Prime95 fatal error

Adam18290 · 29 Jul 2008 at 15:38

I currently have my q6600 @ 2.9Ghz with vcore of 1.3250V. When running ORTHOS there are no errors however when running prime95 two of the working threads stop within seconds of starting the test with "fatal errors".

Temps don't go over 60 degrees on 100% load & the prime95 tests continue to run.

Any ideas?

Cob · 29 Jul 2008 at 15:41

Are you running 2 copies of Orthos? Because it only runs 2 threads, whilst Prime95 runs 4 threads for a quad-core CPU.

And are you running the same test on both?

Adam18290 · 29 Jul 2008 at 15:44

Small FFT's on both. No I wasn't aware ORTHOS just did two cores and I'm only running 1 test of ORTHOS.

Cob · 29 Jul 2008 at 15:48

That'll be why Orthos isn't failing. It only tests 50% of the CPU. Use Prime95 instead.

But it shouldn't be failing at 2.9ghz. Have you set the correct memory voltage?

Adam18290 · 29 Jul 2008 at 15:50

2v with timings at 4-4-4-12

Cob · 29 Jul 2008 at 15:51

Which motherboard and memory? And what does the CPU voltage drop to under Prime95 load?

Adam18290 · 29 Jul 2008 at 15:52

Asus p5k and I have 2GB Corsair DDR2 XMS 6400 C4

Cob · 29 Jul 2008 at 15:55

What does CPU-Z say the vcore is after you start Prime95?

Set the memory to 2.1v and try again. If that doesn't help then set the FSB Term and NorthBridge voltages to +0.1v above their lowest setting.

Adam18290 · 29 Jul 2008 at 15:59

cpu-z says my vcore is at 1.18V when prime95 starts.

Whereas coretemp says its at 1.325V constant.

Cob · 29 Jul 2008 at 16:01

1.18v is a heck of a vdop & vdroop. Manually set the vcore to 1.375v in BIOS and try that.

Adam18290 · 29 Jul 2008 at 16:05

I have already manually set vcore to 1.325V in bios but it stills drops like hell in cpu-z when prime95 is ran.

I just upped the FSB a tad and upped the memory voltage to 2.1v - which in turn has made the vcore voltage in cpu-z increase to 1.2v when prime95 is ran.

P.S I have enabled the Dampener too.

Sitting idle cpu-z says my vcore is 1.26/1.27 even though its set to 1.325 in bios

Cob · 29 Jul 2008 at 16:08

The dampner doesn't work in the standard P5K, so you need to manually compensate for the huge vdrop, so manually set it to the 1.375v I recommended above.

Did Prime95 fail at the same point after upping the FSB Term and NB voltages?

Adam18290 · 29 Jul 2008 at 16:11

Ive just upped the FSB Term and NB volts and will run prime95 now.

Adam18290 · 29 Jul 2008 at 16:13

Nah prime95 still fails on two of the threads near enough a second or 2 into the start of the test. Ill up vcore to 3.75 now and report back.

Adam18290 · 29 Jul 2008 at 16:19

Right, so Ive upped vcore to 1.375 and am now running prime95 - worker thread 3 failed after about 10 seconds - which lasted much longer. All 3 others are still running after about 2 mins.

CPU-z states my vcore is at 1.224v when prime95 is running.

Seems to have got better but still as above 1 thread is failing.

Max temp atm is 72 degrees on core 1

Cob · 29 Jul 2008 at 16:20

Up vcore to 1.4v and try again.

The vdrop is dreadful.

Adam18290 · 29 Jul 2008 at 16:22

Vcore in cpu-z is lowering as the test progresses, now down to 1.21v.

Upping the vcore to 1.4v now.

Cob · 29 Jul 2008 at 16:23

Once one thread fails you may as well stop the test.

Adam18290 · 29 Jul 2008 at 16:30

Thread 3 failing after 2 minutes into torture test. All others are staying on atm at 1.4vcore. Temps picking up to 74 degrees after 4 mins in. I have stopped test.

cpu-z reports vcore to be at 1.24v when prime95 is running and 1.33 when idle. in bios vcore is set at 1.4v

booyaka · 29 Jul 2008 at 16:36

and i thought my gigabyte vdroop was bad!!!