GPU shutting down in the middle of gaming

[Kaizenx]

I have an AMD RX580 with 8GB of RAM.

After I cleaned the GPU with a can of compressed air, the GPU shuts down during moments of intense gaming, games like Burnout Paradise Remastered or HOMM3 will run fine forever, or at least as long as I want to game.

But something newer/intense like Prey will shut the GPU down after a period of gaming, and sometimes on start.

I have the following GPU-Z log attached, the GPU shutdown occurs at line 1776

Do I need a new GPU?

Moderators please move this if this is the wrong place to post.

 

[evernessince]

The temperatures are completely fine. What's odd is that the GPU clock isn't even hitting base clock, let alone boost clock.

What power supply does your system have?

 

[Regeneration]

Check your power supply.

 

[X71200]

Your PSU likely won't prevent you from reaching your base clock, however it might indeed be the cause of shutdowns. Compressed air won't hurt either. If you could write in which PSU you got, that might be more helpful.

 

[Kaizenx]

I have a be quiet! Straight Power 10 700W ATX 12V 80 Plus Gold.

FYI I have never overclocked anything, this is all factory settings.

 

[X71200]

Sounds like the issue lays somewhere else, try capping your FPS maybe. Also, what RX 580 is it and what has the GPU gone through in past?

 

[Kaizenx]

So I just tried the OCCTP app and ran the power test, the GPU shutdown immediately XD, is that conclusive proof that it is the PSU?

 

[X71200]

No, definitely not. Those FSP units are not bad. Unless it's gone through what a modified Peterbilt trash truck of some 2000 HP doing donuts, I'm sure the PSU isn't the culprit. The Teapo caps age quicker than Jap capacitors, which is what your PSU comes with. However, given the release date of your PSU and the system specs, the problem is likely something else. I'm just wildly guessing since you have yet to tell us what RX 580 it is. You know, mining and the whatnot.

 

[Kaizenx]

I have a ASUS ROG STRIX RX 580 GAMING OC.

I mined with it once for a month about 8 hours a day, but that was years ago.

I used to play 3D-es fps a lot (500 hours in a year), but have stopped for at least a year and nowadays get up to some HOMM3 every now and then, mostly browsing and videos.

I also recently patched to the latest WHQL drivers, 20.45.01.18-201113a-361132C-RadeonSoftwareAdrenalin2020

 

[Regeneration]

Try to underclock by 50 MHz.

 

[DeathtoGnomes]

Complete system specs would help here.

https://www.techpowerup.com/forums/account/specs

 

[Kaizenx]

Complete system specs would help here.

https://www.techpowerup.com/forums/account/specs

Updated.

@Regeneration I did the underclocking thing, so far more stable, no crashes (if that is what it is) yet.

 

[DeathtoGnomes]

Good, you have plenty of headroom with that PSU, short of total failure, there shouldn't be any problems with it.

 

[Kaizenx]

So its back to the GPU, I underclocked it and now its stable, what does that say? The GPU is worn out? Drivers? I recently updated the drivers, before this it was stable.

 

[TheoneandonlyMrK]

I would look to see if the fans are working, if they are strip the card down and re apply thermal paste and check the thermal pads on vrm and memory and possibly replace those.

 

[DeathtoGnomes]

So its back to the GPU, I underclocked it and now its stable, what does that say? The GPU is worn out? Drivers? I recently updated the drivers, before this it was stable.

I would look to see if the fans are working, if they are strip the card down and re apply thermal paste and check the thermal pads on vrm and memory and possibly replace those.

If one of the fans is not working, you can replace it, if both are working you can do this ^^. There have been others with the same problem that -50 underclocking has solved, you wont see that much of a performance loss if you decided to keep it that way.

 

[Hyderz]

did you reinstall the drivers after you taken it out to clean?

 

[Countryside]

So its back to the GPU, I underclocked it and now its stable, what does that say? The GPU is worn out? Drivers? I recently updated the drivers, before this it was stable.

I recommend using DDU to clean out the gpu draivers and the install again, it's unlikely that psu is at fault here you have more than enough power i run my rx580 oc on 500W psu with no problems, that Be quite is a good psu only thing i dont like about it is that it uses 85C Chinese capacitors.

 

[Kaizenx]

Thanks for the suggestions, will need to order in thermal paste to strip the card down, as I believe the fans are working.

 

[jboydgolfer]

if the shutdowns are under heavy load, & cause the PC to straight up shutdown & restart with no warning or message on screen or in event viewer, its likely to be the PSU.

 

[agent_x007]

Shutdown can also be a result of overheating/overtaxed VRM (too much vdroop under instant heavy load). Can happen in both CPU and GPU VRMs.
Also, you cleaned card with compressed air :
Q : How sure are you of dust/dirt/grime NOT going between MOSFETs or under caps/inductors during your cleaning ?
Did you clean slot as well ?

 

[evernessince]

I would look to see if the fans are working, if they are strip the card down and re apply thermal paste and check the thermal pads on vrm and memory and possibly replace those.

His log indicates that a re-application of thermal paste is not needed (GPU core temps are well within the safe zone). VRM and memory could be a potential issue as those are not represented in the log.

So its back to the GPU, I underclocked it and now its stable, what does that say? The GPU is worn out? Drivers? I recently updated the drivers, before this it was stable.

By underclock you mean just the core clock and not the memory clock correct? Looking at your GPU-Z logs it appears your GPU only has issues hitting certain core clocks. If you didn't need to touch the memory clock this may rule out GPU memory as a potential issue.

Also need some clarity on one of your previous statements:

"After I cleaned the GPU with a can of compressed air, the GPU shuts down during moments of intense gaming, games like Burnout Paradise Remastered or HOMM3 will run fine forever, or at least as long as I want to game."

Did this GPU work fine prior to cleaning with said can of air?

 

[TheoneandonlyMrK]

His log indicates that a re-application of thermal paste is not needed (GPU core temps are well within the safe zone). VRM and memory could be a potential issue as those are not represented in the log.


By underclock you mean just the core clock and not the memory clock correct? Looking at your GPU-Z logs it appears your GPU only has issues hitting certain core clocks. If you didn't need to touch the memory clock this may rule out GPU memory as a potential issue.

Also need some clarity on one of your previous statements:

"After I cleaned the GPU with a can of compressed air, the GPU shuts down during moments of intense gaming, games like Burnout Paradise Remastered or HOMM3 will run fine forever, or at least as long as I want to game."

Did this GPU work fine prior to cleaning with said can of air?

You take it apart, you re Tim, and I agree on the vrm pads, and it can be an unrepresented thermocouple temp kicking it off because of bubble or crack this isn't hotspot temp.

 

[Kaizenx]

Thanks guys for all the help, I took apart the GPU, blew it clean, reapplied thermal paste, seems ok now. Managed to play Prey at maxed out settings without any issues.

Sorry for taking so long to reply, the thermal paste took a while to arrive.