New quad core server CPUs: AMD Barcelona vs. Intel Harpertown

Hi,
I am the programmer who wrote FlamMap/FSPro.

I believe the reason the barcelonas did not fare so well in the benchmark is becauseI believe the version of FlamMap you have contains Intel compiler optimizations built around the first of the year. I think there's cpu identification code in there that disables SSE with thebarcelonas and reverts back to standard floating point code.

I am also concerned about the size of the benchmark dataset really only testing cache and not testing the memory system enough.

I am in the process of producing a command-line version for benchmarking. I am also trying to produce a dataset of sufficient size that will test the entire system (bypass any cache bias) so results are useful for real world comparison.

Please contact me if you are interested in the new benchmark. I'd also like to verify performance on barcelona systems.

stubrittainATgmail.com

Thanks.