Finding clues of what could trigger "nvidia driver has stopped and recovered..." on Asus strix 970 OC

venge86

Reputable
Sep 21, 2015
28
0
4,530
The title is pretty much a prologue. I also happen to have that infamous "nvidia driver has stopped working and recovered" without the recovery part though, as I just get a black screen right after (I look at the logs though and it says it recovered, then it says, no i couldn't recover it and microsoft driver kicks in).

My specs:
CPU: 4790k i7 at stock speed.
RAM: 2x GSkill, CL9 1866 set to XMP
MOBO: ASUS Z97-PRO GAMER latest bios
PSU: CORSAIR VS550.
GPU: ASUS STRIC 970 GTX OC version

Yes, I'm not innocent, that PSU is a dissonance to the whole system, classified as tier 4 in tom's hardware list. It was intended for a lower spec rig, although it's brand new, and I'll talk about it below.

Everything I did to stop the error from happening was futile. ASUS suggested me a driver tho, 353.62 which eventually fixed (!) the problem (10 hours gaming without problems), but also introduced choppyness and lower fps obviously as an older driver. I think though, the driver was using the card less efficiently, hence the lower power consumption, hence what I am about to say.

I used HWinfo to measure the 12V rail and crossed my fingers it was correct (i'll get a decent multimeter tomorrow). Under no load the 12V is frighteningly on spot, 12.000V idling, which sometimes dips to 11.904V. On heavy system load (prime95 and Heaven benchmark) the voltage diped to 11.504V. It might be within the 5% margin the PSU manufacturers give, but I don't know. I didn't like it at all. Could this be the reason I get the errors? I'm certain the system can't draw more power than vs550 can deliver (supposedly 550 watt), i'm talking about the wattage margin they advertise of course.

Could this gradually degrade components? VRM temperature of 970 reaches a scary 100C on heavy gpu load. I don't know maybe THAT could be the reason. Low voltage could actually increase amperage. Am I correct? Could this also be the reason of extra heat?

So do you have any insights? I rely on you. I like this gpu so much and it's also expensive (400 € in my country which is close to $430 😛 )

Thank you in advance :)
 


No, no, no... The VRM of 970 reaches 100C, the core never got more than 68C, I've got an intake fan directed on the card, pulling air from the bottom. I've got nice airflow in my case, 4x 12cm fans.
 


No, besides the factory overclock of course. In many threads I've read that underclocking a factory overclocked card could fix the problem. Although that falls to the question: Why underclock a card you paid more to be overclocked? And, it also raises another question: Is it the VRM temp, the voltage inadequacy, or the wattage margin the underclock fixes? Because we can't be professionals if we by accident fix a problem. Investigation and proper solution could fix everyone's problem! Nvidia doesn't have (or give) an answer yet... Who knows...
 
Nahh... the PSU is excellent, measured it with a very accurate multimeter and it shows 12.10~12.12V at idle, and a stable 11.96V at full load (same unrealistic scenario test prime95 @ small ffts and heaven benchmark @ everything full). Don't know. The VRMs seem fine at 100C it seems, they are close to a small heatsink as i saw from a naked 970gtx strix.
I really don't know...

watch this https://www.youtube.com/watch?v=2WL7v3eIPgU