Is there a libm (libmath) that is optimised for ARM(v6) processors?
I was looking at the GNU implementation and it doesn't seem to be optimised (but it does have x86 ones). It seems that most implementations of libm do not have ARM specific optimisation.