[RomeSC]

Hey all. I don't know enough about GPU's to answer my own question so I'm posting here.

I own a Galaxy GTX 570 that I have had for about 4 days now.

I was getting stupid high temps around 95C in heaven until I celaned my fan filters and case, and temps dropped to 75C.

I started playing DA2 maxed out last night and all of a sudden I got a full red screen. nothing worked to get out of the application and I had to force a power off.

I noticed after it started up and I quickly got into MSI afterburner (which was not running before I opened the game before) that the temps were at 60C and declining, and as soon as MSI AB loaded the fans jumped up, noticably. I believe, for whatever reason, the fans were locked at 40% during my DA2 session and the card overheated.

Ever since the overheat (which was at 900/1925 @1.088v) the card won't overclock right anymore. I get HUNDREDS of errors in OCCT within 5 minutes. I dropped clocks back to 850 @1.075 and I still get errors in OCCT when I didn't before. I also cannot complete a heaven bench at 875mhz @1.088 when previously I was completeing them at 900 at the same voltage.

Can overheating one time cause permanent problems with the GPU? Or is soemething worse happening here? How can I figure it out? You know I sent my UPC off the box two hours before this happened, so no returning to NewEgg for me...

Please help!

 

[AblueXKRS]

Yes, you can damage a card without killing it in the fashion you described.

Even if you can't return it to the Egg, you should be able to get a replacement from Galaxy if it dies outright. However, slightly diminished overclocking potential is not a valid reason fro a replacement.

 

[Defoler]

Well 570s are notorious to die from OCing.

If you have any issue with the OCing, go back to stock and start testing the card using a few benchmarks, see if its fine in stock.

If stock is fine, I suggest back off with the amount of volt you are putting the card on.
That and the fact that it overheats, might kill the card too early.

Also review your case cooling. If you got it down by 20c because of cleaning filters, make sure you don't have other obstacles which hinders the cooling.

 

[RomeSC]

Quote:

Originally Posted by Defoler;13037846
Well 570s are notorious to die from OCing.

If you have any issue with the OCing, go back to stock and start testing the card using a few benchmarks, see if its fine in stock.

If stock is fine, I suggest back off with the amount of volt you are putting the card on.
That and the fact that it overheats, might kill the card too early.

Also review your case cooling. If you got it down by 20c because of cleaning filters, make sure you don't have other obstacles which hinders the cooling.

No the case if fine now. Full cable management and dust free. Apparently the 95C temp in heaven didnt bother the card much, because it OCd fine after that. Its only been screwy since the red screen lock. I assume if I blew a VRM somehow the card would not work at all?

Even at stock I am getting crashes in games that previously had never crashed (DA2). Sigh... is it normal of afterburner to not let the fan increase unless the program is running in the background??

 

[Defoler]

Quote:

Originally Posted by RomeSC;13037985
Even at stock I am getting crashes in games that previously had never crashed (DA2). Sigh... is it normal of afterburner to not let the fan increase unless the program is running in the background??

MSI AB shouldn't prevent the fans from increasing unless you setup a fan profile wrong, or you accidentally set the fan speed to manual and kept it at 40%.
If it was on auto, the GPU drivers are the ones setting the fan speed by the default profile, which usually means the fans are at 40-50% unless the card really over heats.

If you now get crashes on stock, could be that something got burned. I suggest you take the card out and examine it (don't open, just go over the back of the PCB) and see if there are any marks.

 

[LivingChampion]

Maybe you just got one of those Gray Screen of Death except it was red instead. Was it overclocked when this happened? Try downclocking/voltage tweaking a bit.

Edit: So it happened when it was at stock too, somethings up with that card. RMA might be in order.

 

[RomeSC]

Quote:

Originally Posted by LivingChampion;13038574
Maybe you just got one of those Gray Screen of Death except it was red instead. Was it overclocked when this happened? Try downclocking/voltage tweaking a bit.

Edit: So it happened when it was at stock too, somethings up with that card. RMA might be in order.

The red screen thing only happened the one time when the fans weren't increasing and the whole system locked up.

On stock I am getting CTD crashes, not the red death screen thing. But games are crashing that never previously crashed namely DA2 and BFBC2.

When I get home I will check the back of the PCB. Can't believe I cut the UPC off and mailed it RIGHT before that happened... so much easier to return to newegg. Maybe Galaxy has an advanced RMA where they mail out the new one first? (yeah right...).

 

[jonjryjo]

Make sure the card is seated right. Last time I was playing I started getting artifacts and crashing in Just Cause 2, even at stock. I opened up my case, took the GPU out, and put it back in, making sure the bracket was tight. Now, it works just fine. Granted I have a very heavy full copper waterblock on it, but it's still possible.

 

[Booty Warrior]

The saga continues!

 

[PCSarge]

ati made bad drivers and good cards.
now ati has good drivers and good cards

and nvidia makes cards that die if you OC them.

maybe nvidia hired the guys who used to make atis crap drivers to make thier card designs?

 

[whipgeez]

i hope bassplayer doesn't have to add another to the list of dead 570's

when i get my replacement it's either staying at stock or being sold for a gigabyte 480 soc

 

[Fletcherea]

That much trouble in less than a week = me getting the rma in full swing, and ridding myself of what ifs, why, and headaches.
Optimism might leave you with a dead card come a week after resellers rma/warranty time. I'm done with patient now I think lol =D

 

[RomeSC]

Quote:

Originally Posted by Fletcherea That much trouble in less than a week = me getting the rma in full swing, and ridding myself of what ifs, why, and headaches. Optimism might leave you with a dead card come a week after resellers rma/warranty time. I'm done with patient now I think lol =D

If only resellers RMA was an option, I would have done it last night. I cut the UPC off the card's box and mailed it in for the rebate only a couple hours before I started having this issue.

An RMA now goes through Galaxy, who does not cross-ship. If I send it back I'm without a card for 2 weeks. I don't even know what to do in the interim, I sold my 2x 460's, 5770 and 8800GT I had around. And my mobo doesnt have onboard video. Great...

To the person who told me to look at the pcb for marks... what am I supposed to be looking for?

 

[whipgeez]

any sign of burning would be the obvious thing to look for

 

[Freiya]

Quote:

Originally Posted by RomeSC If only resellers RMA was an option, I would have done it last night. I cut the UPC off the card's box and mailed it in for the rebate only a couple hours before I started having this issue. An RMA now goes through Galaxy, who does not cross-ship. If I send it back I'm without a card for 2 weeks. I don't even know what to do in the interim, I sold my 2x 460's, 5770 and 8800GT I had around. And my mobo doesnt have onboard video. Great... To the person who told me to look at the pcb for marks... what am I supposed to be looking for?

I feel your pain.

 

[Shredicus]

When my 570 fried, I took off the heatsink and sure enough right where the 4 little square black VRMs were, one was melted to ****. You will smell a pretty distinct burnt rubberlike smell if you fry a VRM too.

 

[RomeSC]

Quote:

Originally Posted by Shredicus When my 570 fried, I took off the heatsink and sure enough right where the 4 little square black VRMs were, one was melted to ****. You will smell a pretty distinct burnt rubberlike smell if you fry a VRM too.

its my understanding that if you fry a vrm the card will not work at all. Am I incorrect? I just finished a heaven bench at 85[email protected] and played bfbc2 for an hour at the same settings. I doubt that would work with a fried VRM?

 

[Shredicus]

Quote:

Originally Posted by RomeSC its my understanding that if you fry a vrm the card will not work at all. Am I incorrect? I just finished a heaven bench at [email protected] and played bfbc2 for an hour at the same settings. I doubt that would work with a fried VRM?

Actually not quite. Weird story. With my second 570 I was benching it on the new furmark when the latest version was first released and smelled the melting smell, then I shut the PC off at the PSU. Weird thing is it booted up ok and ran fine for about a week. No artifacting or anything, no more burning smell. I gamed with it at its normal overclock repeatedly. The card finally kicked the bucket after a week when I was playing Metro. I took the cooler off and sure enough, melted VRM. The explanation for this from what I read by local GPU wizard pizzaman is the reason these cards are dying from the VRM phases melting is because the metal liquifies and then actually leaks onto the PCB and contacts another elctrical surface causing a short.

 

[Defoler]

Quote:

Originally Posted by Shredicus Actually not quite. Weird story. With my second 570 I was benching it on the new furmark when the latest version was first released and smelled the melting smell, then I shut the PC off at the PSU. Weird thing is it booted up ok and ran fine for about a week. No artifacting or anything, no more burning smell. I gamed with it at its normal overclock repeatedly. The card finally kicked the bucket after a week when I was playing Metro. I took the cooler off and sure enough, melted VRM. The explanation for this from what I read by local GPU wizard pizzaman is the reason these cards are dying from the VRM phases melting is because the metal liquifies and then actually leaks onto the PCB and contacts another elctrical surface causing a short.

I'm not sure your wizard pizzaman knows that to "liquify" the metal in the caps require a bit more than 100c, and he is wrong by a few hundreds of degrees there.

Actually what happens is that the cap of the VRMs melt (they are made of cheap plastic along with the PCB), and the VRM internals does a little "kaboom" from excessive volts and heat. There is no metal "liquifying" of any sorts.
Because of plastic melting, there might be a contact exposure between the layers, yes.

Along with that, the soldering, which is the weakest link, might have answered the question "which is the coolers card around?" wrong and got kicked out by anne robinson, and the VRM is now off from its place and is just being held by the cooler only.

And to the OP:
Search for any burn marks, anything which might look like a melt soldering, or the sort.
I don't know how galaxy warranty, so don't know if its ok to open the card or not.

 

[Shredicus]

Quote:

Originally Posted by Defoler;13046029
I'm not sure your wizard pizzaman knows that to "liquify" the metal in the caps require a bit more than 100c, and he is wrong by a few hundreds of degrees there.

Actually what happens is that the cap of the VRMs melt (they are made of cheap plastic along with the PCB), and the VRM internals does a little "kaboom" from excessive volts and heat. There is no metal "liquifying" of any sorts.
Because of plastic melting, there might be a contact exposure between the layers, yes.

Along with that, the soldering, which is the weakest link, might have answered the question "which is the coolers card around?" wrong and got kicked out by anne robinson, and the VRM is now off from its place and is just being held by the cooler only.

And to the OP:
Search for any burn marks, anything which might look like a melt soldering, or the sort.
I don't know how galaxy warranty, so don't know if its ok to open the card or not.

It's entirely possible that I misquoted him actually, or just outright misinterpreted what he was saying. Think he mentioned the solder though... let me try and find the post.

http://www.overclock.net/12493635-post274.html

yeah, leaky solder