SLI O/C failure on GTX680

[Magnussen]

Hi all- new to the forum, so please excuse the newbie question.

I am having some inexplicable crashing on both Furmark and in-game (War Thunder and Skyrim specifically) at various overclock settings.

I am following a guide on overclocking a Gigabyte (reference) GTX680 2GB card found here: http://www.hardocp.com/article/2012...rce_gtx_680_overclocking_review/#.UtaOA55dWHg

I have an Intel i5 2500k overclocked to 4.5GHz on a P8P67P MoBo with 8GB of DDR3 PC1333 RAM and two Gigabyte GTX680 2GB GPUs. I am fueling this with a seven year old Corsair TX750 750 Watt PSU.

Whenever I attempt to overclock at anything over 50Mhz on either core or memory, I get crashes from any 3D program I run. That is, I can run them for anywhere from a few seconds to 10 or 20 minutes, but it inevitably either freezes up completely (requiring me to restart the computer completely) or crashes me to the desktop. This involves both adjusting the voltage using Precision X up to its maximum when attempting an overclock in the +150-200Mhz range on the core (I was getting inconsistent stability using Furmark that allowed me to move it up to as high as about 191MHz) as well as when I don't adjust the voltage at all, and only bump up the core to a very modest +50MHz core.

I have a very well ventilated case, and GPU temps on the 1st card (hottest) never go north of 73 (under extreme load).

In a (maybe) related matter, I find that when I leave my computer on overnight (I have it set to go into sleep mode after about 1:30 of inactivity), and try to restart it in the morning, it fails to 'system resume'; I need to restart the computer to get it back.

I don't really know what's causing the crashes, especially when I haven't touched the voltage.

Hence, I'm putting it out there whether it might be the PSU? It is by far the oldest part of my computer (bought in 2006/7), and while it's never really given me problems, I am thinking it might be having some trouble with power regulation (that's an official newbie term), but I'm unsure.

I kind of feel like the doctors not knowing what some kind of ailment is and just saying 'must be lupus'.

Any thoughts are appreciated!

 

[Johan45]

Have you tried the cards in the board seperately and oc'd them to see if you may have an iffy card. Or tried a different driver??

 

[Magnussen]

Haven't tried OC'ing one at a time- I'll give it a go after work today though now that you mention it.

But first, I'll run DriverCleaner on the system to completely remove all old drivers and re-install using the latest NVidia Drivers. I have a sneaking suspicion I might not have done that when I slapped my new 680 in there just before Christmas.

I'll report back if it worked. Thanks!

 

[Johan45]

One more thing I do when swapping hardware around is to clear the CMOS on the board.

 

[Magnussen]

What would the CMOS store that might affect the hardware configs?

I haven't ever wiped the CMOS, and I've changed video cards a couple times in the last year, which is to say, I'm probably well due!

Are there any risks in wiping the CMOS?

 

[Johan45]

No, there's no risk just make sure you have your OC profile saved. Maybe it's all in my head(I don't think so) but clearing the bios allows it to set up fresh with the new hardware, this is particularily important with ram I find. It's simple too. I power down set the jumper then swap my parts , put the jumper back and power up. Go into bios and input my settings restrt and off to the races. I just find I have less funny little issues this way.
But I swap parts like ram ,cpus and GFXcards quite a bit.

 

[Magnussen]

So, I pulled the CMOS battery, switched the jumper for about 15 seconds, rebooted, uninstalled the old drivers, followed the Driver Cleaner Pro instructions to the letter, re-installed fresh Nvidia Drivers and it would appear that I've regained some level of functionality at overclocks. I still can't seem to push it much above +100mhz on the core though when trying to play War Thunder (temps don't go above 60), but at least I can get that, which is a damned sight more than I was able to pull previously (which was nothing).

Another thing I noticed- both cards now are running at similar temps. Before, the lower card was running consistently about 20 degrees cooler than the upper card. Now, understandably, the upper card is still warmer, but only by 5-6 degrees, which leads me to believe they are both being utilized effectively.

It's the fact that the setup fails when temps are in a very reasonable range, such as the situation above, that makes me thing power draw might be an issue. I would still like to push these cards further, and I know that they're at about half of where they could be.

Part of me says just cut my losses and take what I can get, but the O/C'er inside is saying otherwise ...