That depends on what you mean by "calculating math."

If you are talking about writing high-performance software, most high speed math libraries are written in C, and most C compilers provide optimization choices to help speed up math routines. Depending on whether or not you will be working in floating point also makes a difference -- there are libraries optimized for math co-processors and parallel processing, and for nearly any math functions you might desire. Fortran is still widely used because it is easier to program (for mathematicians) but not as efficient or flexible as C/C++.

If you are talking about the need to quickly implement a variety of math calculations, processing speed is not as important a factor as the ability to use existing routines that are known to do the job properly. There are many specialized math programming systems, most of them pricey, favored by different folks in different fields: Mathematica, Matlab, Gauss, etc.. R is a good choice for statistics, but is inefficient on large data sets.