RX 590 woes

Hardware GPU
1

Hi

  • All drivers are up to date
  • Motherboard bios is up to date
  • I’ve disable my CPU overclock for testing

System:

  • AMD FX 8320 (OC to 4.5Ghrz, but OC disabled for testing. CPU Currently running stock)
  • Asus SABERTOOTH 990FX R2.0 Motherboard
  • 12Gb of HyperX DDR3 1866Mhrz CL10
  • Superflower 550w PSU
  • XFX RX 590 Fatboy

I bought a RX 590 in February and all seemed fine for the first 2 weeks or so when my whole system crashed while gaming. It just completely locked up and my display went dead with the sound repeating.

So I hit reset and I could hear the system boot, but the display was still dead. I waited a little to be sure I was on the login screen and then pressed the power button so it would just do a normal shutdown which it did.

After powering the PC again the display was back. (A cold boot is the only way to get the display back after such a crash)

My first thought was it might be overheating… these things do run pretty hot after all. So I fired up hardware monitor on my second screen and loaded up the game again. No crash and it seemed fine.

Another 2-3 weeks pass and it happens again. During this time I also got a new ssd and I was hoping a fresh install of windows might solve the issue. It didn’t. Over the last 2 months it’s happened about 5 times before yesterday (8 April 2019).

I’ve run different stress tests and benchmarks which easily got the card up to 81 degrees and it seems fine. Seems to be game related.

Yesterday (8 April 2019) I started up War Thunder for the first time in what must be 6 months. So it’s the first time I run this game with this card. After about an hour of play it crashed. I restarted the machine, went back into the game, played another 5-10min and it crashed again. I was keeping an eye on the temps and saw a max of 81 degrees. I tried lowering the quality settings of War Thunder and it would still crash.

Spent quite some time on the net looking for possible answers, but didn’t really find any. In one day it’s crashed about 8 times so far while playing War Thunder. NOTE… it’s crashed in other games as well. It seems War Thunder just causes the crash more reliably.

Today I opened the side panel, fired up the aircon and placed 2 large fans next to the case. I also set the fan speed control to manual in Wattman and had it go to 100% at 75 degrees. With this setup I was able to keep the GPU below 75 degrees.

So I fired up War Thunder again and after 1 hour and 30min of play… no crash. Max GPU temps was 73 degrees.

I know about undervolting, but so far I didn’t really want to touch that as the thing should work fine as is out of the box IMO. However, I did notice a max voltage of 1,212 in HW and the default should be 1,150. After setting the voltage to manual it did have 1,150 by default in Wattman.

So I left it as is… did not change any voltage numbers. All I did was change it from auto to manual… that’s it. In theory, nothing should actually have changed, right?

I run War Thunder again and after about 10min the screen went black, but no actual crash… then it came back with artifacts on the screen and the system seemed to be locked up.

Again, doing a burn test on the GPU and CPU at the same time caused no issues with the card getting up to 82 degrees and the CPU getting to about 50 when stock and about 70 when overclocked.

So it seems heat related at this point, but considering that 80 degrees is where this card runs under normal gaming (As seen by multiple reviews on the net) I would suggest the card is unstable and thus faulty.

Now, I’ve worked in IT for almost 20 years. I know that if I send the card back, they will do similar stress and benchmark tests to see if it’s faulty and will find nothing as I have.

So:

  • What Temps do other 590 users see while gaming?
  • And does anyone here have any ideas about what I can do to prove it is in fact faulty?

Thanks

2

I’d suggest that you open a case with XFX.

I had a similar problem with a GTX780 card. For several months, everything was fine. I got a new game that made the card work hard and it started randomly locking up with a black screen, with the sound playing in the background. Temps were OK and the lock ups seemed completely random. No lockups with other games/programs. The problem started happening more frequently and instead of a blank, black screen, I started getting odd colored artefacts on the display when it locked up. Thinking that it was definitely heat/load related, even though the temps displayed OK, I pulled the cooler off. EVGA had only applied thermal compound on 1/2 of the GPU die.

Long story short, as soon as I used the “A-word” (artefacts), the EVGA rep said that it was definitely the card and arranged an exchange.

Best of luck!

1 Like
3

That is a thermal problem indeed.
Might be thermalpaste related, might be due to poor cooler design.

Open up a case with them and before you send it off, take pictures of it (fans intact, nothing bent, package condition, etc.)

4

Oh, and don’t remove the cooler, like I did. Most manufacturers will cancel, or at least try to cancel your warranty if you intrude into the inner sanctum. EVGA is the exception to the rule. At least they were, beck when I had this problem. I attached pics of their paste job to my open case and all they could say, was D’oh!

5

So I’ve been doing tons of more testing and I think I might have a theory to the problem. I think the vram modules might be overheating after long periods of running the card close to or at 80°C. After watching Gamers Nexus videos about the XFX 590 and seeing how they point out weak areas in cooling for the vram modules, the fact that having a huge fan blast air into the open side of the case resulting in no crash seems to make more sens now. With so much extra airflow the vram modules don’t overheat as the GPU runs cooler and as a result the PCB would also be cooler.

This card also seems to suffer from tons of heat soak. After a cold boot the card would idle at around 35c, but after a gaming session it would idle at around 55-60c. The card is holding so much heat that it just won’t cool down without shutting down the system or forcing the fans up.

My theory is that gaming adds more work for the vram than a synthetic test would. My PCIe slot also seems to run a little hotter when gaming than with a synthetic test.

Now this is just speculation on my end… I don’t know enough to claim anything here with any certainty. I do know this, the card is not working as it should out of the box. If it gets too hot, it should thermal throttle (which it does), but not crash my entire rig unless it gets to like 90c+… which it never has. I managed 84c with a stress test, but no crash.

Below are the videos in question.

-In the tear-down at 11:50 he starts to talk about the cooler and the vram cooling short comings.
-In the actual review, at 18 min, they point out potential heat issue for the vram ( and I did flip the bios switch to see if it makes a difference, but all it does is ramp up the fans earlier. It will still eventually crash).

Tear-down: https://www.youtube.com/watch?v=wN0q-BiTve4
Review: https://www.youtube.com/watch?v=BBN2KDiw-V8

I wanted to contact XFX, but their contact page wants me to log in and if I click the register button they want all kinds of personal info which I’m not sure I want to give.

Waiting for the retailer I bought the card from to respond to my findings and will have to see what happens then.

1 Like
6

The VRM overheating would explain it.
As the cooler contacts the VRM in a way, fixing it yourself will be rather difficult.

Keep us updated.

Benchmarks tend to take 5 minutes max, a gaming session lasts for hours.