New Oc Passes Small Ftt, But Fails Blend In Prime95, Whats The Culprit

[cmf0106]

First thank you for taking the time to check this.

 

I am currently running with these settings trying to achieve a 3.8 OC on my q9550 (stock 2.8).

 

FSB- 1800

CPU-8.5x

Link and Synced

5-5-5-15-2T

 

CPU- 1.375v

FSB-1.325v

mem-2.1v

SPP-1.45v

MCP-1.6v

SPP-MCP- 1.4v

 

Idle temp: 38

Load temp: 55

 

With this the system performs marvelously in Prime95's small FTT stress test. But when I try to run a prime95 Blend test, it only

lasts about 3 minutes until encountering a fatal error typically on thread 2 or 1. The error reads "fatal error rounding was .5,

expected less than .4"

 

System Specs:

Intel Q9550 2.8

Heatsink: Tuniq Tower

EVGA 780i SLI FTW

Nvidia GTX 280 x2 SLI

G.SKILL 4GB (2 x 2GB) DDR2 @ 1066

SILVERSTONE OP1000-E 1000W ATX 12V 2.2 & EPS 12V

Windows Vista 64 bit

Edited October 12, 2008 by cmf0106

[crazy8s846]

Turn your northbridge up to 1.4, and see if it passes, id say its your memory if it is passing small ffts, on blend it uses more memory. Turning your north bridge voltage up will help it coup with the high fsb of 1800 that you are running, if that does not work you might have to go to 1.5. Id keep moving it up from 1.4 till it becomes stable but not over 1.55.

[bg8780]

Turn your northbridge up to 1.4, and see if it passes, id say its your memory if it is passing small ffts, on blend it uses more memory. Turning your north bridge voltage up will help it coup with the high fsb of 1800 that you are running, if that does not work you might have to go to 1.5. Id keep moving it up from 1.4 till it becomes stable but not over 1.55.

 

Mine is set at 1800 fsb and thats what my voltage has to be set at. actually i think i am at 1.45. Whichever one is before 1.5 (pretty sure the 1.5 text is in red in BIOS)

[ccokeman]

NVM