Overclock.net banner

1 - 13 of 13 Posts

·
Registered
Joined
·
386 Posts
Discussion Starter #1
ok, so i built a machine for a buddy whos in the university of NY, he had their science lab buy from me because a comparable dell was almost double and would have been crap parts. they are using this machine to run some kinda protien modeling software that runs @ 100% cpu utilization for days and even weeks on end.

the machine is an asus p5n-t deluxe mobo, 2gb of kingston hyperx ddr2-1066, q6600, ac freezer pro 7, supertalent 600w psu, and a evga 9500 gt. the software they use is highly cpu intensive, but not memory or video intensive which is why those specs are not huge. its running on xp pro.

now, before i sent them the box i ran prime95 for about 12 hours to make sure it was stable. it passed with flying colors. since they have had it, their software crashes after about 6-8 hours. the same software works on other mult-core pc's without any problems.

i had him install OCCT and it crashed in about 4-6 hours when running on all 4 cores. so, he started running the modeling software on individual cores and it ran all night each time without a problem on cores 0 and 1, but then when he runs the software on cores 0 and 1 together it crashes.

we've ran memtest x86 all night with 0 errors - is there any way i can determine if its a faulty cpu or a faulty mobo? im confident its gotta be one or the other. again, the machine is at bone stock speeds - it should certainly not be crashing OCCT.

thanks for you help!
 

·
Banned
Joined
·
11,136 Posts
did you use a 29.00 PSU? What's your Idle core Temp? Have you bumped up your Vcore a smidge and your Northbridge another smidge.
 

·
Registered
Joined
·
386 Posts
Discussion Starter #3
voltages are stock, if i remember correctly idle temps are around 30-35 and full load was like 40-45 (going from what i remember)

the PSU is a supertalent 600w "atomic juice". supertalent makes good memory, and the amperage ratings on the supply looked good.
 

·
Registered
Joined
·
714 Posts
what makes you think it is the CPU/Mobo??? Try swapping a different PSU in...that's easy enough...although I wouldn't think even a generic CPU would have THAT much trouble powering just a 9500gt and a q6600.
 

·
Registered
Joined
·
386 Posts
Discussion Starter #6
ok we found out that the mobo defaulted ram timings and voltages to ddr2-800 specs.... i walked the customer thru bumping the timings up to their rated 1066 specs including a voltage hike for the ram and its been running OCCT for 5 or 6 hours now without any crashes. we'll see what happens tomorrow morning if it's still running.
 

·
Registered
Joined
·
386 Posts
Discussion Starter #7
update* it crashed after 9 hours. better than its ever been but still not right.

anyone have any ideas?
 

·
Registered
Joined
·
217 Posts
Maybe the wrong voltage was set at the factory? Try upping it a couple of notches and see if that helps. It should still keep it well within Intel defined specs.
 

·
Registered
Joined
·
5,319 Posts
9hr of OCCT is stable enough for 99.9% of what most people do. I'd say it's fine.
 

·
Registered
Joined
·
386 Posts
Discussion Starter #10
Quote:

Originally Posted by awdrifter View Post
9hr of OCCT is stable enough for 99.9% of what most people do. I'd say it's fine.
i guess my customer is in the .01% then

Quote:
they are using this machine to run some kinda protein modeling software that runs @ 100% cpu utilization for days and even weeks on end.
i can try upping the cpu voltage, we'll see where that takes us before i start ripping things apart and rma'ing things
 

·
Registered
Joined
·
386 Posts
Discussion Starter #13
bios is up to date, and occt shows the voltages right where they should be with virtually no ripple. i hate mysteries like this grrrrr
 
1 - 13 of 13 Posts
Top