From: sjr@debian.org   
      
   Hi,   
      
   I think this should be part of the same transition, as without it, gcc   
   will often prefer x87 instructions even if SSE is available, because   
   that is the cheapest way to be compatible to the 32-bit ABI.   
      
   For example:   
      
        double square(double num) {   
            return num * num;   
        }   
      
   without -mfpmath=sse becomes   
      
        square(double):   
            fld     QWORD PTR [esp+4]   
            fmul    st, st(0)   
            ret   
      
   even if SSE2 is available, because the return value is expected at the   
   top of the FP stack. With -mfpmath=sse, it is explicitly loaded there:   
      
        square(double):   
            sub     esp, 12   
            movsd   xmm0, QWORD PTR [esp+16]   
            mulsd   xmm0, xmm0   
            movsd   QWORD PTR [esp], xmm0   
            fld     QWORD PTR [esp]   
            add     esp, 12   
            ret   
      
       Simon   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)