Analysis of Parallel Computing Environment Overhead of OpenMP for Loop with Multi-core Processors

This paper analyze the parallel computing environment overhead of OpenMP for loop with multi-core processors including the case of data-race. The different solutions of data-race are discussed in present paper, such as critical pragma, atomic pragma and reduction clause. A new method is also presented in this paper which produces the least parallel environment overhead compared with other methods and the codes are presented too. The conclusions and discussions are beneficial to save computing time and improve the computational efficiency.