The paper deals with the analysis of the implementation of special mathematical functions in popular applied and standard libraries of C and C++. Implementations of the error function and complementary error function in the C and C++ standard and common free applied libraries were tested. Intervals of the argument, in which the value of the absolute calculation errors of the considered functions are significant and cause the additional calculations errors, have been calculated. To conclude, for gcc compilers, to reduce additional computational errors without reducing (or with acceptable reducing) their performance is preferable to use quadmath and boost libraries.
special functions, standard and applied libraries, computational error, the error function, performance computing