Gigabyte 7950 Caught Fire...

[Jmjam]

So I was mining for the past 2 days straight and today I noticed my PC was off. I go to turn it back on and the gpu started to emit a bright orange light... This is the result after I quickly turned it back off,

http://i.imgur.com/pXqUOAo.jpg

 

Bought it January 2013, Tried to input the serial number that was on the GPU on the Gigabyte RMA site but the warranty didn't show.

 

CPU: AMD FX 6300

MOBO: Gigabyte 990fx ud3

PSU: Silverstone ST60F-ES 600W

 

GPU: GIGABYTE GV-R795WF3-3GD

 

What may have caused this to happen? I've seen similar gpu's do the same thing on even more efficient psu's than mine. Was it just a bad card pushed too far? Temps while mining never hit 70C, avg 66C, Intensity was 19 at stock clocks

I've never done an rma before so any help would be appreciated

[TheNinjaNextDor]

I didn't see you were mining, so yes a VRM almost 100% likely overheated from the constant strain.

[helping]

call up support and tell them you were playin' the games and it tried to burn your house down and kill your family 

 

2) send card in as per instruction

3) receive new or "new" card in 1 to 2 weeks. 

[TheSLSAMG]

Core temp wasn't high, but the VRMs were probably strained from the constant power being pushed through them.

 

I guess that's what you get for mining? If you RMA it, don't tell them it was mining. They may not take it back.

[CoolBeans]

It shouldn't be at 70C for 48HR straight

[algoat]

Same thing happened to me on 2 of my Gigabyte 7950s and both within a week of each other.  I don't know if it was a bad batch or what.  I RMA'd them both of them and put the F43 bios on each.  This bios locks the volts to 1.09V rather than the 1.275V on the current bios.  The replacements have been running well since about October.  

[Jmjam]

Same thing happened to me on 2 of my Gigabyte 7950s and both within a week of each other.  I don't know if it was a bad batch or what.  I RMA'd them both of them and put the F43 bios on each.  This bios locks the volts to 1.09V rather than the 1.275V on the current bios.  The replacements have been running well since about October.  

Yeah I figure I should undervolt the next one I get, If those cards were to go poop now would they still accept an RMA even though you flashed the F43 bios?

[algoat]

Yeah I figure I should undervolt the next one I get, If those cards were to go poop now would they still accept an RMA even though you flashed the F43 bios?

 

For this type of hardware issue, they would have to rebuild the board in order to see what type of bios was on it.  They just sent the replacement immediately after they received it, so I'm sure they just gave it the once over and approved it.

 

edit:  My current cards are overclocked to 1130/1250 the memory is stock.  During the winter they ran at about 60C, and about 69C (just glancing over at it right now.)

[Jmjam]

For this type of hardware issue, they would have to rebuild the board in order to see what type of bios was on it.  They just sent the replacement immediately after they received it, so I'm sure they just gave it the once over and approved it.

Read more on this and appears to be a very common issue, might as well sell the new one and get a 280x?

[algoat]

Read more on this and appears to be a very common issue, might as well sell the new one and get a 280x?

 

Yes I considered that as well after looking at the different experiences people were having.  But one thing I didn't mention was that a 3rd separate GB 7950 that I purchased months after recently did the exact same thing.  After a long wait, they said that they didn't have any more 7950s in stock and offered me a GB 670 as a replacement.  I declined that offer and said that I would wait.  So they asked if I wanted a GB 7970 instead, which I agreed to, and it is on its way soon.

[flibberdipper]

Yup, one of the VRM's went boom. RMA it and don't tell them it was mining.

[Jmjam]

Yes I considered that as well after looking at the different experiences people were having.  But one thing I didn't mention was that a 3rd separate GB 7950 that I purchased months after recently did the exact same thing.  After a long wait, they said that they didn't have any more 7950s in stock and offered me a GB 670 as a replacement.  I declined that offer and said that I would wait.  So they asked if I wanted a GB 7970 instead, which I agreed to, and it is on its way soon.

well I guess free upgrade for the wait! hope they do something similar with my RMA

[DarrenP]

A moment of silence for the fallen video cards to mining. *puts hand on chest* 

[SkilledRebuilds]

Just because the core runs at under 70*c all the time doesnt mean the VRM's are not increasing past 99*c and up to if not past 115*c, if thats also constant.... then yeah.

[defploit]

Well if you wanna do constant heavy duty ops on your pc... id suggest you get a workstation grade mainboard . 

[d33g33]

I bet they are getting heaps of these!

 

Do they actually state anywhere that they wont RMA the card if its used for mining because if they don't and they knock you back you could throw the book at them.

 

To be on the safe side though and actually get a no fuss replacement might be worth doing as others have suggested.

[Mateyyy]

That's why I never bought Gigabyte cards in the past.

I heard many people saying, even here in Romania, that their Gigabyte cards burned down. 

You can see here some examples, you're not the only one: https://www.dropbox.com/s/qx4z1jtn6xzp6lg/2013-04-25%2014.06.57.jpg

http://i.imgur.com/8vQ4hi9.jpg

http://i.imgur.com/S1P9d88.jpg

Core temps were fine, the VRM was overheating and melted. You don't really have too much to do. You could try and send it for a warranty repair, hopefully your card still has warranty. 

[Eatmyshorts]

It was never designed to run under that kind of load for that long so the only problem is your expectation of the product. For specialised work the cheaper option is hardly ever the better option.