The sequence is: try to wake PC from sleep mode, fans spin, lights blink, but nothing on the screen. Force restart. As it goes through the pre-OS boot, there are weird artifacts (lines of @^__~ type symbols) amongst the normal lines of text, as well as some colored background bits (lines of little marks like -------- -------) in CGA colors behind the text. When the OS starts to boot, it goes to 4-color 640X480, and the background lines remain (they're generally in whatever negative color corresponds to what the screen should be). It boots OK, and the windows device manager tells me that "Windows has stopped this device because it has reported problems. (Code 43)." This morning, I used this working state to back a lot of stuff up onto my external HD. Had to leave for work at that point before I could experiment further.
When I got home, I opened the case and checked all the connections. After doing that, the PC booted completely normally, and there was much rejoicing (although I didn't try to run anything more graphically intense than Firefox). I went off to cook dinner, etc., thinking that the problem was resolved. The PC went into sleep mode. When I tried to wake it up, everything went the same way that it did this morning (and there was much swearing). Further jostling inside the box has done nothing.
The system is the one that I built 3 years ago, and hasn't changed meaningfully since then:
Since I just wrote all this out in the PC pics thread, I figured I might as well throw it out here, too.
CoolerMaster Centurion 5 case
EVGA 680i mainboard (A1 revision)
Core2Duo 6400 @ 2.13ghz
Samsung 18X DVD-R drive (SATA)
Seasonic M12 600W PSU
2 ghz Patriot PC-6400 RAM
EVGA 8800 GTS 320 ("Superclocked" version)
Seagate Barracuda 7200.10 320GB HD
Netgear WG311v3 Wireless adapter
LITE-ON SK-1788 keyboard
Logitech LX5 wireless USB mouse
Logitech X-230 2.1 speakers
Samsung Synchmaster 941BW 19" widescreen monitor (1440X900)
Vista Home Premium
So, my initial inexpert diagnosis is, not surprisingly, that I have a video card failure. But I am concerned that it might instead be a mobo issue. I don't have another PCI-E card that I could use to test the motherboard (this machine was an upgrade from an AGP system). My concern is that the replacement GPU I would buy for this system is quite different than the GPU I would buy for a system with a new motherboard.* So I ask you: is there any particular reason to suspect that my mobo (or even my power supply) is causing this problem, rather than the GPU?
The plan, for the moment, is to order a new GPU in the $150 range and drop that in to see if it works. Later on this year, I'd also like to wipe the HD, move to 64-bit Win7, and add more memory. (I figure that should last me pretty well until the next console generation rolls out, which is when I expect game spec requirements to start increasing in a meaningful way.) But I'm going on vacation in a little more than a week and don't want to mess with all that right now.
* In short, I'm going to be processor-bound until I upgrade that, and I'm probably not going to bother upgrading a LGA-775-based CPU that's only going to last me another year or two before I start over with a more modern mainboard. (Also, I hate removing/installing CPUs and the associated heatsinks/fans.) Thus, sinking more than, say, $150 into a GPU to use on this mobo just seems wasteful. On the other hand, if I were replacing the motherboard, I would invest a bit more in the GPU for future-proofing purposes.