Help with odd nVidia card stability issue

I have an odd issue with video card stability.
With a GTX260 installed both of my systems are stable with any of the power supplies I have on hand.

My Q9500 system is stable with the GTX260 or a GTX280 using any of the power supplies.

My i7 system on two separate motherboards running the same GTX280 or an MSI GTX470 will reboot when playing Left 4 Dead, Batman Arkham City or running FurMark using any of the three power supplies. Furmark is the quickest way to replicate the issue. Most other games run fine.
Both systems show this issue when running the GTX470.
The reboots only occur if I am running a higher resolution than 1680x1050.
The card temperatures when reboot occurs can range from 62 to 93C.
Boosting the fan speed through Afterburner decreases the temps, but doesn't prevent the restarts. Ambient temps in the room are around 21C and the cases have good airflow.

Does anyone have any further theories for things I could try to get this system stable?
System information is below.


Dell 24" 1920x1200 monitor.

System 1 i7
Intel i7 950
6x2GB Gskill F3-12800CL9D
Gigabyte x58-USB3 or EVGA x58 SLI3
MSI GTX470 or Zotac GTX280 AMP edition
GTX260
BFG 1000W psu or Corsiar 750W PSU
Antec Eleven hundred
5 Antec 120mm fans
1 Antec 200mm fan
Corsair H50 cooler
G.SKILL Phoenix Pro Series FM-25S2S-120GBP2 2.5" 120GB
2x300GB WD Raptor Raid 0
seagate 7200.11 1TB

System 2
Intel Q9500
8 GB DDR2 800
two 7200 rpm drives + DVD drive
Corsair 650w.
Antec 900 case 4 antec 120mm fans + top exhaust fan.

I was going to point my finger at the PSU first but with you having tried several (decent) ones I doubt that is the issue.

93C is definitely pretty hot for any video card... Might want to look into ways that will reduce temps (improve case airflow, number/speed of fans, speed of fan on the video card).

What versions of the nvidia drivers are you running? Old copies from the CD? website? Are they up to date?

colinstu said:

I was going to point my finger at the PSU first but with you having tried several (decent) ones I doubt that is the issue.

93C is definitely pretty hot for any video card... Might want to look into ways that will reduce temps (improve case airflow, number/speed of fans, speed of fan on the video card).

What versions of the nvidia drivers are you running? Old copies from the CD? website? Are they up to date?

Click to expand...

I am currently running 285.62, but I have had the same issues for all versions of the drivers I have used over the last year.
The Antec Eleven hundred has plenty of airflow. There are two 120mm fans blowing straight on the video card with ambient air at about 21C.
Using Afterburner to increase the fan speed on the card drops the card temps, but it doesn't stop the issue. I have seen the same issue when the cards are running less than 65C.
I originally bought two of the GTX470's for an SLI setup, but I never got to try it because I couldn't ever get either one stable by itself. MSI RMA'd one for me and is suggesting another RMA but I son't see that helping anything since three of the same model have done the same thing.

I'm currently running EVGA GTX 470s in SLI. When playing a demanding game such BF3, Crysis, Batman, etc... my top card can get up to 92 degrees Celsius (fans @ 75%) and the bottom card will be around 10 degrees lower. I don't think temperature is an issue since the 470s can go up to maximum of 105 degrees Celsius. I did see the 470s go above 100 degrees in Furmark after 10 minutes and I stopped it. Furmark is a bit unrealistic since I've never seen any game max out my cards like that.

My guess is that your driver might be crashing (check windows error). When you update your driver be sure to remove the older version. If you remove it via control panel, it won't completely remove the old driver and could lead to instability. I learned that the hard way. You have to remove the folder and files manually. Hope this helps.

psy81 said:

My guess is that your driver might be crashing (check windows error). When you update your driver be sure to remove the older version. If you remove it via control panel, it won't completely remove the old driver and could lead to instability. I learned that the hard way. You have to remove the folder and files manually. Hope this helps.

Click to expand...

There are no Windows errors in the logs related to the crashes other than "windows recovered from an unexpected shutdown." It beeps then basically acts like i pressed the reset button on the case.
I have removed the drivers through control panel and deleted folders. I tried the clean installation option from the nVidia drivers as well. Is there some further step I should take for a clean driver installation?

It appears, there are problems with a lot of the GTX 470s (which may be why they stopped making them). I guess I lucked out. But that doesn't explain the problems your having with the GTX 280...
Check this older thread below. Someone suggested upping the voltage.

http://hardforum.com/showthread.php?t=1537390

psy81 said:

It appears, there are problems with a lot of the GTX 470s (which may be why they stopped making them). I guess I lucked out. But that doesn't explain the problems your having with the GTX 280...
Check this older thread below. Someone suggested upping the voltage.

http://hardforum.com/showthread.php?t=1537390

Click to expand...

I tried turning off the Adaptive Power management in the nVidia control panel as suggested in the link, but it didn't help. The stock voltage is already higher than the suggested change.
If I have it underclocked from 607/1215 to 500/1000 and set the fan to 100% it will stay running, but it is slow and sounds like a vacuum running in the room.

Could it be the PSU? Does the PSU have 1 or multiple rails?

Sometimes manufacturers will split the rails in a weird way, so you cant get enough power to some parts if you connect everything to 1 rail. If you have 1 rail, as long as you don't exceed that amount you will be good. But, if on a multirail system, if you have graphics cards or other things all connected to 1 rail that pulls more current than what is stated for the rail, while nothing is using the other rail you can't run it.

The amps on the 12+ rail(s) and the required number of amps by the Video Card are what matters and NOT the number of Watts. You may wish to look into this just in case. This is my understanding of how it works. If I am mistaken, ppl feel free to correct me.

I got a single 12 volt rail with up to 80 amps. Sometimes PSU will have multiple 12 volt rails up to 20 amps which might lead to problems if you connect everything to that rail. A GTX 470 will use a 18 amps (maximum) and the GTX 280 needs 38 amps on a single 12 volt rail.

Other than that, I'm all out of ideas...hope you find a resolution that your problem.

This guy in the youtube video sounds like he knows what he's talking about...
http://www.youtube.com/watch?v=4EtNTjrX8Ug

So MSI suggested another RMA and they replaced the card. The replacement once again has the same issue. I tested my system with a friends EVGA GTX580 and had no stability issues. It passes Furmark without issue. I am once again talking to MSI about the issue.

I agree with the others. 93C is very hot, and I've actually had a similar issue with a GTX260 which would poweroff the system at around the same temperature.

I doubt highly that it is any of the PSUs as most modern graphics cards have temperature safety shutoffs to prevent damage to the card and the system itself.

Follow up update.
MSI was unable to replicate the issue on an AMD system during the third RMA. I don't know the results of the follow up testing on an intel system. After the intel tests they offered me a GTX560 as a replacement. I took it and it runs great in my system.
I picked up another 560 for cheap so now I have two GTX560's in SLI with a GTX260 for PhysX and the system is totally stable.