Improving the Performance of an Iterative Learning Control Program

Abstract

In this paper, we present a performance study of a five-axis iterative learning control program. Several approaches are proposed to enhance the performance. Eliminating redundant computation, parallelizing programs using OpenMP, and loop-invariant code motion are used to enhance the performance. For a single learning iteration with 5021 positions and using compiler option "-O2", the experimental results show that the execution time of the optimized program has a 4.84 times speedup compared to the non-optimized version.