I have 2x GTX 580, in SLI... whilst playing games, usually world of tanks or guild wars 2, after about an hour or so of play, one of my cards just randomly stops doing any work. :|
Looking in nVidia Inspector, I can see the graph for "GPU0" sits dead still at 15% GPU usage, or whatever usage the GPU happened to be on at the time. - and my FPS practically halfs, so you can easily tell that 1 card has stopped doing any work.
Now the wierd part comes when I try to isolate the problem and stress test each card individually.
I ran both a TessMark and a FurMark (Furmark is rubbish btw, as the nvidia drivers throttle the cards) on each card individually (without the other card plugged in), and they both passed with flying colours!
Temperatures his a maximum of 78c, the GPU utilization was at 99%, both core and voltage were at 100% (not throttled), and they both happily ran like this for almost an hour without any sign of messing up.
So, the cards aren't at fault? - I've already swapped the SLI link cable for a brand new one out of the packet, so it can't be that.
I need to re-run my testing but using the other PCIE slot on the motherboard, in case it's a PCIE slot error. - however the motherboard is quite literally only 1 month old...
if it's not the motherboard at fault, the only 2 things I can think of next is that either my OS is completely screwy, or my PSU can't handle both cards running in SLI at full load?
however this problem doesn't appear always, I mean it -does- appear at least within an hour or two, every time I turn my PC on, but after restarting it usually goes away until tomorrow,.. sometimes it comes back again. - so what, my PSU can handle the cards perfectly fine, until the system his a random interval that requires juuuust a bit more power than usual which makes my PSU trip out? I don't know.
Hence why I've come here! - do any of you have any suggestions?
is my 750watt PSU too low for my system?
have you had this problem before and how did you fix it?