nVidia 320.18 drivers killing GPUs?

[newtekie1]

Microsoft doesn't test the drivers with cards if that's what you're thinking, they just look at the code to see if it conforms to WDM.

WHQL doesn't even do that, all Microsoft does is look at a bunch of logs generated from the manufacturers testing. Microsoft doesn't do anything for WHQL really, they basically just take the manufacturers word for it.

Yes, I was thinking that they actually tested it with sample hardware. Clearly WHQL certification has its limits and a junky driver can still sail through.

Easily sail through.

Since when software damages hardware like this is intended to show... in a driver update?

It's happened before, nVidia messed up a driver update back in the G92 era, the driver wouldn't spin up the fan and the card would overheat extremely quickly.

It could be a similar issue happening here.

 

[Outhere]

WHQL doesn't even do that, all Microsoft does is look at a bunch of logs generated from the manufacturers testing. Microsoft doesn't do anything for WHQL really, they basically just take the manufacturers word for it.



Easily sail through.



It's happened before, nVidia messed up a driver update back in the G92 era, the driver wouldn't spin up the fan and the card would overheat extremely quickly.

It could be a similar issue happening here.

I believe u are right, i have a gtx580M and installed the drivers and played a bit, no problema playing but i realized the temperature was a little high 90ºC. When i stopped playing the notebook while idle wouldnt temp down, it remained at 87-89. i rolled back the drivers to the 314.xx and the temperature immediatelly went down. The fried GPU people are talking about are probably and overheating issue that they didnt realize while playing. luckuly i didnt play for long.

 

[qubit]

Easily sail through.

Indeed.

So, that 'Q' in WHQL, is it just a way of maintaining control on the market with a digital signature, perhaps? Otherwise, what's the point of it if it doesn't actually produce a decent quality driver?

 

[TRWOV]

It basically means that the driver shouldn't break Windows.

 

[qubit]

I do love that "shouldn't".

So, going by that definition and nvidia's latest and greatest driver, it's ok to kill the hardware as long as Windows keeps running! Sorry, I can't resist irony, lol.

 

[Widjaja]

WHQL labeling makes the average consumer assume it is safe, will make everything work as it should and not going to break anything on the PC compared to a beta driver.

I know from my own experience with AMD WHQL drivers since using a 7xxx series AMD card, that WHQL does not mean any of the above.
12.10 AMD WHQL GPU drivers were the worst ever drivers for my card.

 

[crazyeyesreaper]

oh well switch to a different driver or let the GPU die, for me id let the card die and get a new one via RMA lol

 

[radrok]

I do love that "shouldn't".

So, going by that definition and nvidia's latest and greatest driver, it's ok to kill the hardware as long as Windows keeps running! Sorry, I can't resist irony, lol.

Fried card with consecutive no display output found but Windows still running anyway n downloading your Steam updates.

???

Microsoft mission accomplished

 

[qubit]

Fried card with consecutive no display output found but Windows still running anyway n downloading your Steam updates.

???

Microsoft mission accomplished

Oh yeah, just make sure you get all those Steam updates for that badass game, regardless!

 

[radrok]

Don't you love when you are playing something off Steam (QuakeLive mostly) and suddenly your ping skyrockets... you think about it 2 seconds and yell DAMN STEAM, NOT AGAIN!!!

Thank god they implemented bandwidth limit, took them so much damn time lol

 

[qubit]

Yes, I've been there. Nothing like a choked internet connection to make you feel all warm and fuzzy inside. :shadedshu

 

[newtekie1]

oh well switch to a different driver or let the GPU die, for me id let the card die and get a new one via RMA lol

Or control your own fan profiles and don't trust "the man" to do it for you.

Don't you love when you are playing something off Steam (QuakeLive mostly) and suddenly your ping skyrockets... you think about it 2 seconds and yell DAMN STEAM, NOT AGAIN!!!

Thank god they implemented bandwidth limit, took them so much damn time lol

I've never had that problem, ever.

 

[crazyeyesreaper]

I was being sarcastic newtekie im using the drivers and experienced 0 issues they work great for me, and even in the off chance they do kill my card it just means i get a shiny one back and i can sell it and up grade lol

Always gotta find that silver lining

 

[silkstone]

I was being sarcastic newtekie im using the drivers and experienced 0 issues they work great for me, and even in the off chance they do kill my card it just means i get a shiny one back and i can sell it and up grade lol

Always gotta find that silver lining

Unfortunately, most of the time, you get a crappy refurb back from RMA

 

[PersonWithTech]

Whether you overclock them like that or not really boils down to money and how much you can afford to replace a destroyed card. Most of us are not as rich as you and have to take care of our stuff.

Those waterblocks will keep the temperatures down, but the VRMs, GPU and other components are still getting massively stressed out from the out of spec extra voltage and clock speed and will die a lot sooner than stock speeds. Don't kid yourself that they "can take it" forever.

I agree. The 590 usually got hot very easily due to using two fermi 580s on a single board. The 580 alone drew enough power and produced enough heat. Kepler and the 690 was an exception, due to the improvement in electricity and heat. That is why the 690 is more similar to two 680s, and I'd OC with that (a bit).

If you really wanted to OC, just get two of the single-chip edition, as they handle the heat on separate boards and coolers. Also, it would mean that you'd not lose the whole lot if one GPU failed.

 

[newtekie1]

Unfortunately, most of the time, you get a crappy refurb back from RMA

Not in my experiences. Generally the part is marked as a refurb, yes, but they aren't crappy parts. But they are "reparing" your card, so technically it is a refurb. But a lot of the time with GPUs, I get a newer, more powerful, card back.

 

[silkstone]

Every time I have gotten a refurb, it has broken again fairly quickly.
With 2 exceptions, when I RMA'd a 8600GT I got a crappier version back until I complained, and with a HDD that had a instant replacement policy.

 

[qubit]

I agree. The 590 usually got hot very easily due to using two fermi 580s on a single board. The 580 alone drew enough power and produced enough heat. Kepler and the 690 was an exception, due to the improvement in electricity and heat. That is why the 690 is more similar to two 680s, and I'd OC with that (a bit).

If you really wanted to OC, just get two of the single-chip edition, as they handle the heat on separate boards and coolers. Also, it would mean that you'd not lose the whole lot if one card failed

Indeed, that 590 runs very hot. Mine's in an open case with lots of ventilation and you could fry an egg on it when it's running a game, with some making it run hotter than others. In fact, it came with a warning label saying to let it cool down before removing it from the PC to avoid getting burned and they're not joking. The card runs pretty warm at idle, too.

So, why the hell would I want to stress the thing out even more with an overclock/overvolt?!

Not in my experiences. Generally the part is marked as a refurb, yes, but they aren't crappy parts. But they are "reparing" your card, so technically it is a refurb. But a lot of the time with GPUs, I get a newer, more powerful, card back.

It's ironic that giving you back a better card isn't always a good thing, since it would break an SLI or CrossFire setup, lol. I guess one could buy another card to match it.

 

[PersonWithTech]

In fact, it came with a waning label saying to let it cool down before removing it from the PC to avoid getting burned and they're not joking.

Don't all graphics cards say that?

 

[shovenose]

I'm runing 314.22 on my two GTX550Ti cards on my secondary computer, just to be safe. It works just fine, so I don't see a need to update.

When my new computer with GTX 670 is done I'll use the latest on that

 

[radrok]

The card runs pretty warm at idle, too.

Yeah dual GPUs run very hot on stock cooling when on desktop/browsing and even while idling.

I recall the 6990 was in the mid 65c degrees while browsing with dual monitor

 

[qubit]

Don't all graphics cards say that?

My point is that they're really not exaggerating with this one. A single GPU card like a GTX 680 will get hot, but you may still be able to get away with handling it when hot. My old GTX 580 on the other hand most certainly had to cool down first, however. With two Fermi GPUs on the 590, you get even more heat.

Yeah dual GPUs run very hot on stock cooling when on desktop/browsing and even while idling.

I recall the 6990 was in the mid 65c degrees while browsing with dual monitor

That's pretty hot. Mind you, not quite as smokin' as one of my avatar babes.

 

[radrok]

I remember this article from the time being

http://www.hardware.fr/articles/825-4/dossier-nvidia-repond-amd-avec-geforce-gtx-590.html

Also 6990

Lol that generation of dual gpu should have been marketed with waterblocks only

 

[remixedcat]

650Ti SSC issues anyone?

I'm still on 306.97.

 

[cadaveca]

650Ti SSC issues anyone?

I have MSI 650Ti BOOST OC in SLI, no problems with this driver.

Also MSI GTX670 PE in SLI, no problems either. Not playing very many different games with them, however, Grid2, Rocksmith, BF3, WarThunder...that's about it.