H4
reduce4(__m128i H1, __m128i H2, __m128i H3, __m128i H4,
H4_X4_lo = _mm_clmulepi64_si128(H4, X4, 0x00);
H4_X4_hi = _mm_clmulepi64_si128(H4, X4, 0x11);
tmp3 = _mm_shuffle_epi32(H4, 78);
tmp3 = _mm_xor_si128(tmp3, H4);
__m128i H, H2, H3, H4, Y, T;
gfmul(H,H3,&H4);
reduce4(H, H2, H3, H4, tmp4, tmp3, tmp2, tmp1, &X);
reduce4(H, H2, H3, H4, tmp4, tmp3, tmp2, tmp1, &X);
reduce4(H, H2, H3, H4, tmp8, tmp7, tmp6, tmp5, &X);
__m128i H, H2, H3, H4, Y, T;
gfmul(H,H3,&H4);
reduce4(H, H2, H3, H4, tmp4, tmp3, tmp2, tmp1, &X);
reduce4(H, H2, H3, H4, tmp4, tmp3, tmp2, tmp1, &X);
ctx->state[4] = H4;