I've written a small numeric program(600 lines) with c under linux. I compiled it under 64bit linux using 64bit gcc-3.4(without linking other external library), the program runs very slow. Then I managed to compile it under windows using VC on a machine which is exactly the same with mine(amd64 2800, 1g ram). The program runs very fast, the speed is nearly 1000 times faster. I don't know how to solve the problem, as my main develop environment is linux. I used gcc as follows: gcc-3.4 -g0 -march=k8 -O3 -fomit-frame-pointer -funit-at-a-time -fpeel-loops -ftracer -funswitch-loops -mfpmath=sse (do not use -ffast-math, as it will cause output error) with vc6.0 I just use the default options. I use clock()/1.e6 to determine the time elapsed in second under linux, output : 1.39 s under windows, output: 2.5e-3 s
