Bitcoin Forum

I know the topic name sounds insane but seriously I was just setting up my forth rig and I had a 7950 that I had tested in one of my test machines. The machine had been off for about a week now and the GPU was just sitting there. When I got the GPU out of the machine the image below is what I found... I have NEVER seen anything like this? Any idea what the hell caused this ? When I got the card out of the machine I looked at it and it was completely twisted because the top plastic case piece was still screwed on so the card was twisted to all hell also. It appears that the copper heatpipe/sink exploded like a balloon but I have never in my life seen something like this happen.

https://i.imgur.com/cZXcuNc.jpg

Heh, if it's new then it looks like an rma is in order.

Brand new lol but I am curious as to what the hell would even cause that.... let alone what it is that I am looking at lol

Take more pics and post them in a smaller size. I can't even tell what I'm looking at. Where's the other half of the heatsink? Is it a Gigabyte card?

That is pretty rare.  The heatpipes contain water and should be good past 120C degrees.  I would imagine multiple failures would be required for the heatpipe to burst such as:

1) fan dies
2) high temp shutoff fails
3) temp passes 130C
4) heatpipe bursts

Weird. Your vapor chamber seems to have expanded. Maybe it was a manufacturing defect? The wrong type (or too much) of fluid was put in the vapor chamber and pressure became higher than it was designed for?

Thank you for your reply...

What I was beyond confused about is that the computer had been off... The room itself had gotten to about 95 degrees with the other machines but that specific one had been off. I didn't even know that the heat pipe contains water so I guess you learn something new everyday eh?

This could be a factor as well.  Fan failure combined with too much fluid.

Ive got 14 sapphire cards using that stock cooler, guess it's time to find a replacement.

Almost certainly a manufacturing defect. I would imagine just showing them the card will get you through the RMA process and on your way to a new card.

He was running the card 100% load 24/7, how does that warrant an RMA? No card that it used for bitcoin mining should be entitled for a "free" replacement. How many lies will you tell them to get a new free card?

Did they say you could not use it for that? Fact AMD encourage you to use it for such by providing OpenCL.
If anyone gives you any RMA trouble mention it's a fire hazard and maybe you should send it to a safty regulator so they can issue a product recall.
They'll soon send you another one to get the evidence out of your hands.  ;)

If he tells them what he was using the card for, the clocks, the temps, the ambient room temp, the psu and what else was hooked to the psu, how he had the card attached to the mobo(risered or not/powered or unpowered) and any other little details and they still offer a replacement then I will take back what I said.

Let's see the warranty info first.

Usually it doesn't say anything about ambient temps and other hardware (except PSU power) and might only mention overclocking or something similar but some manufacturers even offer warranty for overclocking (but usually without hardware modding, BIOS flashing or increasing voltage).

What happened in the picture should never occur to a non-defective product.

edit: you work for some GPU manufacturer or sell GPUs yourself?

He is probably from the IBM hard drive section that tried to claim their old "deathstar" series was only meant to be used a maximum of 5h a day.

Bullshit, a card is made exactly for that: TO BE USED

Or do you buy cards as ornaments?

People use GPUs in simulations. Heavy number crunching. I doubt there are different warranties for workstation cards and gaming cards.

Holy shit mother of god!!!
I did not know that heatpipes contain water...

This is redicilous, how the frick can such thing even happen?
I think that it is quite an safety hazart, god knows what happens if you got 4 cards on one MB and a cooler just plain "explodes", it will most likely fuck more cards up becouse of the fact there is no room between cards.

Is this the first of a kind failure documented in this forum? If so, then it seems to be very unlikely to happen to someone else.

that's the first one I saw.....holy.......
I am still running 3 references without the plastic cover and a 20' fan behind them 24*7, I have to check them more often.

which brand is this?

That would be entitled to an RMA.  Pushing it past 100% (overclocking/overvolting) breaks the terms of the warranty.  Using it right up to the terms 100% does not. 

That's the first time I've ever seen heatpipes blow out. I work as a computer tech and see lots of nasty problems, but I've never seen anything like this.

Normal heatpipes can't really expand the way a vapor-chamber can, they usually just crack at either end when pressure gets to high and whatever medium was inside them leaks out.

so the medium leaks out on your motherboard, nasty  ::)

I'll take two eggs over-easy please...

Again, AMD ecourage overclocking in the Catalyst driver, see Performance > AMD Overdrive.
Not volt moding though. (I wouldn't mention that)  ;)

Will people still please stop saying heatpipes contain water *sigh*

Overclocking doesn't with GPUs, overvolting DOES.

Again, why do you think overclocking is readily available in Catalyst, with no warnings about warranty?

Wow I never heard or seen this happen before.

Was it a Sapphire 7950 Vapour-X card?

I've seen one other case of this. The heatsink/card had been bent ever so slightly, but enough to segregate the vapour chamber into two areas. Most of the heat was being applied to one area, which lead to it ballooning. No idea how the bend was stronger than the material though.