Irregular Reboots in 3D Games / Apps

This is a weird issue I just cannot even reproduce at will:
While running any 3D application, my computer may (or may not) shutdown as if I pulled the plug, only to restart five seconds later. I have searched around and found several people who claimed to have a similar problem, but after trying all the suggested solutions I came to a conclusion that this is probably a bit different.

My specs:
Intel Core2Duo E6600, GeForce 8800 GTX, 2GB RAM, ASUS P5K mobo.

The problem is not consistent in any way I could find. I could be playing a resource heavy game for hours on end without incident, or use some light 3D app with this happening. It seems completely random!

I thoroughly dust cleaned the computer, the PSU and the GPU and I have made sure the GPU fan is at 100% constantly.

I tried stress testing the GPU, and the temperatures remain in the normal (below 90 degrees C) and the computer did not crash.

I have even replaced my 600W PSU to check if the fault lies there, but with no luck.

I have, of course, updated all the drivers and tried several GPU drivers.

The main problem is that the computer reboots at what seems like completely random occasions, which means I cannot reproduce the problem by myself and have to live in fear of it happening at any single moment. The only thing I do know is that this problem only occurs when I am using a 3D application of any sorts.

This is driving me nuts, please help.

I actually forgot to say that I did a big stress test for the CPU and memory using Orthos. Full windows reinstall I have yet to do. I don't think this is software related.

You mean for cooling the memory?

Check the windwos event viewer after one of these crashes. Let me know what is there (if anything).

Crashes like this, without a BSOD, are 99% of the time hardware related. Usually an interuption of power.. could be a problem on the MB, since you replaced the PSU probably no issue there.

The last time I had a problem like this it was faulty memory. But I heve seen it before with a bad PSU, bad MB (southbridge got cooked), and a fried CPU... years of overclocking things and assembling computers for others has given me a long list of components i've watched die .

The southbridge is generally for storage control (RAID, Sata, IDE, etc..), it doesnt have anything to do with the memory.. well.. not directly.. the northbridge controls RAM.

daedalus: There are no special events, because this isn't an operating system crash. It's some sort of a power crash. I am beginning to think that my replacement PSU is faulty as well. Maybe the store is ripping me off somehow.

Hmm.. ok, that's something I haven't tried yet. Is there any standard way to monitor the temperature of the south bridge?

Too bad I will only be able to buy new gear next week because of a holiday...
The thing that bothers me is that my PC ran smoothly for a year and a half, so why suddenly would this problem show up if its just the south-bridge overheating?

Sorry to hear you are still having problems.
My idea was that since everything else seems OK and the Southbridge deals with the PCI-E interface it might be overheating under heavy 3D load, because that would, obviously stress the southbridge more.
So, a shot in the dark: You have a E6600 I take it with stock cooling? If so, you should be able to set the fan to run at full speed, which will cool the motherboard voltage regulators (VRM) which is something else that may be causing the problem.
The controls are best set in BIOS, failing that try with Speedfan.

coozie the northbridge handles pci-e, not the southbridge (it handles slower pci, usb and audio).