Research on MpiP Performance Based on Parallel Programs
Chuanying Li, Zhe Quan, Xiong Xiao, Dajiang Yi · 2018
With the wide application of high-performance computing, more and more users use parallel execution programs to improve the execution speed of large-scale programs and improve overall execution performance, which is accompanied by a well-portable messaging interface. widely used. We use MPI-based performance analysis tool mpiP to analyze the performance of the program, and use the matrix multiplication example model as a test case, carry out a large number of targeted experiments, and improve the performance of mpiP in performance analysis. The main work we do is to add data sorting module, heap memory footprint collection module, generate timeline module and I/O port data collection module to improve the overall performance analysis efficiency, and also to obtain the entire program execution process more conveniently. The execution status of each process.