Best ram for AMD platform

[Cecil]

Is it possible that you could test that with a higher CPU_NB clock, say 2600 or 2800? At 1600, memory bandwidth is bottlenecked by your NB clock...thats why nothing is really changing other than standard deviation when you tighten timings.

All runs were done 3 times, and the average posted. Single drops to a single timing arent expected to gain much, but from the 7-8-7-20 I started with, to the 6-7-6-18 I ended with, there was a 9 second drop in Spi 32m.
I was just seeing the difference each drop made.

I can try a high NB speed, but Ive never noticed better scores from it before, but that was with DDR2.

EDIT:
2600NB dropped 1m pi times on 6-7-6-18 by .4 seconds. Guess with DDR3 there is more of a difference then. Thanks for pointing that out.