NoxioN said:
i tried a 300w sparkle, 420w turbolink and 400w allied. allied had best most stable rails and the sparkle was pretty bad. turbolink was ok but the 12 rails flucuated to much as did the ram voltage. that was bouncing from 2.99-3.09 which is way to much flucuation for ram. the allied keeps my ram voltage at 3.07-3.09v
I used the winbond hardware monitor that comes with abit boards. i would use a DMM if i knew where to take the readings.
A DMM may not be fast enough to SEE the line fluctuation which varies much faster. The best way to use a fast sampling oscilloscople (not available easily) over a period of time. DMM is good for measuring DC. Many believe the software probes are not accurate, but I think they allow one to sample the voltage line over a period of time to estimate the % line fluctuation. Since measuring % line fluctuation does not depend on the absolute accuracy of voltage measurement (even with a 5-10% possible absolute error).
From my experience with a 1700+, I was able to go to 2.4 GHz using a noname PSU, but no more. After changing to an Antec TP 430, I was able to go to 2.5+ GHz. Antec TP has 3% line regulation specification, which I think is important to run these chips at high overclocking, due to the current fluctuation, otherwise would result in large Vcore fluctuation (> +- 50mV) leading to system instability, I think. The NF7-S may regulate better than the A7N8X-DLX that I used. Actually measurement I did showed that the line regulation is right at 3% at high CPU overclocking taking into account software probe error for the TP430. Most other PSU's have 5% line regulation specification.
But your various voltage lines do not seem to fluctuate that much, probably that was at 2.4 GHz. You have to see whether they fluctuate more when clocking to 2.5+ GHz.
CPU overclocking does not depend on the FSB and memory, as long as the motherboard allows you to set the multiplier accordingly.
The difference between 2.4 and 2.5 GHz is not that significant in running programs, except for showing benchmark numbers. To spend on a new PSU for 100 MHz is not price performance efficient, except for academic or overclocking curiosity and satisfaction. That is, don't go out to buy a PSU because of that 100 MHz. If you can borrow an Antec TP 430/480/550 to try, it would be the best way to test it out.