XFX 260GTX Black Edition - pink screen crash

[Gak]

Sometime around last April i bought a XFX 260GTX Black Edition and untill two weeks ago i have been completely satisfied with it. Recently it as started to crash.

At first i didn't thought it could be the graphics card but after some tests i found that if the card goes anywhere above 77/78Cº it just crashes. I get a pink screen and that's it. Reboot.

When i bought it i run Furmark and all those benchmarks and the temp sometimes went past the 80's. Not for long though cause i don't like to stress my hardware unnecessarily and things were fine. Despite this behaviour it works flawlessly. On most games temps don't go above 65Cº, but on those who require a bit more juice it easily used to sit at around 78/79Cº. And it used to be of no problem.

Until recently, when it started to crash. I do know that normally stock coolers are kind of lousy at doing their job. But as this one kept the card idle at 32Cº and when playing on an average between 65/75Cº i thought that is was doing an acceptable job.

Cranking the fan up won't solve the problem either. It will only take more time for it to crash. Eventually, with the right conditions (game or benchmark), the card will it the mentioned temperature and crash. Not to mention the noise.

The only solution i've found is to downclock its values to the reference ones for the 260GTX cards. With those, theres no kind of problems whatsoever.

I've researched already and i've found others with the same problem. And it's not that rare with these 260 Black Edition cards.

What's the position by XFX on this? Does XFX accept a RMA based on such problem and at this lifetime of the card?

Kind of disappointed...

[Gak]

Almost 6 months after the original post, some news.

The thread has been quiet, and from XFX support i haven't had a reply which wasn't the usual polite speech.

Being this the main card i have got for gaming, sending it to RMA would mean a couple of weeks without it. And the possibility of returning with the same problem. If life has taught me something it has been not to trust anyone until contrary proof and always expect the worse.

With this in mind, and knowing the risks involved, i decided to put to practice some thoughts before going the RMA way.

All symptoms pointed to two possibilities:

-the card was overheating

or

- the factory overclock wasn't stable

So, opened it up, gave it a thorough clean and replaced that thermal adhesive/goo/tape with some better compound. Temps were a lot better. Took a lot longer to go so higher but still crashed when reaching that peak temp.

As i already knew, downclocking it would make things run smooth. But that didn't pleased me. After some research i came across a modded EVGA VT. Gave it a try and undervolted the card.

Shazam! No more problems with temps (at the factory defaults)! That means overclocked. And it even also allowed me to go a tad further.

I've read a lot, before and also after, about the possible negative vs or and positive effects of undervolting. Even did the math.

After some thorough tests and more research done, decided to mod its bios. Once more, since i was already using a different fan profile allowing me to play without reaching those higher critical temps.

It now hits around 65Cº with most demanding games, lower with lighter ones. The fan noise IS considerably lower. Not to mention that when idle i've been able to reduce even more the fan %, which means less noise when surfing the web or working on non GPU related tasks.

Things have been working stable, and i am happy with it. This configuration is far better imho than its factory defaults.

[alengrosevic]

Nice to see that you have fixed your problem, but... If you need to undervolt the card in order to function properly, then I think you have bad chip on card. If the chip isnt working on his standard voltage, then I think you have defective GPU chip. Maybe now is working OK, but what will happen in mounth or two? If I were you, I would try to RMA the card (if there is no traces of opening the card and cleaning it). good luck!

[Gak]

Nice to see that you have fixed your problem, but... If you need to undervolt the card in order to function properly, then I think you have bad chip on card. If the chip isnt working on his standard voltage, then I think you have defective GPU chip. Maybe now is working OK, but what will happen in mounth or two? If I were you, I would try to RMA the card (if there is no traces of opening the card and cleaning it). good luck!

If that does happen, i'll slap it the original bios and thermal tape again, if it is still covered with warranty. If that period has passed, oh well, i'm not to worried about it. RMAs around here aren't so quick like we are used to see here about the US.

The reference voltage required by this chip was a value i never found when researching. However there are similar cards running with 1.05v so if those can, this may be able also. And it has been working perfectly with a similar voltage

I wouldn't really call it a defective GPU. From factory it is overclocked and overvolted to churn out the proper Mhz. The cooler is (i guess) the same like the one in "normal" cards just with different stickers.

Unfortunately it can't cope correctly with the added frequency and volts. Amongst all my research i found hard to see cards from other brands with similar specs. reaching temps so high like the ones this XFX got to. And it's easy to find people complaining about XFX's Black Edition cards, always with the same symptoms.

Bottom line, the card isn't presenting any symptoms that would lead me to conclude that it is dying. It's factory overclock just isn't stable with so much heat. Reference nvidia settings are no problem, so the chip is fine i guess. If there is someone to blame it would be XFX for not assuring a better QC of their overclocked cards.