You have to use at least one cache line for memcpy. But you need more for pipelining the accesses probably to get good speed. Benchmarking memcpy though is comp

You have to use at least one cache line for memcpy. But you need more for pipelining the accesses probably to get good speed. Benchmarking memcpy though is complicated.