Need some advice!

Ok here I am with my 2500+ :

12 X 200 @ 2.0V (2400mhz)

Rock stable at prime 95

The problem is that my FSB is not optimal. With my NF7, it is able to do 223 mhz fsb rock stable.

If I go 11 X 223, it is unstable (even with 2.06V and I think 2.0V it's enough with watercooling)

If I go 10,5 or 10 X 223 , it is unstable (my NF7 doesn't seem to like the 10 10.5 multiplier)

So to have the optimal fsb, I must go 9,5 X 223 = 2118 and it's a huge difference!

If the Barton is limiting the overclocking at 2400 MHz due to voltage and/or cooling, then I would set the FSB to 218 or 219 MHz, depending what the system can take.

218 x 11 = 2398 MHz
219 x 11 = 2409 MHz

5 MHz of FSB from the best of 223 MHz affects about 2% of memory bandwidth. The 218/219 FSB MHz setup would be better than the setup of 223 x 9.5 = 2118 MHz at which the CPU would lose 11%.

At FSB 218 or 219 MHz, it would allow you to tighten the memory ras/cas timing to x-2-2-2 and also to make sure CPU Interface is enabled in the bios.

Use Sandra memory bandwidth to make sure you are getting 95-96% memory efficiency. If just aiming for high FSB (MHz) and not paying attention to memory efficiency, is defeating the purpose of high FSB.

It is a known problem that some NF7-S rev 2.0 boards are having problem with multipliers 9.5, 10. At low multiplier (high FSB), I found that memory efficiency number are not good (~ 90%), equivalent to losing 10-15 MHz of FSB.

Reference:

Summary on overclocking the NF7-S rev 2.0 (with Tbred B 1700+ DLT3C and ...) (page 15)

Basically what he said. When it comes to AMD procs and Ocing them, hitechjb1 it the man.

I was gonna talk about how you might have reached the limit of the Barton. Ocing is kinda hit and miss, not every chip will get the same OC. One suggestion that I have would be using CPU Burn-In. It might help you get a little extra outta the proc.

Yeah thanks for the fast answer!!!

I never made a burn-in before....... if it work , how much mhz should I win?

And the method is like running the chip at 1000 mhz on 2.0V ?? or I am missing something?