GPU problemsFollow

Was playing WoW today and I got this message "Display driver nvlddmkm stopped responding and has successfully recovered." Did it again once in Witcher 2 (but oddly didn't mess up game play or anything of that nature), and a stress test. Tells me Driver xxx has crashed and recovered.

I was using the latest nVidia beta drivers, I then downgraded to the next beta drivers and finally to the latest released drivers.

It crashed on everyone of those in WoW. Sometimes it would recover and sometimes it would just freeze the computer up so I had to hard reboot.

Got sent to this page on quite a few stress test http://forums.nvidia.com/index.php?showtopic=214770

Just checking to see if there is anything I can do to alleviate this problem without having to RMA the card.

The card is a 570 GTX.

Thanks

What PSU are you using and how old is it? Also, how long have you had the 570?

I have the on-board sound.

Onboard Audio

Audio Chipset
SupremeFX X-Fi 2

Audio Channels
8 Channels


I run Windows 7 Ultimate 64, my system idles at aroun 36C, I keep an eye on it constantly with GPU meter gadget. I've only ever seen it get as high as 60C and that was with stressing it, a lot. It has dual fans, so it stays pretty cool. I ramp the fans up with MSI Afterburner when it starts to get above 45-50C.

I checked my power cords today after it started doing that, I have two lights on it and they both lite up green.

I have a 23 inch monitor. I run it in 1920 x 1080. I run one 570 GTX

My PSU is a Antec HCP 1200W.

I used to have an 470 GTX and never ran into this issue. I've been doing some reading, and it seems that since WoW patched to 4.3 that alot of people get it. Granted mine could be caused of something totally different.

Anyways, thanks for the help. Appreciate it.

Edit: Because I forgot to add what sound system it was using.

Edited, Feb 17th 2012 7:15pm by DecendentMonk

Ok, will check that out and download an update if they have it. Thank you very much.

I have a feeling maybe WoW and Blizzard broke something (seeing as how many post I saw on the subject, mainly seems to target 570-580 GTX), but much rather try everything I can to fix it. Blizzard takes a bit to fix somethings.

Will let you know if it fixes it.

Thanks again, Kaolian

Are you running Firefox in the background, even minimized? Because I had a problem with this error for a while, where my screen would blank for a second then come back up with the error while surfing the webs. I found a work around on Nvidias forums that seems to be doing the trick now.

If you you run into the problem where it was because of Firefox. Open Nvidia control panel, under manage 3d settings, select Firefox and change power management to Prefer maximum. Then in Firefox go to advanced settings and turn off hardware acceleration. Seems to have fixed the problem for me.

Edited, Feb 18th 2012 11:59pm by BeanX

there is a sort of emulation driver for the Realteks that runs the XiFi drivers or something like that, so you get to run a psuedo XiFi on their chips. Tinkered with it once on my last motherboard when a Xi-Fi Fatal1ty PCI started acting up. One channel would keep dropping in/out under heavy load, presumably a bus bandwidth issue or something with the old PCI designs that they never resolved--but it was CPU intensive and wasn't quite the same and I eventually replaced it with an Asus Xonar/DX.

As for the Windows Driver Model (WDM)...that's part of the system for managing the various Audio/Video Input/Output interfaces. Various vendors will release new driver models from time to time to fix glitches here and there, so it may be worth trying a different one--especially when you have HDMI capable devices. I installed one from ATI and it softened some extreme clipping I was getting in FFXI.

Edited, Feb 21st 2012 8:56am by BDHERTZER

I switched out my 570 with my 460 and I never ran into the Timeout Detection and Recovery problem at all playing WoW or a in stress test. I then switched back to my 570, WoW again did the TDR, then ran for a good 30 minutes and again TDR.

I ran my 570 on a stress test with 8xAA for well over 30 minutes, my GPU ran at 98+% the whole time, never got above 48C that I saw. Not one TDR.

Earlier today I did manage to TDR my 570 in a benchmark while trying to OC it (seems I can't get it to OC above 850/1700/2000 @1.063V, yet I saw a website that tested it get one to 1006/2012/1148, not sure of voltage, on air. How is that even possible Link to OC) I like the TDR when I'm trying to OC, helps save me from hard rebooting all the time. But annoying in WoW.

Haven't had anymore problems with any other games, The Witcher 2 has ran fine since the other TDR (think I had it OC then, a wee to much), Skyrim, AoC, EQ2, and so on. No problems, but WoW gives me hell.

Updated my audio with some Creative program that ASUS had on website.

I've seen that I can do this, but not sure how much it would help.

How to Disable TDR:

1) open your vista registry editor (Just type regedit to show your registry)
2) go to the key : HKLM\System\CurrentControlSet\Control\GraphicsDrivers\
3) highlight it then right click on this key
4) Add DWORD TdrDelay
5) Set value of TdrDelay to "0"
6) Reboot

This will stop that msg from displaying. I also have a Nvidia GTX 570 and this fixed that problem.

Found on the WoW Forum: Here
Would this help any?

I can RMA the card, but I don't really believe it is the card as it seems WoW is the only program I run into this problem (minus OCing and stressing while OC).

Thanks

Edit do to failing at linking.


Edited, Feb 21st 2012 8:36pm by DecendentMonk

Ah, alright. I was wondering how that worked out. I can get to 850, and even then I may need to back off some.

Would it be worth it just to RMA the card and try a new one to see if that works? I have my reserves that is it, but I guess it would rule it out.

Or perhaps Return it and get another 460GTX for SLi? Think you have told me before that would be a bit better than one single 570 GTX? I may have misunderstood.

Like I said, I don't really think it's the card as I don't have any other problems with any other games (except OCed to much, but that is more of a my fault kinda thing).

I guess I could avoid WoW till they get this sorted out, I'll contact them tomorrow and see what they say. But according to that forum post, they are kinda shrugging it off as not their problem. And to find a work around, but I'll see what Tech support says tomorrow.

Anyways, thank you for the response. Appreciate it.




Edited, Feb 21st 2012 10:39pm by DecendentMonk

Thank you for explaining that. After reading that, I think for now I will hold onto the 570 and possibly upgrade to another 570 in the future. Maybe I can talk my wife into it sometime. May have to wait and see if the price drops some.

With SLi, it just needs to be two 570 GTX, correct? It can be a different vendor, or does the card need to be the same vendor and model?

Like right now I have the ASUS ENGTX570 DCII/2DIS/1280MD5 GeForce GTX 570, now would I need another one of those exact same cards, or could I perhaps go with a EVGA 570 GTX?

Thanks again, much appreciated.

Alright, I'll wait till the price drops on the ASUS a wee bit and see what I can do about getting another one.

Today, I ran WoW for 2 and half hours without one hiccup (holding my tongue that it holds out like that), the only thing I can even think of that made a difference was I switched from DX11 to DX9. Not one TDR. Will play again later and tomorrow, and if it doesn't TDR anymore, hopefully that means that is fixed. I can live without DX11, it makes games look very nice, but I would much rather my GPU not crash every 15-30 minutes.

Thanks again, Kaolian. Will definitely look into getting another 570.

Edited, Feb 22nd 2012 4:57pm by DecendentMonk

A standard entry level case these days is going to have one 120mm intake in the front and 1 120mm exhaust in the rear. For a gaming case, especially one with a water cooling radiator involved, you will have more intakes and exhausts, but the basic principles are the same. You want a balanced airflow, with a slight bias towards overpressure (more intakes than exhausts)if you have to have an inbalance. There are also sealed and unsealed cases. In a sealed case, the fan intakes are the only inputs, and there are no passive air vents. These are fairly rare these days because they tend to cook your equipment if a fan fails. More common are the unsealed cases, which generally will have additional mesh vents in front or on the sides of the case. The worst possible combination of a case is a sealed underpressured case. The DF85 is an unsealed case though.

The other thing to consider is that heat rises. So the coolest air available to your case will be the lower front intake. The warmest air will tend to be the upper back intake. this is often compounded by computers being under desks and in semi-enclosed spaces The air on the back side of the case with 120mm fans will tend to go out, then spool around the sides if there isn't enough room for diffusion. In an unsealed case, this often gets sucked in through side vents.

Also, you shouldn't really consider your power supply as an air mover when considering pressure balances, even though it has a sizeable fan, its also generating heat, and there is alot of component drag so it doesn't really move as much air as you would otherwise expect generally.

I generally exhaust over my radiators rather than intake, but it doesn't seem to make that much of a temperature difference if you reverse that assuming you have a decent pressure balance. In your case at the moment,I'd probably personally run the radiator off the side panel intake and set the rear fans to exhaust. that gets to be a pain with tube routing and power and whatnot if you don't have quick disconnects. That or just set the rear fan to exhaust with the radiator anyways. The 140's move a considerable amount more air than a 120, so it's close, but you are right to be concerned about that rear fan sucking in some warm exhaust air. Its on the bottom though so it won't be a huge issue. In the grand scheme of things, either position should be fine with the amount of fans you have in that case.

Its kind of a toss up again. Flipping the H80 around to exhust is going to give you better overall case airflow, and lower temperatures for your motherboard and ram, but is going to slightly raise your CPU temperatures. Either way is fine for normal temperatures. When you throw in likely 110+ degree days though, you are going to need to pay extra attention to cooling. For one thing, you should abandon any overclocking during the peak hot days. Your danger areas are going to be the northbridge chipset, your ram, and your voltage regulators on your motherboard, not to mention the capacitors surrounding the CPU socket. If you have an Asus motherboard. many of them come with this fan: http://www.ebay.com/itm/NEW-ASUS-Fan-for-Water-cooling-or-Passive-PU-Cooling-Only-Adapters-H4-/150715470571?pt=LH_DefaultDomain_0&hash=item23175792eb which clips on the heatpipe assembly for use with water cooling CPU coolers. For extremely hot temperatures, that fan, or something equivelent, is not optional. It adds very little noise. Goes on about like this:
[img]http://pcper.com/images/reviews/179/fansinstalled.jpg[/img]

you may also want to consider modifying your side panel door to accept a 200mm intake fan poitioned around the center of the motherboard. On normal days that will be complete overkill. On 110+ degree days if your AC isn't very robust, it may save your motherboard.