Help with mysterious crashes...

Diggsy

Diggsy

Diggsy

Soldato
Joined
31 Aug 2007
Posts
4,856
Guys, I need your help or advice here. My system is as per my sig.

I am having a strange problem lately with mysterious crashes. My system will be ticking along nicely, then all of a sudden it just restarts. No BSOD, no lockups, nothing. It just seems to turn itself off and back on again.

System log etc doesn't show anything unusual, except I keep seing ftdisk and disk errors randomly, but they are listed as coming from my data (D) drive, not my System Raid (C).

Now, I have tried playing around with voltage settings etc in the bios and it still happens. The really strange thing though is that it is so random. I got one tonight, but that was the first for almost a week.

I have no idea what is going on. The only thing I can think of is my PSU (Tagan TG480-U01) but surely that would show other symptoms?

I can run Prime95 Small FFT's for hours and it doesn't cause the issue. I can also loop 3dMark06 over and over again. It seems to happen more when I am just browsing the web or something else that doesn't use many resources. Really weird.

For info I am running XP Pro fully updated. My temps are 40c on the hottest core idle, rising to 65c while Priming.

I would appreciate any help and/or advice. I have been building/tweaking for coming on 10 years now and this is the 1st I have been really stumped.:confused:
 
Last edited:
To be honest with you I haven't ran it long enough at default to know! :D I just assumed that when it will Prime all day long and run everything else I chuck at it at 3ghz that it was stable, but obviously there is something not quite right.

I guess I could try it at default settings, but given how random the crashes are, it might not help me much, especially as I will quickly get the itch to bump it up again!

One thing I did notice the last time I had the case side off arranging some cables was that the heatspreader on one of my RAM modules appears to be bowed in the middle a little, meaning that one small section isn't making good contact with the chips. I wonder if maybe that chip is overheating? Could that cause the restarts? Maybe i'll run Memtest and see what happens, or is there anything else that will generate loads of heat on my RAM, preferably within Windows?
 
Last edited:
Yeah, all well connected. I actually had the whole thing apart a week or so ago to reseat the North/South/PWM Heatsink because the PWM temps were aweful! I checked all the connections then and they are spot on.

Hmmm, just a thought, maybe my North/Southbridge is overheating since I reseated them? Would that be a possible cause of restarts, or would they cause lockups first? I don't think they are the problem as I checked that I had good contact before re-applying the paste, but you never know!

I also used Ceramique, so it's not as if it was rubbish paste. And my PWM temps are 30c lower than before, so it's unlikely.

EDIT: To help diagnosis, here's what my voltages are set at....

CPU - 1.31v
CPU VTT - 1.3v
RAM - 1.95v
Northbridge - 1.4v
Southbridge - 1.55v
HTT - 1.25v

Watching uGuru at load, my CPU dips to 1.26v, everything else is pretty much rock solid. 12v line never goes below 11.94v. Maybe the CPU just needs more juice? But why doesn't it crash in Prime etc???

Also memory is currently running at 5-5-5-15 2T, until I figure out what the issue is.

C1E and Speedstep are both enabled.
 
Last edited:
Yeah, will try disabling them. And no I haven't tried without the RAM, but I might slap a fan over it and see how it goes. Thanks for the help thus far.
 
Seems to be fine again today, without making any changes, but unfortunately that doesn't mean it's stable, because the crashes are so random. Ordered a OCZ XTC Memory Cooler and gonna slap that over my memory and see if it helps.

I guess it could also be that I am not giving the Northbridge enough voltage. I have read that running 4x 1gb can really tax the Northbridge. Perhaps if I still get random crashes I will up it another notch. There's a 40mm fan on there and on the PWM's, so cooling should be fine.
 
Okay, just got another crash tonight. This time on reboot it decided it needed to run diskcheck, which found and had to repair some errors. Now I am guessing these errors were caused by the sudden power loss, would I be right?

I have now taken my RAM back to 2.0v to see if it helps stabilize things.

I am also gonna boost the Northbridge to 1.5v and see how it goes. The really weird thing is it almost always happens when browsing. I have yet to have it happen while gaming. Could it maybe be the C1E or Speedstep?

This is becoming a major headscratcher. I would appreciate any input you guys have.
 
Thanks again for your advice. Yeah that sounds like a distinct possibility. Going to try that the next time I reboot, probably later today. Hopefully it will stabilise things. Only problem is that if it does, then my chip will run a lot hotter at idle. But I would rather have it stable than cool to be fair.

I guess it's just going to be a case of trial and error with a few settings until I find the mystery cause.
 
Tagan TG480-U01, the older and IMO better of the Tagans, before they started splitting the rails.

I know it could be at fault, but given that it's flawless at full load, I doubt it. Might change it soon anyway as I am sure it has to work almost at max with this setup at full load.
 
mattbill said:
You could try going Right Click on my computer>Properties>Advanced>Start up and recovery... and then untick the box that says automatically restart. That should hopefully leave up a BSOD if there is one. My friends machine rebooted whenever disks where inserted etc and never displayed a BSOD until we unticked that box.

Hopefully that might help you discover the issue. :)
Click to expand...


Yeah, funny I had forgot to do that, thanks. Will see what happens, although IIRC it should write an entry to the system log too, which I haven't seen.
 
Okay, something of an update really. Since installing the OCZ XTC Memory Cooler (which looks really good btw) on Friday morning, I have not had any crashes. Now, that doesn't necessarily mean I won't get any, as they were so random, but so far, so good.

So, perhaps it was something to do with the memory heatsinks after all. The only thing I can't understand is if it was heat from the memory, why did it not crash when at full load? Maybe I should RMA them? They were only bought from here in July.
 
Hmmm, seems I spoke too soon. Just had another reset this morning. Strange thing is it reset just as the system finished booting into Windows.

Yet again the restart was just as if I had pressed the reset switch on the case. No lock-up, no error message, just a clean restart. Now this has really got my thinking....

What is used most during boot, memory or CPU and which of these would be more likely to cause a clean restart?

Maybe it's just my PSU starting to show it's age, could a failing PSU cause these restarts?

Obviously I want to get this issue sorted out, but I also don't want to start replacing components that don't need replaced, so I appreciate any and all suggestions.
 
Cheers again for the tips. Voltage rails are virtually rock solid. I have been watching them whilst Priming and the lowest the +12v rail got to was 11.90. All other voltages are rock solid too. I just don't get it.

Maybe, just maybe it's Speedstep lowering the voltage too much, although checking Intel's spec sheet indicates that the SLACR needs between 0.85-1.5v, and mine drops to 1.15 when idle, so it's nowhere near their lowest value.

Gonna go for the default speed/voltage option and see how that goes. If that fixes it though, it could still be anything from CPU to voltage to PSU to memory troubles :rolleyes:

Oh the joys of PC issues. :D
 
Yeah, god only knows what the issue could be. I might even take the heatsink off again just to check the contact on the North/South Bridge chips, although surely they would be causing the crashes at load rather than idle?

I just don't know what to think, but I guess it really can't be anything other than a voltage issue or a PSU issue, given that full load is fine.

Time to fiddle around with the voltages a little more I think. Might even buy a PSU tester just to see if it picks anything up.
 
mp260767 said:
take it that the silence as a good sign ?? :D
Click to expand...

:D You would think so wouldn't you? But I am taking nothing for granted! It hasn't crashed since the last one I mentioned. But then again it is so random that it's hard to tell if it will go again or not.

The funny thing is though, since that all I did was to have a play with the overclock. I took it to 3.2, then 3.3 and did some benching. I didn't like the temps so I dropped it back to 3.0ghz at the previous voltages I was using and it's been rock solid ever since.:confused:

Maybe the CPU just needed a damn good roasting:p

Time will tell, but I will be sure to keep you posted...
 
Well, after what - more than a week without any issues, whallop! Off goes the PC a minute ago. Same as always, just as if I pressed the reset button.

I have to be honest, the sheer randomness of this has me completely stumped. It just does not lend itself to any form of troubleshooting.

Think I might buy a new PSU and see if that helps, but I hate the thought of handing over £60-90 just to see if it works.

Oh, in reply to memory - I have installed the only reliable Audigy driver that I know of. It took ages to find a decent one with my old install, so I have stuck to that ever since and it always worked well.
 
Okay, another bit of an update. The crashes began recently becoming more and more frequent, and in fact started happening both in game and at the desktop.

So, I was playing Half Life 2 Episode Two today and it crashed 3 times within an hour. Off I go into the bios and I dialled the CPU back to stock 2.4ghz and it has been flawless since.

Again, can't say it's cured due to the randomness of the crashing previously, but it seems ok for now.

Which leads me to my next question....why is the overclock causing this issue?

Could it be a poor clocker? (possibility for sure but it will Prime all day at 3ghz)

Could it be too little voltage? (not tried any more than 1.35)

or

Could it maybe be my PSU is just at it's limit and the overclock is tipping the scales? (find this strange though as it crashes at the desktop sometimes)

I truly am stumped. My gut feeling is that the PSU is just about it's limit and that's the problem. But I hate the thought of paying £90 or so for something that might not fix the problem. Decisions, decisions....
 
mp260767 said:
if it wasnt for the fact the problem seems to be so random and irregular i would have said it sounds like it could be the psu on the way out.

when u say back to stock is it just the speed or voltages as well ?
Click to expand...

I know mate, likewise, it's a really weird one.

Yes, everything is back to stock, all voltages too, although I had never increased anything else by very much, + or - 1 point was about it.
 
You must log in or register to reply here.
Back
Top Bottom