View Single Post
Old 2004-06-10, 23:17   #9
delta_t's Avatar
Nov 2002
Anchorage, AK

3·7·17 Posts

It's not the amount really, but the speed between the RAM and processor. As an example I have a dual processor Athlon 1.2G with PC2100 RAM and when I run two instances of LL there is a slow down in both due to the fact that both are competing with each other for the RAM. My dual Xenon 2.4G has dual channel DDR333 and I see no slow down because the dual channel RAM is much faster so I am able to run 2 instances of LL tests with minimal speed degration. Actually on the dual Athlon I run 1 instance of LL, and 1 instance of trial factoring. That is because the trial factoring isn't RAM intensive because it does everything in L2. You should try some different combos and see for yourself. There is a benchmarking feature built into Prime95. Also, check out the Perpetual Benchmark thread and see what kind of performance people are getting.
delta_t is offline