KARKI JITENDRA BAHADUR wrote:
> 
> pfrants@cs.vu.nl (Frants P) writes:
> 
> >Hi,
> 
> >I am writing a 3d-engine with Watcom C/C++ and it produces *horrible*
> >code for casting a float to int when compared with gcc. The program now
> >spends more than 80% (!!!) of it's time casting float's to int and it

Try something like this in C, or translate it to an inline assembly 
macro. As far as I know, this was originally thought out and posted 
by Terje Mathisen on comp.lang.asm.x86. Works like a miracle for your 
fp to int conversion performance.

inline long float2long(float d) {
    #define MAGIC (((65536.0 * 65536.0 * 16.0) + 32768.0) * 65536.0)
    double dtemp = d + MAGIC;
    return (*(long *) &dtemp) - 0x80000000;
}