mat4_0_addasm.cpp

inline static void sse_mat4_add(__m128* m0, __m128* m1, __m128* dst)
{
  __asm
  {
    // First load the args addresses
    // into ecx, edx and edi registers.
    mov ecx, m0
    mov edx, m1
    mov edi, dst
    // Do a an aligned move of m0
    // to xmm0 vector register
    movaps xmm0, [ecx]
    // Add the first rows of m0 and m1
    addps xmm0, [edx]
    // Store xmm0 value on dst first row
    movaps[edi], xmm0
    // Here we repeat the same for each row
    // of our 4x4 matrix.
    movaps xmm0, [ecx + 10h]
    addps xmm0, [edx + 10h]
    movaps[edi + 10h], xmm0
    movaps xmm0, [ecx + 20h]
    addps xmm0, [edx + 20h]
    movaps[edi + 20h], xmm0
    movaps xmm0, [ecx + 30h]
    addps xmm0, [edx + 30h]
    movaps[edi + 30h], xmm0
  }
}