I am trying to use foreach to parallelize some matrix calculations. However, I found it gives similar performance to apply and for loop, despite I used 10 cores.