Optimization Athlon xp

I have many computer to calc, and i’m probe the pgf90 compiler in my machines.
when i probe one program compiled with pgf90 (-tp piv -fastsse) in a proccesor P4 (1,6 GHz) , the time of end the program is 48 seconds;
and when i probe a same program compiled with pgf90 (-tp athlonxp -fastsse) in a proccesor Athlon XP (2,08 GHz), the time to end the program is 1 min 40 seconds.

What is the problem, or what is the parameters that i need to optimize the compiler for my Athlon xp.
I use Debian 3.1 (Sarge)
Kernel 2.6.10

Thanks
(sorry for my english)

Hi DonKijote,

The P4 is able to use SSE2 instructions which the XP does not support. Obviously this makes a big difference with your code. Try using the “-Minfo” flag to see what optimizations are being performed. You’ll most likely see more vectorization with the P4.

  • Mat