Nvidia drivers killing GPUs, Don't install anything after 320.18. Edit: Issues are now fixed.

[DDIF]

One or two issues do not make drivers buggy. It has to be widespread. Those one or two guys might be experiencing artifacts due to other parameters.
There were issues only in 320.18, which has been rectified in 320.49.
What i'm pointing out is that the point of this thread is pointless. I'm also forced to report your original post, due to spreading of misleading info and even feeding it, after people showed the reality with enough proof.

And what intentions you are talking off? I just gave an analogy wrt AMD and their driver issues regarding crossfire only. No issues in single gpu.
They are about to launch the fixed driver versions on this month only.

Buddy I'm not a noob, I did my homework and wasn't able to pin point the problem, then Digi told me about driver problem and the pics with faulty rendering. That explained everything, frequent crashes playing Killing Floor (furious_gamers is witness, we were playing MP), faulty rendering in KF, after one GPU dead, the only thing I changed was my drivers.
Any idea what caused this?
Drivers
Digi is not getting any money telling this to people, it was just his good intentions, how can you say he is misleading and feeding wrong info? Are you dev working on nVidia drivers? if yes, then you may be right that Digi and me are wrong but if you are not then stop reporting posts and wasting everyone's time.

 

[d3p]

^ that's scary.

 

[vickybat]

Buddy I'm not a noob, I did my homework and wasn't able to pin point the problem, then Digi told me about driver problem and the pics with faulty rendering. That explained everything, frequent crashes playing Killing Floor (furious_gamers is witness, we were playing MP), faulty rendering in KF, after one GPU dead, the only thing I changed was my drivers.
Any idea what caused this?
Drivers
Digi is not getting any money telling this to people, it was just his good intentions, how can you say he is misleading and feeding wrong info? Are you dev working on nVidia drivers? if yes, then you may be right that Digi and me are wrong but if you are not then stop reporting posts and wasting everyone's time.

You are unnecessarily making this worse. I never denied that the drivers weren't the issue. 320.18 are indeed faulty and had some adverse effects in GPU's, reported over multiple forums by consumers. The issue was taken into consideration and 320.49 was released to fix issues. How can you be so sure and pin-point on the fact that 320.49 was the cause of your problems?

Your problems could have initiated when you were ( maybe) in 320.18 drivers. It could be some other issue as well. Besides , defective GPU's caused by rogue drivers are easily RMA'ed. Its a pity you're not able to do so, as you didn't buy it locally(India). People get new cards as a replacement for defective ones. Its not like they are simply sitting ducks. You should find ways to RMA your card, from the place you bought instead of cursing the drivers and telling members not to install new ones, even though the issues were fixed. The card you see in my siggy was damaged after 10 days of usage. Had to RMA that and got a new one within a week. GPu's are extremely vulnerable to defects, more than a cpu.

Btw, one does not have to be a driver dev to question for or against their integrity. Faulty drivers will be bashed ( same happened with 320.18) based on simple observations and vice versa.
Digital Fragger claims are misleading because he's bringing 320.49 into the picture. The thread title mentioning 320.18 is perfectly fine though and so are the pictures posted by him, which cater to the same 320.18 drivers

 

[ico]

This happened before with some 190.xx series as well.

 

[Hrishi]

Yes, they are absolutely safe. WHQL is standard referred as "Windows Hardware Quality Labs" and is a quality certification for windows platforms.
GPU companies ( AMD and Nvidia) first release un-certified drivers known as beta drivers, which is close to the final version but not polished. They are released for a try-out by general public
and comment on its pros and cons that eventually make way to the final or WHQL version.

Getting BSOD(nvlddmkm.sys) while using 320.49WHQL drivers. Wasn't even gaming or using the dGPU. I was just web browsing , and it crashed with a BSOD.I have the dump file with me.
Never had BSODs before past couple of days.
Third time , it happened.

Reverting back to 314.xx , better have less perf. than a dead card and data loss.

 

[powerhoney]

What the hell's wrong with you vickybat???
What was the need to report the thread... Just cause you have more posts doesn't mean you are a know-it-all... Deal with it, man... Jeez!!!
On topic though, I had a BSOD with 320.49 too while playing F1 2012!!! Damn scary, considering I imported my lappy... Reverting back to older drivers!!!

 

[Hrishi]

What the hell's wrong with you vickybat???
What was the need to report the thread... Just cause you have more posts doesn't mean you are a know-it-all... Deal with it, man... Jeez!!!
On topic though, I had a BSOD with 320.49 too while playing F1 2012!!! Damn scary, considering I imported my lappy... Reverting back to older drivers!!!

Have you upgraded from 320.18 or direct jump from 314.xx to 320.49 ??
I have read in forums that some users who upgraded from 320.18 to .49 continued to had issues . This may also be because 320.18 already did enough damage to the card.

BTW , if possible play the F12012 with old stable 314.xx drivers and care to share the results .. Thanks.

And thanks for posting this Thread "Digi" . At least those who were on 320.18 and read it must have been warned. .
And VickyBat , take it lightly bro.

 

[Digital Fragger]

What the hell's wrong with you vickybat???
What was the need to report the thread... Just cause you have more posts doesn't mean you are a know-it-all... Deal with it, man... Jeez!!!
On topic though, I had a BSOD with 320.49 too while playing F1 2012!!! Damn scary, considering I imported my lappy... Reverting back to older drivers!!!

Getting BSOD(nvlddmkm.sys) while using 320.49WHQL drivers. Wasn't even gaming or using the dGPU. I was just web browsing , and it crashed with a BSOD.I have the dump file with me.
Never had BSODs before past couple of days.
Third time , it happened.

Reverting back to 314.xx , better have less perf. than a dead card and data loss.

good. glad the thread is useful to at least few people.

 

[gopi_vbboy]

Linus was right about nvidia.

 

[pratyush997]

Linus was right about nvidia.

Hmm Yeah!

 

[vickybat]

Getting BSOD(nvlddmkm.sys) while using 320.49WHQL drivers. Wasn't even gaming or using the dGPU. I was just web browsing , and it crashed with a BSOD.I have the dump file with me.
Never had BSODs before past couple of days.
Third time , it happened.

Reverting back to 314.xx , better have less perf. than a dead card and data loss.

Install driver sweeper and do a complete sweep, i.e remove all traces of nvidia driver. Restart the system and again install the latest 320.49 WHQL.
Report back here if you get any BSOD's. Btw, numerous review sites are stress testing gpu's using 320.49 which came with the launch of GTX 760.

No one has reported any such issues with these drivers. Were you on 320.18 before?

 

[ico]

Linus was right about nvidia.

yeah. Especially with AMD having Video Decoding (UVD) support and power management with their open-source linux drivers + Weston/Wayland working fine as well.

 

[Hrishi]

Install driver sweeper and do a complete sweep, i.e remove all traces of nvidia driver. Restart the system and again install the latest 320.49 WHQL.
Report back here if you get any BSOD's. Btw, numerous review sites are stress testing gpu's using 320.49 which came with the launch of GTX 760.

No one has reported any such issues with these drivers. Were you on 320.18 before?

Have already Uninstalled previous drivers and have used driver sweeper in safe mode.
And yes I was using 320.18 before coming to 320.49.
From the reports I have read about 320.49 feedback , it seems the overvolting issue caused by 320.18 got fixed and it longer causes hardware damages (probably) , but there's still some bugs and unusual crashes/freezing on 320.49.

BTW , do you have any idea about the ideal voltage for kepler cards at highest performance level state. ? Mine is 640m.

 

[vickybat]

Have already Uninstalled previous drivers and have used driver sweeper in safe mode.
And yes I was using 320.18 before coming to 320.49.
From the reports I have read about 320.49 feedback , it seems the overvolting issue caused by 320.18 got fixed and it longer causes hardware damages (probably) , but there's still some bugs and unusual crashes/freezing on 320.49.

BTW , do you have any idea about the ideal voltage for kepler cards at highest performance level state. ? Mine is 640m.

Well from what i know, gpu boost was behind the damaged gpu's due to 320.18. You know in kepler 6xx series, gpu boost works with respect to the card's tdp.
So until the threshold tdp isn't reached, the clocks rise. Now the tdp is maintained by a power state logic in the gpu, like c1 , c2 (eg). If c1 is the power state for max tdp, c2 is supposed to bring it down to idle. Now supposedly, the 320.18 drivers allowed clocks to rise so that the c2 state was never reached and clocks rose irrespective of relative tdp. This caused gpu damage due to heat and voltage. It was a serious issue.

The 7xx series cards feature gpu boost 2.0, which removes tdp and instead relies on temperature. So clocks will now rise and fall according to temperature. So a card with a better cooling solution, will achieve higher boost clocks. So the 320.18 never affected 7xx series cards afaik.

I think your 640m would have suffered, but i doubt your gpu has gone bad. Do you get the same crashes in 314.xx??

 

[jenimukanna]

i have installed 320.49 before two weeks in my asus gtx 660 v2 please help me out how to revert it ...

 

[vickybat]

^^320.49 is absolutely fine. No need to revert it back imo. Were you in 320.18 before?? If yes, then for how long??

 

[Hrishi]

Well from what i know, gpu boost was behind the damaged gpu's due to 320.18. You know in kepler 6xx series, gpu boost works with respect to the card's tdp.
So until the threshold tdp isn't reached, the clocks rise. Now the tdp is maintained by a power state logic in the gpu, like c1 , c2 (eg). If c1 is the power state for max tdp, c2 is supposed to bring it down to idle. Now supposedly, the 320.18 drivers allowed clocks to rise so that the c2 state was never reached and clocks rose irrespective of relative tdp. This caused gpu damage due to heat and voltage. It was a serious issue.

The 7xx series cards feature gpu boost 2.0, which removes tdp and instead relies on temperature. So clocks will now rise and fall according to temperature. So a card with a better cooling solution, will achieve higher boost clocks. So the 320.18 never affected 7xx series cards afaik.

I think your 640m would have suffered, but i doubt your gpu has gone bad. Do you get the same crashes in 314.xx??

Well according to this logic , I assume that my 640m wouldn't have been damaged. It's because I was using a modified vBIOS which changed the stock clock to something higher ,from ~624Mhz to 810Mhz.
The GPU boost never worked , because the clocks were like 135Mhz~450mhz~810mhz. 3 Different states. If it were for GPU boost the 2nd maximum frequency should have been around 750Mhz or so.

Maybe My GT640m survived because of the modified vBIOS.

Also , I was looking into GPU overclock thread for reference and found out that at Clock speed close to 850Mhz and more , the VDDC was maxed at 0.987V , on 314.xx drivers.
But with the latest drivers 320.49 , I was getting that value beyond 1.xx volt for same clock speed , even for a li'l bit lower ,you think it's overvolting ???

 

[topgear]

yes it's overvolting.

 

[jenimukanna]

^^320.49 is absolutely fine. No need to revert it back imo. Were you in 320.18 before?? If yes, then for how long??

i didn't know exactly how long in 320.18 now updated 320.49 wqhl on july 1 ..... my asus gtx idle temp is 38-39 load 59 is there any problem in it ...... ?

 

[vickybat]

Well according to this logic , I assume that my 640m wouldn't have been damaged. It's because I was using a modified vBIOS which changed the stock clock to something higher ,from ~624Mhz to 810Mhz.
The GPU boost never worked , because the clocks were like 135Mhz~450mhz~810mhz. 3 Different states. If it were for GPU boost the 2nd maximum frequency should have been around 750Mhz or so.

Maybe My GT640m survived because of the modified vBIOS.

Also , I was looking into GPU overclock thread for reference and found out that at Clock speed close to 850Mhz and more , the VDDC was maxed at 0.987V , on 314.xx drivers.
But with the latest drivers 320.49 , I was getting that value beyond 1.xx volt for same clock speed , even for a li'l bit lower ,you think it's overvolting ???

Its overvolting because of the modified vBIOS imo. Drivers don't cause gpu's to automatically over volt. Either you do that manually or takes the value of the current installed bios profile.
I'd say revert back to your default bios or decrease the voltage manually to previous value. That could be the cause of your BSOD's. Decrease the voltage to 0.987V and then see if it still crashes or gives BSOD's in 320.49.

IMO it shouldn't because there 320.49 is bug free. Do that and report here mate.

i didn't know exactly how long in 320.18 now updated 320.49 wqhl on july 1 ..... my asus gtx idle temp is 38-39 load 59 is there any problem in it ...... ?

If your system isn't showing any signs of problems, then i don't see any reasons to revert back, if you are on 320.49 WHQL.