_mm_shuffle_epi32
m = _mm_shuffle_epi32(m, 0x4e); /* swap halves */
rk = _mm_shuffle_epi32(rk, 0xff);
rk ^= _mm_shuffle_epi32(prkhi, 0x80);
rk ^= _mm_shuffle_epi32(prk, 0xfe);
rk = _mm_shuffle_epi32(rk, 0xff);
rk = _mm_shuffle_epi32(rk, 0xff);
m = _mm_shuffle_epi32(m, 0x4e); /* swap halves */
c1 = _mm_shuffle_epi32(r1, 0x39);
c2 = _mm_shuffle_epi32(r2, 0x4e);
c3 = _mm_shuffle_epi32(r3, 0x93);
r1 = _mm_shuffle_epi32(c1, 0x93);
r2 = _mm_shuffle_epi32(c2, 0x4e);
r3 = _mm_shuffle_epi32(c3, 0x39);