drivers: motherboard, GPU, CPU, chipset, Bios.
I tried setting power Tuning to +50% in the AMD program.
I reinstalled games.
I did windows updates.
I cleaned my PC inside from dust. (Not inside the GPU or the GPU fans, because I need to remove my card to reach them, should I do that?)
I've installed a new heatsink & fan for my CPU because the temperatures were somewhat high.
Tried using an old AMD driver.

This is an error message I've gotten while the AMD program is active:


And when it is not running, if I try to start it after a crash it says this:

And also this:


As you can see techpowerup GPU also records no sensor reading from the GPU here anymore. (Restarting the system removes the errormessage and techpowerup records all sensors again).

I feel I've tried everything. I looked around on the AMD forum, I've tried all solutions or tips I could find. Anyone has any ideas, suggestions or ways to solve this issue? Several games are simply unplayable because they cause so many crashes.

Thanks a lot.

[elstaci]

Try stress testing your GPU card and see if it passes.

You can use OCCT GPU test and GPU MEMTEST to check your GPU card  and RAM for errors. Or you can use Furmark and Furmark's Artifact test.

Is your GPU card overheating? That would cause it to crash during game play?

Also is your PSU sufficient to run your computer with a RX590 GPU card installed? You should have a PSU of at least 500 -550 Watts.

[CitizenVeen]

Thanks for the help.
I did a OCCT 3d stresstest and it crashed my system after 2minutes 49 seconds.


I have a 450W PSU, I calculated my power draw here: https://outervision.com/power-supply-calculator and it says 372Wats is what is uses max. So there is still alot of room for error there, don't think I am short on power.
OCCT didn't detect any errors (before my system crashed).

Thats leaves heating then. Temperatures go to 81-83 celsius instantly when the GPU is under load (says 81 in the screen, but I've seen 83 as well), and will stay there. Is this too high? Should I try to undervolt my card using the AMD software? Or turning fan speed up? It's all on default tuning setting now.

thanks again for the help

[elstaci]

I believe it is a running hot under stress but depends on the GPU Card. I read, for instance, the RX580 can support temps up into the 90C but most mention it should run at or below 80c.

Try increasing the Fan curve to a very aggressive fan curve and see if it still crashes during the stress test or open the side computer case panel and put a floor fan aimed directly at the GPU card and see if it passes the test. That should eliminate overheating as a problem if the temps stays at 80C or below while under stress.

Also what was the PSU Output during the tests for 12 & 5.0 & 3.3 Vdc?  They all should have been -/+ 5% tolerances. For instance the 12 Vdc should have shown a minimum of 11.4 Vdc or higher during the testing.

Run OCCT GPU MEMTEST and see if your Ram check good without any errors or run Furmark.

Here are some GPU testing software from this tech site: https://beebom.com/best-tools-stress-test-gpu/

also run MEMTEST86 or Windows Memory Diagnostic just to make sure your System RAM Memory is physically good and not going bad.

If  the PC is crashing using a Stress test, at least, that indicate you have some sort of issue with either the GPU card or driver or some other hardware.

[CitizenVeen]

Thanks once again for the reply and time you're taking to help me out.

I cleaned and applied new thermal paste to my GPU (which wasn't really needed, barely any dust, and old thermal paste still seemed fine) and made a much more aggressive fan profile, which goes up to 95% at 79 celsius (rather then 85% only when the card gets to 85 celsius, which is default). This now keeps my card at or under 80 degrees, even when benchmarking on 1080 extreme (used Unigine Superstition).
System doesn't crash anymore when doing a stresstest, but the driver still crashing when playing video games, so, as you said, I think it's a driver issue, more then a hardware / temperature / voltage issue.

So got my card a little cooler now, which is always nice, but issue is sadly still unresolved.

Did Windows Memory Diagnostics and OCCT memory stresstest. Both went fine, yielding no errors and no crashes.

I am not sure where I should check the PSU output in OCCT or anywhere else. I can only check the voltages that my cards are consuming, not the PSU output, or I haven't found it.

[elstaci]

This is my image of OCCT for my PC. As you see my PSU Outputs are well within 5% tolerances for a PSU that is over 8 years old:

Well it seemed like part of your GPU crashing was do to getting to hot under stress. Now it does look like a Driver issue or could  also be a Configuration issue in Radeon Settings or Game Settings and also in Windows Settings - Graphics.

Someone else would need to help you since I am not really a gamer.

[Bobbyhardnut]

Well I've read so many of these over the last year while I battled my rx 590 fat boys  random nonsense. This caught my eye as its recent and all too familiar.  I assume you still have the issues. Ii run mine with a ryzen 5 2600x, may be relevant. Not sure. 

Don't follow my route.  I've replaced everything. Psu upgraded. Ram swapped case open and closed.  New fans and cooling, windows 10 and 11. Replacing the cards bios.  Multiple bios tweaks on my mobo, and finally today.  A new mobo. Went from gigabyte to Asus.   My dismay when it crashed after playing rust for all of 10 minutes :(. 

Its a royal pain in the xxx.  I honestly may never buy a amd card again and I've been gaming exclusively on amd hardware for  15 odd years. 

I do have work arounds but my mission was to identidy why the hell it was like this.  Today's motherboard upgrade was the final piece of a jigsaw for me. 

In a nutshell. Its not the card. I dropped the RX590 into an older rig running a very very old phenom 965.  The rx 590 worked perfectly minus having the ability to hit 82 degrees which is the point at which my card gives in.  By manually setting the fan curve in the amd driver it would game all day long at 77 ish degrees and never once crashed. I can't explain this. It makes no sense to me. 

Right now I belive its somehow the Amd adrenal drivers or the chipset drivers that cause some sort of Conflict.   

M workarounds: please try and let me know if these work for you and our thoughts. 

1#  underclock the card.  Minus 50 on the core voltage and cap the core clock to 1450.  (there is wiggle room for more Mhz than 1450 but ive kept it simple)  

This works 99% of the time but I get new issues pop up with things like alt tabbing from games. The card runs much cooler and performs well enough to stop me throwing it at a wall. 

2# this is a new discovery for me learned only today as I was waiting for the new motherboard to arrive. 

Use amds driver remover tool from safe mode in windows. Remove all amd chipset drivers too.  Make sure you remove the folder @ c:/Amd. 

Do not install the chip set drivers and do not install the graphics drivers. Let windows do its thing and install its own versions.  

Enjoy playing your games with exactly the same performance as with the drivers installed manually.  You lose some of the settings in the Radeon software and you may need to use something like riva tuner to set a custom fan curve to handle temps but the card gamed perfectly @ default clock speeds 1580mhz  for the first time in ages.

I've spent longer trying to fix this than I should have. The only reason I've stuck with the card is the state of the market right now. 

Let me know how you get on and if anyone else can suggest why the card worked flawlessly in my older rig or without the Adrenaline drivers I would love to hear.

Typed via phone. Forgive me. 

[CimmerianRhoba]

Hi, I had this issue recently. Mine I believe was directly connected to temperatures, despite me thinking it was a faulty game/ software issue.

With yours however I do believe it to be the PSU. Sure, the tolerances are looking fine on stable tests, however on your PSU, which is fairly old, I believe it is not able to withstand sudden increases in power draw, and even so, on a 450w PSU, will be quite demanding. I believe these spikes are causing instability, thus causing the device drawing the most power to shut off, in your case the GPU.

Maybe invest in a Bronze rated 550W PSU. Theyre around £40, ofcourse they can be cheaper, but at cost of quality.

[Chester1994]

"1# underclock the card. Minus 50 on the core voltage and cap the core clock to 1450. (there is wiggle room for more Mhz than 1450 but ive kept it simple)"

^ This works. I did it some years ago a friend's GTX 760. The card would TDR after a while. Temps were not high, PSU was 600W certified. However, it stopped timing out the drivers after clocking it down like 10%. It will slightly impact the GPU's performance, but will give you a reliable system, which is always the priority.