[pclmulqdq]

Oh, many embedded architectures don't have a cache hierarchy and instead place dynamic objects in one SRAM. Access latency is constant anywhere you go.