AMD R9 390 - Gray screen of death or?

[Temuka]

Hi everyone,I have some kind of weird problem,I'm in pc building for over 7 years and this is the first time I have this kind op problem.

2 days ago I updated my R9 390 Nitro amd drivers to latest Adrenalin Edition 18.9.1 and the problem started. Once in a while (after 5 minutes playing,20,50,1 hour or 2) my screen goes gray,I hear sounds but can do nothing,only restart pc from on/off button.

Games this happened are: World of Tanks (I really play only this one) and I also tested in Far Cry 5,same result.

What I have done already:
-reinstalled game
-reinstalled game on ssd
-reinstalled drivers
-installed older drivers (with duu utility)
-reinstalled windows and game itself again on ssd with new windows on it
-bios reset
-I took out PSU and GPU and cleaned them with compressed air and also whole pc also

nothing....

Now I installed MSI afterburner and I'm trying to play with clocks,this are results:
1010/1500 crash (stock speeds)
900/1400 crash
800/1300 crash
750/1200 - Now here is stable testing for 2+ hours,GPU temp is only 55c

My guess is something wrong with vram and gpu needs some kind of soldering inside of chip or something? (I'm not too into gpu insides,lol )

Is this just a coincidence that gpu failed with drivers update? ....

Thanks in advance and sorry for long post

 

[mikeymikec]

Obvious question but I'll ask it anyway - do you use this computer for non-gaming most of the time and it works perfectly well with whatever workload in that situation?

The reason why I ask is that a 750ti I owned once would give intermittent graphics corruptions but also sometimes would hang the computer coming out of sleep mode (maybe 1 in 5 chance), and occasionally video playback on say YouTube would result in graphics corruptions, but if the computer came out of sleep mode OK then chances were when not gaming it would run perfectly fine. Replacing the card fixed it.

 

[Temuka]

Obvious question but I'll ask it anyway - do you use this computer for non-gaming most of the time and it works perfectly well with whatever workload in that situation?

The reason why I ask is that a 750ti I owned once would give intermittent graphics corruptions but also sometimes would hang the computer coming out of sleep mode (maybe 1 in 5 chance), and occasionally video playback on say YouTube would result in graphics corruptions, but if the computer came out of sleep mode OK then chances were when not gaming it would run perfectly fine. Replacing the card fixed it.

Yes I use this pc for browsing,watching movies in full hd,youtube etc... no problems with all this stuff But I never use sleep mode...

 

[Krteq]

Is the same happening also with older driver versions?

 

[Temuka]

Is the same happening also with older driver versions?

I used the latest and the previous one,but not older than that. Should I look for some old drivers?

 

[EXCellR8]

I used the latest and the previous one,but not older than that. Should I look for some old drivers?

he's just curious to see if the problem persists if you roll back to older drivers.

I would do a clean uninstall of the drivers, running DDU in safe mode after normal install, and then abandon all OC/UC profiles for now. After, install the latest graphics drivers for your OS and test again. You can create a new fan profile if you wish, just don't touch any clocks for the time being.

 

[Temuka]

he's just curious to see if the problem persists if you roll back to older drivers.

I would do a clean uninstall of the drivers, running DDU in safe mode after normal install, and then abandon all OC/UC profiles for now. After, install the latest graphics drivers for your OS and test again. You can create a new fan profile if you wish, just don't touch any clocks for the time being.

As I've written in first post,I have done all of that already

 

[EXCellR8]

DDU was vaguely mentioned but the process you performed isn't detailed. DDU is meant to be run after normal uninstall from CP and then in safe mode upon a reboot.

Are you using DVI or HDMI?

 

[Krteq]

If it's still happening with older driver versions I think it's a case of cold joint on BGA bumps below GPU/GDDRs

 

[Temuka]

DDU was vaguely mentioned but the process you performed isn't detailed. DDU is meant to be run after normal uninstall from CP and then in safe mode upon a reboot.

Are you using DVI or HDMI?

I uninstalled amd drivers from programs and features
Then i ran ddu utility after i boot in safe mode
I ran ccleaner also
After that I installed drivers again..
I'm using hdmi..

Probably this is the weirdest coincidence with gpu fail and drivers update procedure...

 

[Temuka]

If it's still happening with older driver versions I think it's a case of cold joint on BGA bumps below GPU/GDDRs

Can it be repaired?
We are stepping to 2019 and I really thought that amd managed to stop problems with drivers/faulty gpus and I tried to support them (because nvidia price politics),but I think it's time to say goodbuy,pay little premium and save own nerves.

 

[Krteq]

This is not AMD fault

If there is an issue with BGA soldering it's an AiBs fault, it's also happening on all IHVs cards/products. AMD is not making cards

Well you can try to bake a card/"re-flow" in an oven. There is a lot of guides how to do that on the net.

 

[mikeymikec]

How old is the card? What was the length of the warranty it was sold with?

 

[Temuka]

How old is the card? What was the length of the warranty it was sold with?

GPU is 2+ years old,I don't think there is any warranty left...

 

[mikeymikec]

GPU is 2+ years old,I don't think there is any warranty left...

My 750ti theoretically had a 3-year warranty, that is if Asus hadn't insisted that the card had to go back through my supplier and if Amazon UK hadn't insisted that it wasn't their problem.

 

[Temuka]

My 750ti theoretically had a 3-year warranty, that is if Asus hadn't insisted that the card had to go back through my supplier and if Amazon UK hadn't insisted that it wasn't their problem.

Yeah.. So basically is there anything I can do now ?

 

[mikeymikec]

Yeah.. So basically is there anything I can do now ?

Check the warranty length, if it's within warranty then talk to the manufacturer and see if they have any more suggestions before beginning the RMA process.

 

[Temuka]

Check the warranty length, if it's within warranty then talk to the manufacturer and see if they have any more suggestions before beginning the RMA process.

I can't check,because I don't have receipt or anything,I ordered it online from other country couple of years ago,all I have is box with all original content {passport,disc etc)

 

[mikeymikec]

I can't check,because I don't have receipt or anything,I ordered it online from other country couple of years ago,all I have is box with all original content {passport,disc etc)

If you've bought it online then you'll have a record you can check there. Next I'd visit the manufacturer's website, put the model number of the card in and see whether it says the warranty there (perhaps in a downloadable pdf), or maybe they have a single warranty statement that covers all their graphics cards. Or contact the manufacturer and give them the order date.

 

[kawi6rr]

I've had issues where I updated an older card with the latest driver before. I rolled back to the driver I was using before the new one and all the issues went away. You can find older drivers on AMD or the AIB's website.

 

[Temuka]

I've had issues where I updated an older card with the latest driver before. I rolled back to the driver I was using before the new one and all the issues went away. You can find older drivers on AMD or the AIB's website.

I'm sorry, but did you read my first post in this thread ?

 

[Temuka]

If you've bought it online then you'll have a record you can check there. Next I'd visit the manufacturer's website, put the model number of the card in and see whether it says the warranty there (perhaps in a downloadable pdf), or maybe they have a single warranty statement that covers all their graphics cards. Or contact the manufacturer and give them the order date.

Ok,I will try that,but i wouldn't bet on that. So I continued testing and card works normally on this clocks 750/1200,stocks are 1010/1500. As I guess it's some kind of hardware issue ?

 

[LTC8K6]

TIM need to be redone on the GPU and/or the ram?

 

[Temuka]

TIM need to be redone on the GPU and/or the ram?

I haven't changed thermal paste on gpu for 2,5 years if you mean that,but what about RAM ?

 

[mikeymikec]

Ok,I will try that,but i wouldn't bet on that. So I continued testing and card works normally on this clocks 750/1200,stocks are 1010/1500. As I guess it's some kind of hardware issue ?

I've been refraining from giving a technical opinion since most of my experience with graphics cards is with regard to low/medium range cards (the r9 380x is the highest end card I've owned relative to its time, which is what I have now). Admittedly it sounds to me like the card is at fault and you've tried everything I can think of.

Contacting the manufacturer even if it's not in warranty may result in more useful suggestions as well.

Running at stock - if you run it for a few minutes with heavy load, what temps do you get?