If you overclock, but forgot to lock pcie at 100mhz, it could cause stability problems.

Unfortunately haven't had the chance to OC yet, so that's not the issue.

I've also attempted this on fresh installs of both Vista32 and 64, and was sure to completely clean out my drivers every time I installed a new version of catalyst. Not sure if another windows install would show anything different, but I guess it may be worth a shot. If I had a store in town that had any NVIDIA cards in stock I would have purchased that and see if its another issue outside of my GPU that I've yet to be able to discover.

is it the Driver has stopped responding and successfully restarted crash you are getting or something else?

UPDATE: Handycat i've made sure my Direct X drivers are the latest, but now i've even seen texture glitches in windows and basic programs that have caused the drivers to stop responding. I have yet to be able to find a Memory Hole Option in my BIOS so did not test that. However now i'm completely out of ideas because I went and purchased a GTX 260, did a full format, installed nothing but the newest NVIDIA drivers, and 3dmark... and it still crashed during the tests. Its obvious now that its not the GPUs problem, but something else. Any help is appreciated.

Hold on. If you have a setting inside BIOS that allows you to increase the signal voltage going to your VGA card you need to up the voltage a little bit at a time because there is a lot of noise on the PCI-E bus but be very careful because you don't want to fry your video card.
Also you might try slowing down the PCI-E bus to around 95MHz.
Also, make sure that you're installing the video card in the correct PCI-E slot which the manual will inform you about.

Oh yeah, I almost forgot. If your operating system is a 32bit, you might want to try removing one of the sticks of memory. 32Bit OS's can only handle 3.5GB of memory. When you have over 3.5GB of memory in a 32bit OS, it can causes havoc on some systems.
If none of this works, then your motherboard is defective.

UPDATE: Out of sheer boredom I decided to go ahead and test my RAM again, and have received good/bad results. With the GTX 260 in, I went back down to 2gb of RAM, quickened my timings, and booted up. BIOS was not happy at all, but eventually I was able to get it up and running. And to my suprise, 3dmark Vantage completed all tests successfully, and I no longer crash in Crysis or get the odd textures.

I then added the 2nd stick of RAM again to see if it was just a timing issue all along, but unfortunately the issues came back but much worse with constant blue screens. So I removed both sticks and tried the other 2gb stick by itself. With the second stick by itself I can't even get past BIOS Auto Recovery, even after shorting the CMOS jumpers. Obviously it's leading me to beleive that the second 2gb stick has been bad all along, yet somehow passed Memtest twice. Bleh....

Now i'm going to remove the GTX 260 and go back to my original HD4870 and see if I'm able to achieve the same results with the single stick of RAM.

Anyone have any idea on how I can force my way past BIOS Auto Recovery for a Gigabyte X48-DS4 so I can see if the 2nd stick of RAM will even boot? Or another way I can test the RAM other than Memtest? Or have anymore insight into these issues?

have you got the latest bios for ur motherboard

OP Had the exact same problem twice before and, like yours, mine were caused by a faulty memory stick; Works fine with Word or surfing the Net, but once loaded up by a game, it fell over, or, sometimes, just would n`t boot!
RMA the obviously faulty stick and enjoy the benefits when its replacement graces your rig.