mov eax, xmm2 fld eax psrldq 4, xmm2 mov eax, xmm2 fld eax fadd st(0), st(1) psrldq 4, xmm2 mov eax, xmm2 fld eax psrldq 4, xmm2 mov eax, xmm2 fld eax fadd st(0), st(1) mov eax, xmm1 fld eax psrldq 4, xmm1 mov eax, xmm1 fld eax fadd st(0), st(1) psrldq 4, xmm1 mov eax, xmm1 fld eax psrldq 4, xmm1 mov eax, xmm1 fld eax fadd st(0), st(1) //fpu stack artithmetic after this: //1:2 (2) 3:4 (1) 5:6 (4) 7:8 (3)