[Symmetry]

Well, generally optimized numerical libraries like BLAS are faster than what you write yourself. But a prolbem with calling them from C is that you don't get loop fusion, each call has to go through the data before the next one starts. That isn't true with Haskell's lazy evaluation so it can be faster in some cases.

Of course, C++'s eigen does loop fusion for you with template magic so to go really fast you probably want that.