As Keith said btw, check your psu this is an interesting article on the subject http://www4.tomshardware.com/howto/20021021/index.html also is your cpu overheating? it could be that the heatsink is loose or the thermal compound has been applied incorrectly. Perhaps try running something to get the cpu to 100% and see if it will reliably lockup then.
Well, initially I thought it was overheating, since I have that computer on for about 12 hours a day. I've checked the heatsink, it seems ok. More recently, however, it's dying quicker and quicker- this morning it lasted a good 2 minutes 20s (yes, I timed it!) after the Gnome loaded until it froze. Yesterday it froze as the Gnome screen was loading! But then after a reboot I used it for a good 30 min before it went.
Hmm yes, the PSU could be a problem, altho the memtest86 results are intriguing. It is also strange that I had exactly the same problem a year ago.Interestingly, I have moved them back to the original configuration of RAM in slots 1 and 2, and memtest86 *hasn't* frozen, which maybe does suggest the mobo isn't the cause. hmph.
Saying that, I don't have enough drive bays to add a cd-rw drive, so maybe I'll get that new case (with new PSU, therefore) after all! Thanks for the link. I love that site. Now.. mid or full tower? heh heh heh ;)
I just checked and apparently the nvidia drivers now include 8x AGP support, so I presume they mean in their kernel patch...