64 vs 32?

[playdayz]

I noticed that machines with 4G or more ram with my processor seem to have significantly lower CPU mark scores than those with 3.5G or lower (meaning a difference between 64 and 32 bit windows I think). >4G = 2800-2900 vs <3.5G = 2200; that's with the processor not-OC'ed running at 2500. a phenom 9850.
Does this make sense? Is it the 64 vs 32 that is making this difference? Is it a real difference or just an artifact of the testing methods? Is it something else? Thanks. I bought Passmark to see what the benchmark data looked like and already it looks interesting.

[David (PassMark)]

The amount of RAM isn't so important for the CPU tests. But using a 64bit CPU, 64bit O/S with the 64bit release of Performance can make a significant speed difference.

[playdayz]

Thanks for the reply. As a result I am running Windows XP 64 and my cpu mark went from ~2200 in XP to ~2800 in XP64 (at stock speed). I am happy to have purchased Passmark and it has helped me tune this PC.

Now I have noticed something else. At 2.8GHz, Version 6.1 gave me a ~3150 cpu mark. I just installed and ran Version 7 and I got ~3960. I am pleased to see that of course, but I have the same basic question: Is it real? Are the results in Versions 6.1 and 7 comparable? I checked and my Version 6.1 was 64 bit also.

[David (PassMark)]

The tests are different between V6 and V7. We tried to make the score as similar as possible, but on certain CPUs there is a significant different between the V6 and V7 CPU results. So when possible just compare V6 to V6 results and V7 to V7 results.

[playdayz]

On the graphs then, will V7 results just gradually supplant V6 as people adopt V7?

[David (PassMark)]

Yes, we are getting in 400+ new V7 baseslines a day. So in a few weeks all the high end CPU results will be dominated by V7 results. For teh disk and video graphs, they are only comprised of V7 results.