Guys, what's the right way of testing the CPU-NB of the processor? I'm always testing the following way:
Increase the voltage from 1,25 to about 1.5 - 1.55 in steps of 0.05 and increasing the HT Ref. frequency, while keeping the cpu, HT and ram close to (over or below) their normal frequencies.
After increasing the FSB with some MHz I'm running blend for a while and if no error occurs, increase the FSB again. After some point, I resatart and increase the CPU-NB multi and decrease the FSB again in order to maintain the other 3 frequencies in the same range.
With this method I have a table with voltage scaling of the CPU-NB. The next is testing it together with the RAM. After putting everything together (when tested separately) I'm correcting the frequency and voltage, while looking up the table and testing with Prime again.
Is it ok?
I'm asking this, because I see a big difference, compared to my best C3's...this 3100 with only 1.25V looks unreal, lol. If I need to increase it to 1.3, or even 1.35 after complex overclocking of the system, it's still much better than the C3's CPU-NB.



Reply With Quote
Bookmarks