Recurrent Fatal System Error at start up

[pointlessalias]

Hello, I built a new PC in October last year (first time build), and it worked happily enough until December when I started getting error messages at start-up. I made a note of the first one which was:

"STOP c000021a {fatal system error}

The initial session process or system process terminated unexpectedly with a status of (0xc0000221 0x001003f0)

The system has been shut down"

After some constant rebooting and battering of the F8 key I was able to get it to do the 'revert to an earlier time' thing, which fixes things for a while, but the error keeps coming back.

When the problem strikes (which is pretty much every 1 or 2 days now) the messages seem to be slightly different each time; most recently it was 'paged fault in non-paged area', for example. In all cases the only way to 'fix' it is to hit F8 and keep doing system-repair or 'revert to last known good config' repeatedly until it eventually decides to start working again.

Memtest86+ came up clean after 8-10 hours or so, and the memory timings are reported as:

"RAM: 666MHz (DDR1333) / CAS 9-9-9-24 / DDR3 (64bits)"

...with CPUID confirming this, and listing the voltage as 1.50v, which (along with the previous settings) is what it says on the sticks of RAM in my machine.

About a month ago I finally had enough of trying to troubleshoot software/driver problems and did a full Windows 7 reinstall, and set it up to dual-boot with Ubuntu. I'm still having the same problem, and when the problem strikes it won't load either OS, which I'm assuming rules out a software problem.

I finally hit upon the idea of waiting until the computer wouldn't start and then trying memtest a few weeks ago, since it's one of the options on my bootscreen now. Lo and behold it reported a bunch of errors, literally within the first 10 minutes. 7000 or so on my first try, then over 10000 on my second.

If I try memtest when it's 'working' it passes fine.

So, I've no idea if that points to bad memory, or whether the fact that it's a really high number of errors, produced only when the computer won't boot suggests something else?

Failsafe settings in the BIOS makes no difference. I've disabled 'spread spectrum', too.

Once it loads up, the machine works fine, for hours at a time if necessary. (The only thing it does once booted which might seem 'unusual' is to make a LOT of noise if I run games, but I've read that the stock AMD cooler is very noisy anyway. It does sound a bit like a washing-machine under heavy load but I'm guessing that's just normal until I get round to buying a better cooler and probably unrelated. It's fairly quiet under normal usage).

System specs are:

Motherboad: MSI 880GMA-E45
CPU: Phenom II 955 BE
Memory: 2 sticks of Corsair Dominator DDR3 1333
GPU: Gigabyte GTX460 1GB
Hard Drive: Samsung SpinPoint F3 7200 SATA
DVD-ROM: Samsung SH-D163C 16x DVD ROM 48x CD ROM
PSU: Antec 500 Green
Case: Coolermaster Elite 340

I'm completely out of ideas on this now. (It was Sept last year I bought all the stuff to build the PC, so obviously I don't want to start harassing Scan for replacement RAM unless there's a good chance that's actually what's causing it).

Can anyone shed any light on this, or suggest some further troubleshooting steps to take?

Thanks for reading.

 

[attackllama]

Since your memory tests fine a bad or damaged hard drive makes sense as the cause to me. If you have a spare hard drive try it and see if it will give you the same problems.

 

[banthracis]

RAM could still be the issue. Memtest86 (that's what you're using right?) does not require HD to run, it is in of itself a bootable file.

My advice would be to start part swapping. either place your current parts in a working PC and see which swap causes errors, or swap in known working parts into your current PC until the problem is fixed.

you can start with easy stuff like RAM, HD, PSU, GPU before you do the more annoying stuff like Mobo and CPU.

 

[pointlessalias]

Thanks for the replies. Will try the part-swapping thing.

I've borrowed a defunct machine off a friend, but I'm a bit limited with regard to what bits I have access to as it uses DDR2 RAM rather than DDR3, and an Intel processor rather than my AMD! Not ideal! The second machine also doesn't work terribly well either, which renders things somewhat unscientific but I'll have to work with what I've got.

I'm pretty sure the HD from it is okay though, and there's a replacement SATA cable in there too, so I'll try those two first.

Presumably I can check the GPU option by just removing the graphics-card and reverting to onboard for a bit?

And since the PC I built has two sticks of 2GB DDR3 RAM inserted - would I be okay to just use one at a time to try and identify if one is failing? Or do they need to be paired?

The other thing that springs to mind is that I seem to remember reading (somewhere) about someone experiencing a problem with the type of RAM I have on MSI boards. Could it possibly be a compatibility issue? Does anyone know of any good resources for checking such things?

RAM could still be the issue. Memtest86 (that's what you're using right?) does not require HD to run, it is in of itself a bootable file.

I'm not sure I understand what you mean. I get that Memtest86 doesn't require the HD to run, but why would that affect the test results from instance to instance in such an extreme way? Sorry to be dumb but I'm missing something here.