Hi all, Apologies for posting about something that I think has been raised a while ago, but I wanted to see if there was any suggestions from this forum. I've been essentially been getting "Nvidia kernel mode driver stopped responding and has recovered" messages on a rig I've inherited from a mate. Running an old Asus Sabertooth X58 with an EVGA 770 Classified 4GB card inside. I've heard that this is mainly to do with TDR but also wanted to canvas views on anything Asus-related that I should check out first (e.g. Realtek driver issues, etc)? Thanks p.s. I can post full specs if needed
Erm - yes, was a fresh install of Windows 7...64-bit Ultimate... Drivers were then updated and I'm not sure if the recent nvidia driver is causing this (although rolling back, I still see it). Only thing I can think off is that the prior gfx card was replaced with the EVGA one I mentioned...so I don't know if there are any known conflicts of EVGA products with ASUS mobo's... Or if: - It's a power issue (although I have 750W so I doubt it) - RAM issue (although I have XMP enabled which I think is correct) FurMark can usually trigger the driver crash, but I'm at a loss as to why since the GPU monitoring I have doesn't show anything out of the ordinary going on with the GPU (temps are high but within limits, fan speed is adequate although I may try to artificially up this, etc) Could just be TDR but not sure why TDR is playing up with my 770 and I had no issues with a (much more inferior) much older card before...
There shouldn't be any conflicts between the motherboard and the graphics card. There is some good information here on TDR errors: https://forums.geforce.com/default/...-tdr-errors-display-driver-nvlddmkm-stopped-/ It could easily be a failure in one of the components somewhere along the line but as you can see from the list above, the possible causes are almost limitless. Without trying different configurations of hardware it might be difficult to get to the bottom of. I presume your PSU is a good quality one?
Yeah...Unfortunately I've tried a fair few permutations to no avail...tend to play games for 15-20 mins and then get a fail... Coolermaster 850w PSU so I'm assuming that's not the fault! I'm going to try a different PCIE slot to see of there's any noticeable difference (I doubt it but might as well try!).
Bad GPU/VRAM overclock. Run GPUz and have a look at your clock speeds - I'll bet they're higher than the stock settings. It could be a overclocking program like EVGA's Precision running at startup, or possibly a DIY BIOS mod, which can be done in two minutes these days.
I'm actually having the same issue at the moment with an EVGA Nvidia card in a W7 Pro x64 system. The display driver will crash every 15-20 minutes when playing full screen video or games. I doubt it's related to your issue though, I think the card in question is dying because when it gets too warm, one GPU (it's an SLI card) will shut down completely, forcing a hard reset and requiring it to cool off for a while before both GPUs will enable again. It's being replaced this week.
Hmmm, interesting...I am also tending towards thinking that this could be an overheating issue...although when playing games, tabbing out and viewing my GPU monitor only has it at around 50-60C...FurMark gets up in the high 70sC, so I can maybe understand a crash there...I'd still expect that to be within tolerance though... I'm not sure if, at higher temps, the card is throttling back and causing an issue there (either with voltage or something else).
Sorry, didn't see your earlier post. The EVGA Classified come OC'd...but I'd be very unimpressed if I'd need to underclock a card that's supposed to deliver performance OC'd...could well be the cause though as you say :/
I'm not saying it's defective. I'm suggesting it's been overclocked further, and it's unstable. Have you checked GPUz?
Yeah...at the EVGA 'out of the box' settings... I did run a "Video memory stress test" program and, when testing OpenGL, got an error, code 00000505...no idea what that means though :/
The nvidia driver crash and recovery thing is a common enough sight when pushing the VRAM to the max when overclocking and, coupled with that OpenGL error, suggests an issue with the card's memory. Have you tried stressing the card in another rig? Failing that, try underclocking it in your own rig and stress it, to see if it still falls on it's arse. If it doesn't you have your answer but, if it still fails, you'll need to test it with alternate components - ideally in a different PC.
*Sigh* agreed I'll need to find another rig to try and test it on...although I don't have one :/ EVGA Precision doesn't allow an underclock either, so I'll try MSI Afterburner instead...
Oh right ok, didn't realise that, thanks! Looking at the EVGA specs, mem clock is actually stock, but I'll try underclocking the GPU clock
Cool, whacked the GPU clock down by 50 to get it closer to base clock speeds... Odd how the EVGA GPU monitor app doesn't reflect adjusted clock speeds whereas Precision details original and current clock speeds...confusing!
So...downclocking the GPU clock *seems* to have made the crashing stop (or at least a lot less frequent)...shame given it's supposed to be a stable oc'd card... Have GPU-z logging things as well to see what's going on behind the scenes if it does crash again!
Did you underclock the GPU core or the VRAM? I'd leave the GPU as standard and try easing off the VRAM clock a bit, no sense in gimping your card unnecessarily.
Are any of the components in the system overclocked? (bar the gpu) i have this problem sometimes when my overclock is not stable (cpu, ram), or sometimes even when at stock but the memory controller needs a slight bump in voltage or timings need adjusting. First as suggested try the card in a different system. if it works, put it back in your system and enable defaults in bios, if its still doing the same thing then i would run memtest for at least 5 passes, then run intels processor diagnostic tool. if all passes try to use an older nvidia driver, or borrow someone else's card (if you can) to see if the same problem is occurring. If all of the above fails, post your bios settings and specs of the machine, then myself or someone else might spot something that needs adjusting.
Hi all, So, memory clock is stock 770...hasn't been adjusted by EVGA...you still think it's worth me actually reducing it but leaving the GPU at the stock OC'd levels? I'll try the memtest as suggested...although I was hoping that the Asus X58 memok capability would flag any memory issues...naive of me perhaps!
