_mm_unpacklo_epi32
_mm_tmp0 = _mm_unpacklo_epi32(r0, r1); \
_mm_tmp2 = _mm_unpacklo_epi32(r2, r3); \
__m128 lo = (__m128)_mm_unpacklo_epi32(a, b); /* (a[0], b[0], ...) */
__m128 hi = (__m128)_mm_unpacklo_epi32(c, d); /* (c[0], d[0], ...) */
__m128 lo = (__m128)_mm_unpacklo_epi32(a, b); /* (..., a[1], b[1]) */
__m128 hi = (__m128)_mm_unpacklo_epi32(c, d); /* (..., c[1], d[1]) */