PC switching off when running prime

Associate
Joined
14 Oct 2005
Posts
1,982
Location
Pembrokeshire
Hey all,

Yesterday when cleaning my PC (while it was on), it froze up when I knocked something near the CPU with my hand. System would refuse to post, instead looping between 3 and 4 diag lights about 5 times, and then sounding a long beep. The power light flashes constantly.

Decided to take apart the system and give it a proper clean. PC posted fine (though I did reset CMOS this time, moved wrong jumper last time, silly me!), set bios up how it was, and did a memtest for 8 hours to confirm my memory settings. Passed fine.

Off into windows, ran Orthos and Speedfan. The stress test appears to run fine, until the system just turns off without warning. Tried several times, always happens within 30 minutes or so. Thought it might be the setting in bios to turn off the system if the cpu temp hits 65/70/75, but no :( The memtest didn't cause the system to turn off.

Any ideas guys?? :D

In case it isn't in sig...
DFI nForce4 UT
Opteron 170
2x1GB UTT
X1900XTX
Tagan 530w
 
The settings are the same as they were previously, as far as I can remember them! Not been into the bios for months, actually using the pc for once!

I've no idea what could be the culprit, though memtest86 is fine...
 
Grr it switched off on me while typing a reply!

The CPU is running at 2500MHz...HTT at 4x, FSB at 250MHz. Memory running on a 166 divider so is running at 208MHz DDR. Timings are 2.5-3-3-7-10-1T. All settings are the same as far as I can remember, though I have not ruled out the fact that I've set something wrong...though I've no idea where to start!
Gfx is at stock...
 
I would give the Vcore a bump and see if it makes a difference. Resets when running Prime are typically due to the CPU not getting enough voltage.

Jokester
 
Yesterday when cleaning my PC (while it was on)

Deathwish ? :confused:

If youve physically nocked something before and it caused problems, id see that as a hardware problem to be honest, check over the area you nocked looking for any obvious signs of damage.
 
Never really knocked anything, just assumed thats what happened as the PC just froze and the soundcard just poured out what was in its buffer over and over. I was attempting to reposition the temp sensor that I had around the heatpipes on my Scythe Mine, when it just froze up.

Deathwish? Perhaps! I never intended on cleaning inside my PC. I just saw chunks of dust in every possible hole in the case and cracked it open to see how bad it was inside!
 
Not just yet...any idea what Errata 92 Enhancement does? I enabled it as I'm pretty sure I read it increases stability when running with dual core processors. Disabled it now...
 
Aye, but settings that worked before i stuck my fingers in the pc should still work now! Just priming with that Errata 94 setting changed back to disable, and have reset my CPU voltage back to stock, seems ok so far! *crosses fingers*

Priority 9 on Orthos really slows the system down!
 
Pc's get fussy about almost anything!! I remember when i first put this 3700+ in the pc, it would only work properly with 1 stick of ram. If i put boths sticks in to make 1gb it would just farrrrk everything up big time. Resorted to putting the 3000+ back in and everything was bliss again. Upon getting a new cooler i decided to give the 3700+ another try and to my surprise it worked fine (it was not overheating before, stock was just too loud)

Totally random and to this day I do not know what caused it but it just goes to show how weird these things can be. Sticking your fingers inside clearing dust could quite well have had nothing to do with it!
 
1hr 12 mins so far since I turned that Errata 94 thing off!
So bored, miss my Linux install, but I can't really trust it for stress testing, as I've never used it for it before...
 
Could be a power issue - my rig did that a couple of times when I was using my old (5+ years) Thermaltake PSU while my shiny new OCZ one was being replaced. Try not overclocking...
 
make sure the atx conenctors are in straight and that youu havent banged off any caps arround the socket.

make sure northbridge heatsink is still on good

make sure all fans are still firmly in sockets

reseat gfx + power to gfx.

those would be my culprits
 
Yeah, had a look round inside before doing this run of prime, as well as disabling Errata 94, I also increased the DRAM idle timer from 32 to 64 clocks. Some connectors did push a bit further, which may have come loose during the build, including the silly floppy power connector sitting between the graphics card and CPU. Awkward to reach that one, used the lid end of a biro to push it down.

Been priming for almost 3 hours now, though had to set it to Priority 8 instead of 9 as even changing songs on Winamp was taking over a minute to action, got unbearable! Did leave it running for over an hour before changing it to 8 though. Have also installed 3dm06 and done a few laps of that at the same time as Prime...first time I've ran something graphically strenuous since I got my new UPS, it manages to draw enough power to light up the 2nd load LED!

One of the above things does appear to have done it though! Glad really, as I thought me prodding around trying to resit a thermal sensor had killed something, resulting in excessive spending! (kill a CPU, may as well upgrade the board and ram too ;))
 
Well after 16 hours of Orthos, with 5 hours of 3dm2006 looped in the middle, the pc turned itself off! Some would say 16 hours is 'stable enough', but the fact that it is turning itself off while being stressed is getting to me :(

Any ideas besides going back to stock and running tests with each set of settings? :)
 
Back
Top Bottom