Pseudorandom Computer shut-downs

I

ilikecake

Gawd
Joined
Feb 20, 2006
Messages
759
HI
I am having a rather vexing problem with my computer.

I have been trying to verify that the computer is stable by running stress testing programs and monitoring the temperature (nothing is overclocked... yet). However when I try to do this, my computer will shut down after a few minutes. There is no blue screen, no weird behavior, the computer will just shut off as if I had pressed the off button. When I power on the PC again, there is no error messages waiting for me or anything.

The programs I am running are:
I have monitored the temperature levels, and have not noticed anything unusually high before the shutdowns. In the past (on this computer) I have run these programs separately and not had a problem. Is there some interaction between these programs that would cause my computer to die?

After I post this, I will be trying to run the programs separately again and see if any one of them kills my computer, but I want to get this post done in case my computer dies again. :D
 
OldMX said:
Whats your power supply?
Click to expand...

Seasonic M12-600 (in my sig ;) )

Edit: well, it did it again. It took a bit longer than before but running RTHDRIBL and SP2004 seemed to kill it. The temperatures were

Cores: 52-58 C
GPU: 80C

Do these look abnormal?
Also, is there any way to log the PSU voltages to see if they are the cause?
 
I've just had a problem exactly like this recently. PC would be fine until I tried to stress test it and then after a while it'd just shut right off exactly as you describe.

At first I thought it was my PSU and I was on the verge of spending £200 on a new one until I solved the problem...

It was the ram.

In my case I simply had to loosen the memory timings a bit and now the problem has totally gone away. It was the Trfc and Trc timings that I had to alter. The ram would pass windows memtest as well making it harder for me to pinpoint. But as soon as I adjusted those timings the problem was fixed. So in your case I'd say play with the ram a bit. Move the dimms around the slots or remove one to test etc etc.

It was odd because my ram was totally fine before I added a dualcore cpu over my single core one, so naturally the ram was one of the last things I suspected. It took nearly a month for me to pin it down, during which I've gradually cut and added customised ramsinks to almost everything on my mobo and graphics card and added extra fans to my case. A harrowingly annoying ordeal to say the least. :p
 
Thanks for the response. How did you test for the bad RAM? I have ran the onboard memtest for ~12 hours and had no errors. Do I need to test for a longer time?
 
Well to rule out the RAM completely, test for 24 hours. Question though: what do you mean by "onboard memtest"? Does that mean that your mobo have memtest built in?
 
Danny Bui said:
Well to rule out the RAM completely, test for 24 hours. Question though: what do you mean by "onboard memtest"? Does that mean that your mobo have memtest built in?
Click to expand...

Yeah, sorry for being vague. DFI included a version of memtest86 in the BIOS. I can enable it in the BIOS instead of using a boot floppy/CD.
 
ilikecake said:
Yeah, sorry for being vague. DFI included a version of memtest86 in the BIOS. I can enable it in the BIOS instead of using a boot floppy/CD.
Click to expand...

Twelve hours is not enough. I've seen errors show up 23 hours into a test... and one error is too many, soooo.....
 
In my case running prime95 in blend mode was what revealed the error after 2 hours. Bios based or boot cd based memtest86+ can often pass just fine while windows tests still fail. Sitting these in a pre windows state with minimal drivers loaded and with little cpu overhead creates a relatively cool temperature for your PC as a whole compared to when in windows.

If errors are being caused because of heat, then memtest86+ may not pick up the error whereas prime95 blend in windows will.
 
Thanks for the responses. I am currently running memtest and will let it run for ~24 hours. If that work, I will try to relax the timings and see if that makes it stable.
 
You must log in or register to reply here.
Back
Top