Bitcoin Forum

I had the 7970 in a rig with two other 7970s. Last night at some point it the rig shut off.
This morning the computer wouldn't boot. I thought the psu went wonky.

Unplugged. Replugged. Etc. Determined the Gigabyte was causing the problem.
Moved the card to an empty machine. Set it up. Turned it on. Flames and smoke and crackles!
What a way to start the morning.  :o

Card was only 15 days old so plenty of warranty/return actions available. Hooray!

Anyone care to venture a guess from looking at the photo?
http://i1273.photobucket.com/albums/y416/stumbling-orangutan/IMG_1264_zpscb4b4dce.jpg (http://i1273.photobucket.com/albums/y416/stumbling-orangutan/IMG_1264_zpscb4b4dce.jpg)

http://i1273.photobucket.com/albums/y416/stumbling-orangutan/IMG_1264_zpscb4b4dce.jpg


Gigabyte AMD Radeon HD 7970 OC 3GB GDDR5 DVI-I/HDMI/2x Mini-Displayport PCI-E 3.0 Graphics Card GV-R797OC-3GD

All in good fun I say.

What temperature were u running at?  I have mine running at 100c and it hasn't burned up in over 3 months!  I hope it doesn't wait til 2 years..heheh

75C- 77C. Open crate rig. I don't think it was a temperature issue.
Maybe the voltage regulator heatsink (I think that's what that is.) got a little too close to the board and shorted.

I suspect it is a random event and not a widespread problem.

Perhaps a assembly defect... It's why warranty exists. Good luck for your future new 7970 card...

Was this on a powered riser?

Regular riser.
Everything had been working fine.

powered riser this, powered riser that... everything thinks the solution or the problem is weither or not the riser is powered or not. It's starting to piss me off.

Where was this riser manufactured?

China, Taiwan (another part of China, that has its own special name), Singapore, or Indonesia?

I'm pretty sure it was a guy named Ralph. Riser Ralph they call him. From a little village outside of Hong Kong. He's known locally as an artisan when it comes to risers.

Lol here is a guy good for a laugh.   How much does Riser Raplh charge?

I guess it's just a defect product and successfully bypass their QA
or maybe a bug walked in and short the circuit

That looks like one of your VRMs blew up. That's not the actual GPU core, but actually farther towards the end of the board, correct?

That doesn't really surprise me. These Gigabyte cards are known for running the VRMs really hot, as there's only a TINY little heatsink on the VRMs, and the air that cools that tiny little heatsink has already passed through the main heatsink and warmed up a bit. I had to struggle with my Gigabyte 7970 to keep the VRM temps below 100C, and this included undervolting, high fan speeds, stick on heatsinks on the back, and a fan blowing on those extra heatsinks. See here: https://bitcointalk.org/index.php?topic=102167.5

vapor x are way better

my 7950(1100/1250) with 30.7 ambient temperature, is sitting at 70° with 50% fan, vrm at 79°

Well a 7950's VRMs naturally run cooler than a 7970's VRMs, so that's kinda pointless in this thread.

After putting new thermal paste on my GPU, VRMs, and VRAM, and cleaning the dust out of my heatsink, my temps are now 75C on the core and 86C on VRM1, when overclocked to 1150 MHz at 1.25 V on the core and 1750 MHz at 1.6 V (default voltage) on the memory. Before, they were 85C on the core and 91C on VRM1, when overclocked to 1125 MHz at 1.2 V on the core and 1750 MHz at 1.6 V (default voltage) on the memory. VRM2 remains a nonissue. Ambient temperature is 25C. This is a reference 7970 using the stock cooler, and the fan is at full blast.

In the future, always replace the thermal paste on your GPU, VRMs, and VRAM, and make sure your heatsink is dust-free. It'll keep your temps down.

The area of the burn is to the right of the silver heat sink (I think that is where the voltage regulators are). I vote for a fluke component failure or a short circuit involving the heat sink.

In any case, it was neat and I am glad I was there when it happened. ;)

http://i1273.photobucket.com/albums/y416/stumbling-orangutan/cooler3_zps32bd9385.jpg (http://i1273.photobucket.com/albums/y416/stumbling-orangutan/cooler3_zps32bd9385.jpg)
(not my exact card but layout is the same)

http://i1273.photobucket.com/albums/y416/stumbling-orangutan/cooler3_zps32bd9385.jpg

Ah ok. Your first pic is a lil dim, and not a good angle. You wanna take the heatsink off, and take a better pic? Yes, you can usually take the heatsink off and repaste it without voiding your warranty.

you mining litecoin or bitcoin ? Have card like that in one of my miners, and scared it is going to catch fire too ! have not seen gpu burn like that before.

My curiosity was outweighed by my desire for a refund so the card is already on its way.
This mystery will have to be like Sasquatch or Nessie.

I run that card mining ltc 1045/1500, stock volts, ~65c for 2 months @~650khash

Runs alot cooler and faster than the XFX DD I had. No problems... Yet!

Hi did you ever find out what caused this? I had exactly the same happen last night on the same card. I have 4 of these, is there any way i can stop this happening on the others?

Thanks
Max

Same thing here..

Guess I will steer clear the Gigabytes in the future.
Card was undervolted and running 76C core temp.
It has been mining for a year, and on a powered riser.


http://i44.tinypic.com/2u4o48h.jpg

The magic smoke escaped!
http://i39.tinypic.com/2e5pm5k.jpg

Great... another case here.  :( It seems to be happened to Gigabyte card more often than other. I guess I will just sell off mine ASAP. Worried about it a month ago when found out a lot of similar cases. Now I am just worrying it more. It does not worth the whole damn house being burned to crisp just for a damn faulty card. lol...

In the last image I can see balls of melted solder.
This is an indication the temperature reached 180 and 190 °C (360 and 370 °F) at some point.
The thing being on fire is a temperature issue! Get a thermometer gun. ;)

The fan directly above the vrm heatsink may be the culprit.
When I turn the fans by hand, there is significant resistance in that one bearing.
Also tried hooking them up to a psu, and the one over the vrm's did spin at a much lower rpm.
If i stopped it with a finger while running, it would only sometimes start by itself.

this is why i have a temp gun and run them slower than most and stable. hardly any hw errors and a few rejects. a 100mh/s gain on each card isnt worth it imo, because if one goes out and you have an empty slot for a week or 2 you prolly take more than what that gain has given you.

Simply RMA this card and install a new one, received from Gigabyte!

I lost one about a month ago. Sent it in for RMA. They could not repair. They said they would send me a check if I sent an invoice.

I hope to get a check this week. Have two survivors left in the rig, mining litecoin. Anyone know if the r9 series play well with the 7970's?

I sent RMA request for my dead 7970 yesterday.. I hope they fix or replace it for the same card.

R9 and 7970 should play well together, but you need to a) Update drivers and b) Edit scrypt bat file to change mining settings to add a new R9 card

I received a notice from USPS that Gigabyte has sent me a package. Hmm?  They said they were sending a check because there were not anymore 7970's in stock. If they are sending me a card I hope it hashes at least as well as the 7970.

I will update when package arrives.

Regards.

Update:

The package they sent to me was an envelope with a usps tracking number on it containing the reimbursement check.

Regards,

Yea it could have been the voltage problem. Your VRAMs must be running too hot to control the voltage, hence leading to this.  :-\

i have the same problem.

black smoke and VGA gone away :)

http://shrani.si/f/3G/5t/1ihKXdQl/unnamed.jpg