Parallel Optimization computing in Universal Combinatorial Coding

摘要

Universal combinatorial coding derives from the combination principle and its core is ordinal calculation. The solution of ordinal is a process of computing permutation and combination values. But permutation and combination computing is very time-consuming. As a result, adopting optimized methods or parallel techniques to increase ordinal calculation efficiency is very necessary. Parallel optimization algorithms of the ordinal in universal combinatorial coding are advanced based on CPU multi-core and multi-thread techniques in this paper, including prime splitting and unequal parallel algorithms. Firstly, the thesis simplifies the multiplication and division calculations of the ordinal parallel computing through prime splitting optimization. Secondly, the thesis analyzes the deficiency of equal parallel that the unbalanced computing time of each part will lead to inefficient parallel computing of the ordinal, and then proposes the unequal parallel algorithm to shorten the ordinal number calculation. The experiment results of prime splitting and unequal parallel algorithm proved that two parallel optimizations are better in improving the efficiency of ordinal computing.