Comment
Comment by throwaway4good
What is the point of making the matrix multiplication itself multithreaded (other than benchmarking)? Wouldn't it be more beneficial in practice to have the multithreadedness in the algorithm that use the multiplication?