I've tried lots of performance options for MSVC 6 and Borland 5.5 on Windows 2000 (Celeron Tualatin), but the speed of my program did not change significantly! The GCC 3.2.x (cygwin) compiled version finishes it's calculations in about 8.7 sec, but the MSVC and Borland versions need about 20 to 21 sec.
What is wrong with MSVC and Borland? Did I make a mistake?
Or will I have to buy Intel's compiler to get the same or better performance than GCC?
MSVC: -Ox -G6
Borland: -6 -O2 -O
Thank you for your help.