Help with p95 error

[aviduser]

My system is set up as follows:

 

Phenom 940 BE - (230x16) 3680Mhz, 1.45 vcore. 1.35 Nb vcore

Lapped T.r.u.e. (antec Tri Cool Push/Pull)

2x2gb Gskill pc2 8000

Gigabyte 790gp ds4h

Corsair tx850

 

Running p95 I get a core 1 "FATAL ERROR-rounding was 0.5, expected less than 0.4" after 1 hour.

 

Every other core is fine for 10 hours of stress testing.

 

I had the nb voltage at stock, and core 1 failed after 2 hours, a little longer than with the voltage raised.

 

 

 

 

Does anyone have any tips? or anything I can tweak in the Bios that will help get my OC stable. I'd really like to keep it at 3.7 ghz if possible, as my temps are about as high as I should go for OC'ing. Maybe I should try some better fans on my true and raise the cpu vcore a little more (thoughts?)

 

Thanks.

[Nuclear]

I see you mentioned upping your NB voltage, did you up your NB or CPU-NB voltage? They are different and help in different areas. I would try upping the CPU-NB voltage a tick and see if that helps. Keep an eye on your temps though because upping your NB and CPU-NB voltage will cause the core temp to rise around 2-3C at idle and a bit more once loaded.

[Verran]

Just to clear up a common misconception: The fact that just that one core failed means nothing. That will always be the case. It's a typical "weakest link" scenario. Each of the cores in your chip has a certain OC capability. One of them has to be the worst. It doesn't matter which it is really, because if one is unstable then your whole system is unstable.

 

You would fight this just like any other instability, which is usually increased voltage or improved cooling performance.

[aviduser]

I forgot I tightened the timings on my ram a while back, sp I loosened the timings on my ram, from (5-4-4-12) back to stock (5-5-5-15).

 

Passed 10 hrs of p95 so she seems stable.

Edited April 17, 2009 by aviduser