PC Instability

Cobbydaler · 10-05-2008, 17:14

Lately my PC has become unstable under heavy load, e.g. video encoding, compiling largeish programs.

It seems to be when the temperature of the CPU reaches over 50 degrees, although this might just be coincidental with the amount of processing going on.

Symptoms are; in XP BSODs (usually IRQ_NOT_LESS_OR_EQUAL but not always) and in Linux the program loading the CPU crashes (if it manages to report any error it's usually a segfault).

It's not the memory, as I've run memtest86+ until the proverbial bovines return to the roost with no errors (if that's not a mixed metaphore, I don't know what is).

Anyone any ideas? Is it likely to be a failing mobo, processor or power supply? I seem to have noticed it more since I upgraded my graphics card to one that requires a direct lead from the PSU...

TIA for any suggestions.

10-05-2008, 17:21

single or multi core cpu? you could prime95 it this will blitz the cpu straight away and stress it to the max if it falls over quick it could be the cpu. Any problems showng in event viewer? Whats your motherboard temps? 50degrees on a cpu isnt to high on any of em

You can not really rule out the ram just cuz it passes memtest either ive still known ram to be the culprit even with an overnight pass pass

Druchii · 10-05-2008, 17:22

What are the system specs?
Have you blown all the dust out with compressed air for example? It's more likely to be the chipset (Northbridge) overheating than anything else, as at 50'C the CPU is barely breaking a sweat.

kingbuxton · 10-05-2008, 19:11

AMD or Intel? If it's Intel, check the Northbridge, is it getting hot, if it has a fan cooling it, is it spinning. Run Prime95 on blend overnight (or as long as possible).

If the PC is crashing in games, could be the PSU (as it has the AUX12v cable) if it is crashing not running games (the gfx card shouldn't be drawing a lot of power in 2D mode really) doubt it is the PSU (not to say it isn't).

As you say it started when the card went in, it does point to that. Was it a clean install after the card went in? IRQ_NOT_LESS_OR_EQUAL is a bad device driver if I am not mistaken (and I could be).

From Microsoft - Stop error 0x0000000A (Stop 0x0A) indicates that there was an attempt in kernel mode to touch pageable memory at too high a process internal request level (IRQL). Typically, this error occurs when a driver uses an incorrect memory address. Other possible causes of this error are an incompatible device driver, a general hardware problem, and incompatible software.

10-05-2008, 19:16

Incompatible software and drivers are imo pretty much ruled out by the fault being present on both XP and NIX

Cobbydaler · 10-05-2008, 19:25

System is an Abit KV7 with an Athlon XP 3000+ slightly overclocked at 2280MHz. 2Gb generic RAM, nVidia 7300GT with up to date drivers, 3 x PATA HDDs, PATA DVD-RW, external USB HDD. Not sure if the PSU is 350 or 400W without opening it up.

Mobo temp 38 degrees, GPU ~50 degrees, CPU in normal use 38 - 45 degrees.

No out of the ordinary errors in event viewer. Falls over with Prime95 after a couple of minutes in XP with an 8E BSOD or consult the stress.txt file (which doesn't seem to exist)..

10-05-2008, 19:33

cpu ram or mobo then dude. Good chance its the cpu mate or generic ram as I said even though it passes memtest.Drop the cpu to stock or even underclock it and try prime again perhaps
Might be time for a cheap c2d couple of gig of ddr2 and a new mobo if you ask me

kingbuxton · 10-05-2008, 20:53

Quote:

Incompatible software and drivers are imo pretty much ruled out by the fault being present on both XP and NIX

Hmmmm, true.

Quote:

Falls over with Prime95 after a couple of minutes in XP

Sounds daft, but that is a good thing, especially if it does it every time, at least fault finding should be fast. Is that 2Gb two 1Gb? Try it with only one installed. That will help eliminate RAM. If it is 2Gb then you are going to be a bit stuck without another CPU/RAM to try.

All you can do really is drop the CPU to normal speed. Try the RAM in a different slot (if it is a single stick). Try your original Gfx card. Unplug additional drives and see if you can unload the PSU a bit.

You could install something like Speed Fan, run Prime95 and have a look at the voltages and see if any look like they are an issue. http://www.almico.com/speedfan.php

Cobbydaler · 11-05-2008, 10:30

Quote:

Originally Posted by zinglebarb

cpu ram or mobo then dude. Good chance its the cpu mate or generic ram as I said even though it passes memtest.Drop the cpu to stock or even underclock it and try prime again perhaps
Might be time for a cheap c2d couple of gig of ddr2 and a new mobo if you ask me

Quote:

Originally Posted by kingbuxton

Hmmmm, true.

Sounds daft, but that is a good thing, especially if it does it every time, at least fault finding should be fast. Is that 2Gb two 1Gb? Try it with only one installed. That will help eliminate RAM. If it is 2Gb then you are going to be a bit stuck without another CPU/RAM to try.

All you can do really is drop the CPU to normal speed. Try the RAM in a different slot (if it is a single stick). Try your original Gfx card. Unplug additional drives and see if you can unload the PSU a bit.

You could install something like Speed Fan, run Prime95 and have a look at the voltages and see if any look like they are an issue. http://www.almico.com/speedfan.php

@ zing - hmmm, bit of a quandary then. Worth finding an XP 3200+ & hope it's not the mobo or change everything? To change everything would have to wait for funds

. I've tested the memory overnight several times with the most demanding memtest86+ settings but I suppose it could still be that.

@ kingbuxton - I'll try dropping the overclock & core voltage to normal. I use GKrellM as a system monitor in Linux. Current (ha ha) voltages are:

Vcore 1.76
+12 12.1
+5 5.1
+3.3 3.4
V5SB 5.6

I'll try compiling the Linux version of Prime95 (mPrime), run that & monitor the temps/voltages.

@ Druchii System is clean inside. CPU/Mobo temps at the moment are 42/37...

Thanks for your help...

11-05-2008, 10:40

I did also mention your overclock In post 7.

Cobbydaler · 11-05-2008, 16:26

Well, ran mPrime & it consistently fell over with errors or too many warnings once the CPU temp reached 50C in overclocked mode.

With stock settings it's been running the torture test for 4 hours without any complaints with the CPU temp at a steady 60C & mobo at 38C.

Don't know what changed to make the box unstable, guess it's just getting old...

11-05-2008, 16:28

yeah possibly the cpu is now less tolerant than it was before. Is the vcore still the same ?

kingbuxton · 11-05-2008, 16:33

Thing is, you said you put that Gfx card in, with the Aux12v, the PSU could well be a bit weak or underpowered, and it can't do the overclock and the Gfx card at the same time.

Either way, at least the issue is sorted.

Cobbydaler · 11-05-2008, 17:05

Quote:

Originally Posted by zinglebarb

yeah possibly the cpu is now less tolerant than it was before. Is the vcore still the same ?

Dropped the vcore back to 1.65, it was +5% when oc'd...