GTX 570 melted, what do I do now?

shottie

Honorable
Oct 24, 2012
3
0
10,510
This weekend I did some case modding to my haf 912. Painted black and put in a side window. I had to rebuild the computer from the ground up but that wasn't a problem.

I was using a 6870 and found a good deal on a GTX 570 so I bought it. I am a forum lurker and read everything I can to become more knowledgeable before I dive into anything. I had read about msi afterburner and I planned to OC the 570. The 570 has been installed since Monday and it was playing fine with no OC at all. Temps were 84c and I figured that was due to dust buildup so I disassembled it, cleaned it out and applied Arctic Silver 5 and let it set for a bit. Plugged it back in last night and played Crysis on everything ultra without a problem, again no OC yet and it dropped temps down to 76c while gaming. Today for lunch I ran home to play with afterburner after researching how to OC properly online.

I tried increasing the core and shader clocks by 50 and ran kombustor. It instantly crashed. I tried increasing them by 20. Instantly crashed. Then I thought I might need to increase my voltage. I believe it was at .988 so I increased it to .989. I ran kombustor again and it instantly crashed. I tried rebooting but the screen never came on. I heard windows load in the background but nothing was appearing on screen.

I opened the case and removed the card and booted up with the CPU graphics. I uninstalled afterburner and kombustor and all nvidia drivers thinking I could just plug it back in and start from scratch. Wrong.

I plugged it in, screwed it down, hit the power switch and saw sparks coming from the right 6-pin connector. Then I saw smoke. Lots of smoke. And intermittent sparks.

In a panicked fit I hit the power button and the computer shut down. I stopped for a second to breathe and then the computer turned itself back on. I freaked out again and hit the power button and ripped the power cord out of the back of the computer.

After the smoke cleared (zing) I unplugged everything. I uninstalled the GPU and it had melted above where the fan is. I just got this card Monday and it is already ruined with what I thought was the most mild of mild OC's.

Now I'm terrified to try turning the computer on. I have no idea what will happen. I'm scared something else will catch on fire. I'm scared that other components are damaged and I have no way of knowing what condition they’re in.

What is the best course of action currently? What steps would you take if this happened to you?
 

shottie

Honorable
Oct 24, 2012
3
0
10,510
I re-installed the 6870 and turned the computer back on to see what would happened and all is well. Lesson learnt.

I guess the question now would be, why did the card melt with such a minimal increase?

Specs:
i5 2500k @4.5 ghz
Corsair 8 gb ddr3 1600
Asus p68 v-le
Corsair force 3 120gb ssd
Seagate 320gb 7200 hdd
Seasonic X650 80+ gold psu
 

TheMadFapper

Honorable
Oct 18, 2012
170
0
10,690


I'm sorry but this was a highly entertaining story. I would guess your card was bad in the first place and unstable. Those were some high temps you saw initially, and would have been a good clue that something wasn't quite right. At least in my opinion.

I'm not sure how you could return the card, perhaps just plead with them and explain that their card was garbage and almost caught your house on fire.
 

shottie

Honorable
Oct 24, 2012
3
0
10,510
It's quite funny in the grand scheme of things. My gf thinks it's hilarious and I had a good laugh as well. If nothing else, just reflecting back on the abject terror I felt when I saw my card melting in front of my eyes, I'm sure I looked hilarious flailing my arms around trying to turn the thing off and get it to stop bursting into flames.

I bought it from my brother in law strictly because he upgraded to a 670. The card was working completely fine until I tried the overclock, and it worked fine for him for a long time up until that point.

I agree, the temps were too high for my liking. When I took the card apart there was no dust to be found. The only questionable area was the amount of thermal paste. It seemed to be excessive. After applying my own with the correct amount I saw an instant decrease to what I felt was a reasonable temperature under load. I used HWMonitor and msi afterburner to monitor temps. I have one 200mm front fan for intake and one 200mm and one 120mm for exhaust in the case, as well as a push/pull with 120mm fans on my hyper 212 evo.

I'm starting to wonder if, in my haste, I didn't fully connect the right 6-pin connector before powering on. Since I uninstalled all drivers and all msi software it shouldn't have remembered the oc or overvoltage that I applied. In fact, I didn't have on the option to remember settings at startup. Logical thinking deduces it must have been user error when reinstalling the gpu. It's quite an embarrassing mistake given how long I've been doing this and replacing parts on my own.

It was definitely unfortunate. Hopefully the 8000 series cards come out in Q1 2013 so I can get away from the 6870 and the ridiculously loud fan!
 

matt_b

Distinguished
Jan 8, 2009
653
0
19,010
I assure you, your temperatures were not an issue. The general rule of thumb is that graphic cards can take 90* C, Do you want to run it that high- no, but it's a statement meant to tell you that they are more resilient than a CPU is.

Definitely contact MSI on this one. It sounds more like what happens when ground and a live wire/connector contact one another, why it didn't do this from the start on day one this is beyond me. No telling what happened as this is one of those "wasn't there, cannot tell over the internet" kind of situations/ Here's the problem now though, you've also removed your heatsink and replaced it with a different manufacture's thermal paste - this voids warranty is just about every manufacturers' opinion regardless of the cause. You'll have to play this one tightly in hopes they don't figure this part out.

Edit, after reading your second post, you should understand that AS5 has flakes of silver in it - it's a very conductive paste!. Hopefully over applying or smearing it on other parts of the PCB wasn't part of the problem. If the second PCIe connector wasn't plugged in, most times your card won't even operate as the sensing connector will tell the card it will not have enough power to operate, so it doesn't (in most cases). If your card was not the case like this, then it would just crash constantly.
 

ram1009

Distinguished
I have only a philosophical response for you. IMHO, over clocking has evolved nothing more than a giant pissing contest. The few FPS you MIGHT add will be the most expensive ever and will add nothing to your enjoyment of any game. You will, however occasionally have wonderful stories to tell your grandchildren, assuming you outlive your folly.
 

Xenharmonic

Honorable
Jul 19, 2012
28
0
10,530


This site should have a "like" button. Pure gold :).
 
IMO there is nothing funny about this, regarding the power connection point sometimes the connection female pins slip backwards in the connector plug not making full contact, under those circumstances it could destroy a perfectly good GPU through a faulty connection.

It's more common with the 4 pin molex connectors than the PCI-E connectors but can happen with any connector at all, it's a good practice once the connection is made to continue to push the wires together for a firm solid connection, just in case one wire did not make a full connection.

Sorry for the loss of your graphics card!

However be thankful you did not loose anything else!
Ryan
 

nacos

Honorable
Mar 24, 2012
301
0
10,780

The 36% fps increase I see around the board with my 7970 overclocked to 1.25ghz is a bit more than a few fps.