Prime95

senorstealth · 10 Jun 2006 at 10:20

Is it possible for a hard drive problem to cause prime95 to fail. I have an overclock (see sig) that is occt stable, and prime stable one day, and then fails after 11 minutes the next.

I've underclocked the cpu and run the ram through memtest86 all tests over night with no problems.

I just can't explain the instability, anyone got any ideas?

Cyber-Mav · 10 Jun 2006 at 10:26

use goldmemory v6.92 to test the ram to make sure its not a memory problem.

senorstealth · 10 Jun 2006 at 10:29

Will do, i'm running prime on it again at stock speeds to see what happens, after that i'll try goldmemory

xirokx · 10 Jun 2006 at 18:33

how long did you prime for? which test do you run?

w3bbo · 10 Jun 2006 at 20:54

Temperature related error imo. I'll bet the ambient temp(and therefore CPU temp) was higher when prime failed.

senorstealth · 10 Jun 2006 at 23:01

Its running at 44 degrees under full load with case temp of 38, It was running the "in place...." torture test can't remember the full name, and its running at the minute so i can't look it up.

I've clocked it down to 2450 at 1.4375v and its been prime stable for 6 hours. I think it might just have been a lack of voltage at 2600 (1.55v).

senorstealth · 11 Jun 2006 at 11:04

Ok prime stable for 9 and a half hours, and the memory ran through all tests on memtest86 for 9 hours (12 passes) without problems.

Certain it was lack of voltage now

mc_bob · 11 Jun 2006 at 11:07

Have you lowered the multiplyer?

Your FSB times the multiplyer should always = 1000 or lower I think.

Default is 5x

say if your fsb is 250, change it to x4.

May be thats too high causing instability,

MC_Bob

senorstealth · 11 Jun 2006 at 11:12

FSB/HTT is running at 288, with a 8.5x multiplier, hypertransport of 3
giving 2448MHz stable for 10 hours (before i stopped it)

With 5:6 divider on the RAM giving 445MHz, no errors with memtest all tests in 10 hours. Timings are a bit loose for the speed, but i wanted to be sure there wasn't a problem with the RAM. I'll tighten the timings up, then redo memtest and prime it again

mmj_uk · 12 Jun 2006 at 10:39

if you run blend test and assign almost all your physical memory then yes it is possible, since it will have to page out to the HD.

senorstealth · 12 Jun 2006 at 12:00

I've had it stable for 2 days now,going to run some tests on the primary drive, been looking for a reason to get a new barracuda .10 or two

Cyber-Mav · 12 Jun 2006 at 14:30

i pity the fool who uses prime as a stability test.

senorstealth · 12 Jun 2006 at 22:11

and why might that be, its a well known test for stability