As part of my university studies, I've written a physics modelling program (in C, compiled with gcc, running on Mandrake 10) that's FPU-intensive. Currently I'm running it on an Athlon XP 2400+, and would like it to go faster.

I'm pondering getting an Athlon 64 (3000 or 3200). I've seen plenty of benchmarks for the A64's, but all of them have been either fully 32-bit or 32-bit binaries running on a 64-bit OS.

Will I see much of a performance gain in my program if I get the A64, run a 64-bit Linux distro, and recompile my code using the 64-bit distro's gcc? I know that it'll be faster than the XP 2400+, but will I see any extra gain from running in "true" 64-bit mode with a program compiled specifically for it?

I know it's rude to *bump* posts, but since this is a rather obscure question it's possible it scrolled off the front page before anyone who could answer it dropped by.

So, an apologetic *bump*.

The gcc compiler for AMD64 is pretty much a work in progress - you might want to have a look here http://www.anandtech.com/linux/showdoc.aspx?i=2163

There are some things that are even slower with a 64 bit environment.Of course the processor itself is a lot faster than the Athlon anyway but I doubt if anybody can tell you for sure what kind of gain you'll get with a specific programm that is only known to you ;-)