First post here guys. Just had a few questions regarding overclocking my SLI EVGA GTX 580 3GBs. I have found that they just refuse to overclock at all. I'm using MSI Afterburner 2.2 beta 4 and NVIDIA 270.61 drivers. These cards are stable at stock speeds and voltages (playing games, OCCT, 3DMark, no problems whatsoever). However, all hell breaks loose when I even slightly overclock them. Here are a few of the things I've tried:
 
I found I was getting artifacts and crashing even just pushing the core clock to 800MHz on default voltage, so I decided to start from baby steps.
I push the core clock up to 778Mhz from 772Mhz on default voltage (1.025V) and it crashes in Crysis 2 for me (screen goes black, I can still hear sound, but then I am forced to do a hard reset. At this point when I restart, the BIOS says it has encountered problems with my system overclock and everything is reset to default levels (multiplier, memory, voltages, etc). However as explained below my system OC is stable).
I increase the voltage from its default to 1.038V. I then can play Crysis 2 fine at 778 and 781Mhz (I was trying increments of 3Mhz), but OCCT gives me artifacts in its GPU test at those settings. I found that any increase in voltage from the stock voltage will give me artifacts in OCCT, even at stock speeds.
All throughout these tests, I haven't seen my GPUs break 70C temperature-wise.
Also to rule out system problems, I've tried overclocking these GPUs at stock CPU and memory and voltages, yet encountered the same problems.
 
Am I doing something wrong here? Or do I just have absolutely terrible overclockers? Should increasing voltage cause artifacting even at stock speeds and within reasonable temperatures? OCCT never runs above 50C yet artifacts nearly instantly when I push the voltage up. The rest of my system is as follows:
 
Intel Core i7 2600K (overclocked to 4.4GHz (44x multi), with 1.22V Core and Level 6 LLC, hours of Prime95 and IntelBurnTest stable).
G.Skill Ripjaws DDR3-1600 2x4GB (using XMP profile 1.6V)
Gigabyte GA-Z68X-UD3H mobo
OCZ Vertex 3 240GB SSD
RAID0 Seagate Barracude 500GBx2 HDD
ABS SL 1050W PSU
 

Like I said before though, even at stock CPU settings (no OC, auto voltage in MB), the same crash still occurs in Crysis 2, and artifacting still occurs when increasing GPU voltage. Note that the black screen crash only happens in Crysis 2, all other games I've tried with OC settings have not caused this sort of crash. I am only overclocking cores right now.
 
I'm using the 6+2 and 6 pin connectors from my power supply to power the GPUs, I didn't see any additional power connectors for the motherboard besides ATX and ATX 12V. Should I be increasing other voltages in the BIOS like QPI/VTT or System Agent Voltage? I seem to recall some having to increase IOH/ICH voltage in X58 mobos for multi-GPU setups.
 
I've checked the default voltages on the cards, it seems one card defaults to 1.025V and the other to 1.013V. However, once I apply voltage increases and then reset to default in MSI Afterburner, it defaults both cards to 1.025V.
 
Regarding the PSU, I've used this PSU before to power GTX 480s and to power GTX 580 1.5GBs to 850Mhz overclocks with no problems. How much more additional power does a 3GB GTX 580 use?

Hmm, So I tried disabling SLI and testing each card individually (I have a Surround setup so it was simple to just change the primary monitors to test each card). It turns out only the second card is bad with voltage increases, with OCCT artifacting pretty quickly after bumping up the voltage even a little bit. I just this is just a dud overclocker then? I even tried switching the 6+2 and 6 pin PCIe power inputs between the two cards to rule that out, and the problem still occurs.

Hmm, tried that and still no luck. I even switched the GPUs to rule out any PCIE problems, but it looks like its still happening to the same GPU I switched. Guess I just have a really terrible overclocker here. Tried some testing at stock again, and I've seen no problems.

maniacvvv
If you still get artifacts with gentle voltage increases in Heaven and Vantage.. I would suspect the PSU, it appears to be a 2-3 year old unit that has been run hard and it may be having issues under stress.

If it was a problem with the Drivers or the cards, you would NOT be seeing the error you quoted above.

+1. That PSU you are running is close to the lower end of the power requirements for your set up in the first place.
Add-in overclocks, and it is making it worse.
 
Instead of disabling SLI and switching cards around, have you tried removing the "good" card and running the other one on its own?
 
Some cards won't overclock, unfortunately they don't come with any performance guarantees in that regard.