Diagnose my PC problem!

[Fraz]

Apologies for the length of post... It's confusing and would appreciate your opinions!

Here's the story:

Built the PC as seen in My System <----- about a year ago and it's been TOTALLY solid until now. The graphics card, a GTX 260 I put in recently - about 2 weeks before the problem started. I don't think the GPU has anything to do with what's going on, but I gotta give you all the details.

Anyway, PC was on during the daytime whilst I was at work, doing some crunching for the World Community Grid, as it normally is. I came home, and the PC had completely locked up, and I had to force the power off. From this point on, the PC worked but only erratically. Sometimes it wouldn't even post, other times it would run Windows for about half an hour before locking up.

So... I took the PC out of its case, and set it up to run on my desk so I could swap parts in and out more easily. I been testing the RAM mainly so far, as often when it hasn't posted, my the Abit BIOS code was stuck on "C1" which is "Detect Memory" according to the mobo handbook.

I've got 2 * 1 gigs of Crucial Ballistix, and 2 * 1 Gigs of OCZ. All PC8500. I started to test them all individually, and it seemed at first as if one of the OCZ DIMMs was dodgy, as with just this one in, it would often lock up in Windows. I set Memtest x86+ going overnight on it. Zero errors. Then I found I got a lock up today using only the other OCZ DIMM - It crashed about 5 mins into some Prime95 stress tests.

So now I'm really confused. It's not hung once with just the Crucial DIMMs in so far, but I find it hard to believe that both the OCZ DIMMs have become faulty at the same time. So... now I'm suspecting a motherboard problem.

Wise words from fellow Hexites required.

[TAKTAK]

try increasing the RAM volatge a little bit (OCZ likes it's juice...)
and possibly go up a notch on NB voltage aswell...

[Fraz]

try increasing the RAM volatge a little bit (OCZ likes it's juice...)
and possibly go up a notch on NB voltage aswell...

I'll give it a whirl. It's odd though, as this setup has been stable for over a year.

RAM is currently at 2.2 Volts.

I think I also massively underclocked the system, and set the voltages on auto, and the problem still occurred. I can't quite remember though. Maybe I'll have to do that test again.

[TAKTAK]

hmmm, tis a bit waffy... mebbe the mobo is just on its way out?

do you have a spare PSU you can try in it?

[Fraz]

hmmm, tis a bit waffy... mebbe the mobo is just on its way out?

I'm leaning towards that more and more also

do you have a spare PSU you can try in it?

Sadly not.

It's really frustrating. The problem is so erratic and hard to reproduce, that I'm having problems getting it to crash at all right now with just the OCZ memory in. Also, I'll be really annoyed if it is the Mobo, as it's only a year old and cost a lot at the time All the same, if I could conclusively prove it was one thing, then at least I could get on with replacing it. Arrrgh!

[greencross]

Why dont you think it's the GPU? Seems to me it went in then the PC broke. I'd take it back out and try again. Sorry to state the obvious, but could it be that the new GPU is producing extra heat? I'd try memtest and P95 with all the case covers on. Could it be the PSU is under rated for the new GPU?
What are the temp of all the bits? HWMonitor from CPUID is a stand alone prog that would give you these values.

[Fraz]

Why dont you think it's the GPU? Seems to me it went in then the PC broke.

I don't think it's the GPU because it was running fine with the new GPU for 2 weeks. Didn't encounter any problems whilst gaming, etc. But you're right... it could be the GPU. Frankly, right now I think it could be pretty much anything, which is why I'm so frustrated with it.

I'd take it back out and try again.

I sold my old GPU - because everything seemed ok - literally a few days before I started having the problems, and hence have nothing to swap it out with. Sod's law! I would have tried this already if I could

Sorry to state the obvious, but could it be that the new GPU is producing extra heat? I'd try memtest and P95 with all the case covers on.

No. The problem occurs both when the components are in the case, and also when it is sitting open on my desk with no covers/chassis near it at all.

Could it be the PSU is under rated for the new GPU?

Very unlikely I think. It's 520 Watt Corsair PSU, and is easily rated for my GPU. I also have a watt meter on the wall plug that my entire PC gets its power from. I'm yet to see this go over 350 Watts when gaming + all cores fully stressed.

What are the temp of all the bits? HWMonitor from CPUID is a stand alone prog that would give you these values.

I haven't checked the GPU temps yet, I confess. However, it can't be a temperature issue, because sometimes it won't POST after it's been off for hours. Regardless, I've been keeping an eye on all the other temps. Everything is normal AFAIK.


ARRRRRRRRRRRRRRRRRRRRRRRRHGHGHGHGHGHGHGHGHGHGHGHGH! Goddamn computers!

[Lucio]

I'd be looking at either a dodgy PSU (it could be the extra draw of the GT260 was enough to tip it from stable to unstable), a damaged RAM slot (easy enough to do without noticing when fitting a long GFx card) or a BIOS incompatabiltiy issue between the very new card and the older board.

Unfortunately there's not *that* much testing you can do. You can try Everest Ultimate edition to check the voltages being reported but really you'd need another GFx card to compare it against to see if the extra power was too much.

Damaged RAM slot, you'll probably never notice unless something is visiably wrong

BIOS issues, tough one because very few companys will admit to even widely known bugs.

[Fraz]

I'd be looking at either a dodgy PSU (it could be the extra draw of the GT260 was enough to tip it from stable to unstable), a damaged RAM slot (easy enough to do without noticing when fitting a long GFx card) or a BIOS incompatabiltiy issue between the very new card and the older board.

Unfortunately there's not *that* much testing you can do. You can try Everest Ultimate edition to check the voltages being reported but really you'd need another GFx card to compare it against to see if the extra power was too much.

Damaged RAM slot, you'll probably never notice unless something is visiably wrong

BIOS issues, tough one because very few companys will admit to even widely known bugs.

Interesting. I agree with you on the damaged RAM slot - that is currently my main suspect. A BIOS problem I hadn't thought off. I can update the BIOS...thing is... I'm a bit scared of flashing the BIOS because the computer might decide to hang halfway through.

I'm thinking the PSU/CPU are the least likely causes right now. But I guess you never know.

[sleepyhead]

Just something that came to mind, are all the PC components clean/free of dust? Nothing causing a short somewhere? No chaffed cables?

[Fraz]

Never did find out what was wrong. I suspect some dry solder joint causing an intermittent fault somewhere or orther. Anyway, it seems to have completely gone away for now, and hopefully it'll stay that way.

Thanks all for your help.