I've been searching for weeks and can't find any solution. Either right when I launch a game or shortly after, both monitors go black and for a few seconds. The only error info I can find is windows saying display drivers "stopped and recovered." I've also discovered that it happens only when the GPU is at max clock speed (1405 for the XFX RX 580) and when I overclock or under clock, the crash still happens at whatever I set the clock speed to.
Motherboard: Asus B450m-a
CPU: Ryzen 5 2600
RAM: Corsair Vengeance 16gb 3000mhz
SSD: Samsung 860 EVO 500gb
PSU: EVGA 700b
Occasionally crashes while not gaming as well, even happened while typing this post. I have reinstalled drivers multiple times, messed with wattman settings, reinstalled windows, updated the bios, and RMA'd the card. Nothing makes a difference. I am out of ideas and a little pissed off.
Edit: Still afraid to get my hopes up at this point, but it seems like the issue was with the RAM the whole time. After more troubleshooting, I ended up trying each ram stick by itself. One of them still crashed but then the second one worked fine. I played off and on all night and have yet to experience a crash. I will try an RMA on the second stick, but everything works fine with 8gb of ram for now!
Plan A: Call me crazy.. but I'd look into an RMA for the PSU or just buy a good one. Or.. borrow a buddy's
n.b. although this isn't EVGA's best PSU and EVGA does have a good name ANY PSU can fail or be a problem child.
Plan B: did you run a memtest?
Plan C: Is there nothing else in windows event viewer relating to these issues?
A: This is the second PSU I've tried. Started with a Corsair 450w and thought that the problem was that it was too low, but putting in the 700 made little difference. I kept the 700 anyway though because I later found that AMD recommends at least 550w.
B: I have not run a memtest, what should I use for that?
C: The only other windows event are "DistributedCOM" saying that a program tried to use something it didn't have permission to do. Those usually show up right after the crash logs, but everything I've read about them makes me think I should just ignore that.
I have two 1080p monitor . One is an old hp that has only dvi and vga, and the other is an AOC that has only vga and hdmi, so I am using hdmi for on and dvi for the othe . Ive tried unplugging one monitor, but the same problem occurs on dvi and hdmi alone.
I ran a mem test for about 30 minutes with no errors, I will try again and let it run longer.
I'm running out of ideas. One last thing to try is downclock to ram to 2600. 3000 is an OC setting and once everything is running it probably requires another .5v or something. Don't quote me on that (it's been a while since I messed with ram voltage) and can I assume your currently running the ram with a xmp profile?
XMP is enabled, but the crashing was happening before enabling it. I might have done that on the old psu though, so maybe I'll try disabling it and see what happens.
memtest86 and the windows memory diagnostic gave no errors.
I have one of these cards and can tell you a couple things definitively. Yes it is possible you still have a bad PSU. Not likely though if that 700 is new.
I can also tell you that these cards do not run at default settings for many users, self included. They are factory overclocked and need a higher power ceiling. Unfortunately XFX should but does not account for this in their BIOS.
So game screens will go black or other colors sometimes and you will often still hear the sound going in the background.
You fix this by raising the Power Limit in Radeon Settings / Gaming / Global Wattman - to 50.
Then you will want to use a custom fan curve to stop the card from thermal throttling.
See the picture of my settings below and mimic those. Only use my Power Limit and Fan and Temp curve. My other settings are overclocks and may not work on your card but you can try them if the other settings make things stable first. Good Luck here is the pic: