This is awesome! In case you're wondering, the latency of _mm_aesenc_si128 on Skylake and above is only 4 clock cycles