Weird shutdown problem

[Phage]

The story is that I have the system in my profile and it has been completely fine (6 months or so) until yesterday. After a few hours of gaming I shut down the machine as normal.

NB: A few hours is normal usage, temps never got above 62C. (idle 44C - 100% 70C)

Later I needed to do some other things and went to re-boot. Nothing happened. At all.
Power I thought. No, all seems well, printer, speakers etc all live. The only activity I could see without disassembling the box was the CMOS button was flashing at the back.
So I cleared CMOS and it booted A-OK into the default clocks. Had a quick look, nothing appeared to be wrong at all in the OS or hardware profile. So I re-started, went into BIOS and restored the OC profile. It now appears to be up again at 3.8GHz.

I just ran OCCT for an hour, which it completed with no errors, no BSOD, nothing. Realtemp3.4 reports min 40C max 76C. I used OCCT as I was told that it was a complete package and easier to use than Prime95.

Then it did it again when the kids were playing Sims 3. Wouldn't re-boot until I left for a while. When it did come back it reported a CMOS checksum error on the BIOS. Just been into the Bios and I get an error message that says that there has been a boot failure due to overclock settings and/or voltage changes. Wondering if the profile has somehow been corrupted. I can also see that EIST is still enabled. Is that right ?
However I have now booted back into the Default clocks and it's working fine. Could this be a dodgy RAM stick ? Dodgy PSU ? I have never seen a system just lose all power and be dead like this before.

Aaaargh ! Any ideas ?

[Phage]

Right I've done a system restore back to a week ago, and only taken those critical updates. Let's see what that does...

[Ulti]

It could be the CMOS battery too but that is pretty rare as 6 months is pretty short. I'm not sure if PSU/RAM can corrupt the BIOS though.

[Ruddrage]

Is there anything useful in the event logs? Does Windows know that it was rebooted unexpectedly?

[WhiteGiant]

I'd look at the PSU.

[Phage]

Windows doesn't seem to have written any logs that I can find. The BIOS knows it's had a problem, but there was no BSOD and no error message.
I suspect the PSU as it's the only thing that could do this other than thermal shutdown, and I've eliminated that. Can anyone else think of an alternative ? Otherwise it's new PSU time as I don't know anyone who has a spare.

Interestingly ETQW no longer runs. Tells me that my desktop must be at 32pp (?) and when I apply a last known good through W7 it tells me that it is unable to load fs.chk

This why I think there may have been some sort of catastrophic hit to the software drivers or OS. Hence the roll back. If that fails I might try a rebuild. I can't believe that it's the CMOS battery or the PSU so quickly.

[Ruddrage]

What wattage is your PSU rated at?

[Phage]

Ample 750W - Gets a good review as well at Jonny Guru. Antec Truepower.
Running back to normal now. Doesn't seem to be a stress thing. Sims 3 is hardly stressing the system.

I recently applied a few MSoft updates. I'm looking at a software issue, or PSU.

[Phage]

Hmmmm.
the roll back seems to have worked - so far. It would appear that the system does not like some aspects of the MSoft Live package which was the software that installed.
Moral: Do not use ANY optional updates from the MSoft update service. I've had this problem before with NIC drivers.