In this paper, we describe a neural network program that was originally developed in C, then ported to Ada 2005. We explain several simple modifications to the Ada code that reduce the overhead from 76% to 0%. These modifications could provide significant performance gains to other applications, allowing them to combine the safety of Ada with the speed of C. such a complicated data structure. In Section 3, we explain how we modified the code in translating it to Ada. Section 4 describes how we made simple changes to the Ada implementation to eliminate the additional overhead. Section 5 provides conclusions and insights for future projects.
/lp/association-for-computing-machinery/timing-neural-networks-in-c-and-ada-s1Au6BgMVH