Random drivers crashes - HD7950

[gator37]

Hey,

I have HD7950. Obivously it's bit of a veteran right now (about 4 years iirc) and I'm experiencing random and weird drivers crashes (display driver stopped responding and has recovered). Hopefully it won't be too long, I'll try to keep it as short as I can.

Crashes first appeared around March/April. It started out of nothing, without drivers update (I'Ve been running on those drivers for about 3 or 4 weeks at that time). It was happening only while I was playing one specific mission in XCOM Enemy Unknown. Because of that I thought that it's some compability issue or whatnot and since I finished the game back in the day already I basically gave it no attention and just stopped playing.

After about 1 month it started happening in CS:GO. I couldn't find a way to replicate the crash. Sometimes it would crash after 20 mins, sometimes after 2 hours and sometimes I could play for hours without a single crash. I spent good 20 hours trying to replicate the crash and had no luck with finding a pattern.

Since then no other game started having same problem. I could play stuff like Witcher 3, DCS, BF4 or 1 or Arma 3, running benchmarks and stress tests or working in Cinema4D etc. and I wouldn't get a single drivers crash. I installed Xcom back again to test it and it still suffers from the crashes.

However on friday I started getting crashes even when I'm not running games or programs. After a while I found out that watching streams at twitch.tv at source quality was causing it. After circa 2 mins of watching (full screen) I start getting green bars/artifacts flashing at the top of the screen, it happens like every 20-40 seconds. After couple more minutes (2-5) the drivers crash. Sometimes it just happens straight away and sometimes it happens only when I alt-tab to other window or try to minimaze the window.

PC specs:
I3 6100
HD7950 Windforce 3 3GB
8GB 2133mhz DDR4
MSI H110M PRO-VH
Windows 10

What I tried (throughout the last 4-5 months, it would probably be longer list but I can't remember it all)
- reinstalling drivers (both manually and with stuff like DDU)
- trying old drivers
- most, if not all, suggested operations at AMD troubleshooting site (bar the windows reinstall)
- trying beta drivers
- messing in registers (mostly following advices on forums)
- cleaning the hell out of my PC
- tried different PSU (althought I doubted that would be the problem)
- reinstalling affected games
- messing around with GPU settings

None of that helped nor made it worse.

I also ran Furtest many times at various settings even for more than an hour. No crash at all.

Also I feel like it's good to point out that I don't see any loss of performance.


Now, I haven't tried windows reinstall but that's probably the only thing I can think of that I haven't tried. The reason is that if I wanted to reinstall the Windows it would probably be complicated (I got legit version but it would be tricky, I can elaborate if needed)


My question to you is:

Is this more likely to be HW or SW problem? As I said the GPU is kind of veteran so I wouldn't be surprised if it was on it's way out, but it seems like weird signs of dying imo.

Any idea what I could try without reinstalling the windows? ( I'd like to keep reinstall as a very last resort, same for buying new GPU)

 

[Georgivs]

Looks like a hardware problem.
You can try to turn up the power limit in drivers, but not sure will it help.
In the beginning of production of 7000 series was a problem of the black screen/driver crash, that was caused by the power management plan in vga-bios. There were released new bios versions for some cards.
But the appearance of the problem after many years of using suggests me that is a problem of card oldness.
I'd glad to make a mistake. Maybe another user has a better solution for you.

 

[Shaun o]

Well i know that for some cards of the older Ati series of gpu`s back dating to driver 16.2 fixed the problem.
the latest drivers have a power tweak for the power problem that the new RX 480 range of cards had using to much power draw from the pci-e card slot of motherboards.

If you have not cleaned the card at all over the last four years of use gator37.
And the card you have is the Ati reference design that uses a single fan of the blower type of fan on the card.

You may want to take the card out of your system, you should remove the fan shroud that covers the card all the way along the top side of the card and with the circular hole for the blower type fan.

Between the blower fan of the card and the front edge of the cooling solution fitted a thick matte of dust can block the air flow of the fan pushing cold air down the fins of the cooling solution .
Due to a lack of air flow the card will begin to crash and freeze or the video driver for the card will often stop working in windows due to the Gpu of the card getting to hot because of the dust blockage.

Clear the dust matte build up away by removing the fan, and card shroud of the 7950 card.
And the card will stop crashing when playing games.

It often happens when the Gpu on the card gets too hot, the video driver is designed to self terminate if the gpu is getting to hot due to a lack of cold air cooling it, to prevent damage to the Gpu chip by it overheating to much gator37

So you must check for this.
Click on the link bellow to see a picture of a 7950 reference cooler of the card with the blower type fan.
The dust gets caught and clogs the front edge of the fins of the cooling solution in most cases, and where you need to look.
https://www.google.co.uk/search?q=7950+reference+card&espv=2&biw=2844&bih=1491&source=lnms&tbm=isch&sa=X&ved=0ahUKEwiFr5fbopvPAhULLMAKHcPjBdYQ_AUIBigB&dpr=0.9#imgrc=ZbN8mbSZi4onOM%3A

 

[gator37]

If you have not cleaned the card at all over the last four years of use gator37.
And the card you have is the Ati reference design that uses a single fan of the blower type of fan on the card.

You may want to take the card out of your system, you should remove the fan shroud that covers the card all the way along the top side of the card and with the circular hole for the blower type fan.

Between the blower fan of the card and the front edge of the cooling solution fitted a thick matte of dust can block the air flow of the fan pushing cold air down the fins of the cooling solution .
Due to a lack of air flow the card will begin to crash and freeze or the video driver for the card will often stop working in windows due to the Gpu of the card getting to hot because of the dust blockage.

Clear the dust matte build up away by removing the fan, and card shroud of the 7950 card.
And the card will stop crashing when playing games.

It often happens when the Gpu on the card gets too hot, the video driver is designed to self terminate if the gpu is getting to hot due to a lack of cold air cooling it, to prevent damage to the Gpu chip by it overheating to much gator37

Thanks for the advice.
However the card had been cleaned couple of times during it's lifespan + the whole card was completely cleaned around New year 2016. Since then I did just one more cleaning around June. But I will have a look at it when I get home.

Also if it was the temperature problem wouldn't the crash happen everytime the card reached "Critical" temp? Because I am pretty sure that the card reached much higher temperatures in Furtest than it reaches in CS:GO or when watching streams.

I remember checking temperature after crash couple of months ago and if I remember correctly it was below 60°C, while yesterday during 1440p Furtest it went up to 75 and was working just fine.