Box3D shared and static support for fasm2:

I looking avx2 code multiply two quaternions.

I found sse variant

Code:

macro quatmul a,b {  movups xmm0,xword [a]
                     movups xmm1,xword [b]
    movaps  xmm5, xmm0
    movaps  xmm3, xmm1
    movaps  xmm2, xmm0

    shufps  xmm5, xmm0,27
    movaps  xmm4, xmm5
    shufps  xmm3, xmm1,17
    shufps   xmm1, xmm1 ,187
    mulps    xmm2, xmm3
    mulps    xmm4,  xmm1
    mulps    xmm3,  xmm5
    mulps   xmm0,  xmm1
    hsubps  xmm2, xmm4
    haddps  xmm0, xmm3
    movaps  xmm1, xmm2
    shufps  xmm1, xmm0,177
    shufps  xmm0, xmm2,228
    addsubps        xmm0, xmm1
    shufps  xmm0, xmm0,156
 
   ;normalize
    movaps xmm1,xmm0
    mulps xmm0,xmm0
    haddps xmm0,xmm0
    haddps xmm0,xmm0
    rsqrtss  xmm0,xmm0
    vbroadcastss xmm0,xmm0
    mulps xmm1,xmm0  

}