Three Graphics gone pop? cause?

WhiskyDelta86

Honorable
Mar 19, 2013
4
0
10,510
Hello

Been around toms hardware for some time now, always use it for pc component reviews but Now I am in need of advice.

As the title says I have 3 graphics cards go pop in less than a year,pc specs as follows

Windows 7 64 bit
8GB RAM corsair
GIGABYTE Z77 - D3H Motorboat
Intel i5 3570k OC to 4.2 GHz
Corsair 500r case with 2 extra 140mm roof fans
OCZ 650w X series PSU
Usual dvd drive, 3 hard drives, mouse, keyboard etc
3 Dell 1680 x 1050 monitors

I bought a xfx 7950 graphics card which after installing the card straight away encountered issues. When I powered on the gpu fans would remain spinning at 100% making for a very noisy pc and no signal would go to the monitors, screens would remain in standby. I would have to force a shut down and try again, sometimes it would happen again, others it would boot through....random. when it did boot the fans would constantly be spooling up and down and at idle it ran hot, near 70c. During gaming it got very very hot! Spoke to the retailer and they ram's it with the same 7950.

That worked no issues for around a year with intermittent intense gaming (battlefield 3, splinter cell, bioshock etc) and a lot of 3d modelling, photoshop etc all on 3 screens. Temps were consistently normal 75c ish under load. No issues, one evening shut it down and the next day when I booted I had random, coloured, flashing, spikey lines appearing all over the screen and really bad artifact in. Ran fur mark, temps remained normal with good frame rate but very bad artifact in. Read but replaced with a 7970 as the 7950 was no longer in stock. That was January.

Now since then i have done very little gaming but a lot of 3d modellling with mo issues, temps good and frame rates good. this weekend I discovered war thunder which across eyefinity was epic and had great frame rates. However i noticed (first proper gaming session) temps were quite high, started off around 60c, and steadily rose to the low 90's. I was concerned so stopped and googled but apparently high temps with the 7970 is normal (think a few comments made on this forum) so though ok it's fine, but I did stop with eyefinity and just ran one screen with the intent of reducing the load and the temps. They were hovering around the 80c mark when the pc just shut down without warning. No fps slow downs, bsod, beeping, noises, just shut down. When I rebooted I got coloured lines and artifacting in the bios but when it at temps to get into windows at the point the hardware acceleration would kick in you can hear the gpu shut down completely and the screen remains black. When I rebooted into safe mode (not hardware acclecerated graphics) I got to the desktop but coloured lines were everywhere but could still not get to the desktop in normal mode. Removed the gpu and replaced with a spare 8600 gts and everything was fine.

Now while i believe the retailer will honour a replacement under warranty i am now concerned that something is wrong in the pc. It could be that XFX cards are just crap and i had bad luck but i don't want a fourth.

So could it be the psu? Or maybe the motherboard? The pc has now been reformatted and everything is fine, no hang ups or weird behavior and I have run fur mark and although frame rates are low and power draw is less than the 7950 or 7970 there is no other behaviour to suggest another component is causing the failures. Also all 3 rma are under slightly different circumstances.

Any ideas would be greatly appreciated. Short of spending over £80 for a new psu I am not sure what I can do. Psu is out of warranty.

Regards

Will
 
Solution
If you are not overclocking anything, for 3 cards to go bad in the system I would start with making sure your system has good cooling. Next is the power supply.

I'd do a BIOS update also just in-case, free to do that.
Hi thanks for the response.

I am over clocking the cpu but not bothered with the gpu.

Cooling wise the 500r I'd said to be very good. It has 2 front 120mm fans, a rear 120mm fan, a side 200mm fan and 2 extra 140mm roof fans which I did have in an exhaust set up but have now switched them around so they blow air into the case. With the second 7950 temps were always a reasonable 75c ish the 7970 with the first and only gaming session resulted in temps up to 90c

Regards

Whisky
 
Hey Will,

First off, thank you for writing a detailed post.

It is possible that you got bad GPUs but Its not likely. 3 to go bad in 1 year would make me think that something is up. Have you ran any tests on your PSU to ensure it is supplying stable power?
 


Stop all overclocking while you test things. You want to remove ANY variable aside from the video cards. First thing to do is try the cards in a known good system if you don't want to buy a power supply to test with, although since the cards run fine for a while that may not work for you. Aside from heat issues, the power supply is the most likely to cause the issues you are seeing unless you simply have bad luck with getting good working video cards.
 
I can't speak from experience, but I've seen some horror stories about XFX gpus

Their PSUs seem fine, but for the forum posts I've seen on various sites, they seem to be very unreliable gpus

More recently there's been a collection of reports where the XFX 7950/7970 just explodes after the cooling fails and the vapor chamber just goes boom!
 


I have a friend that went through 2 xfx 7950's last year as well. He has a 850w corsair PSU, so it could very well be the brand of video card. I actually try to stay away from xfx and powercolor myself. For AMD cards I usually would recommend Sapphire first, and then Asus, MSI, and Gigabyte. They all seem to get higher ASIC ratings as well from research I had done, so they are getting better chips overall from what I can tell. I guess that's why the XFX and Powercolor cards are usually noticably cheaper.

 
Sorry...I didn't mean to click solved and now it seems I can't undo it. I was trying to find the quote button to answer his specific question

Anyway....what specific test can I do to verify the psu?

As a precaution I have rearranged my study so now the side vent to the cade is completely exposed and it is raised up off the ground salt desk level with the intent being more air can circulate than it being under the desk. However when gaming a little my 8 600 gts got up to 80c at which point the fan went from 50% to 75%. Nothing to extreme I suppose but considering the relative small power draw of the card it seems reliably high temps.

Regards

Whisky

P's. ..to reiterate this is not solved so can the mods undo that please!
 
Hey,

just to add - from all brands of GPUs I've had through the years the only one to die on me was an XFX so I stay clear of that brand ever since. Maybe just a coincidence though.

P.S. I think I managed to undo the "solved" as you asked. Not sure if it's gonna work though.