论文部分内容阅读
针对ADS颗粒靶概念的研究和设计,中国科学院近代物理研究所自主研发了蒙特卡罗模拟软件GMT。为了提高GMT程序的计算效率,研究了MPI在GMT中的应用和发展,实现了大规模随机数在进程中的随机分配,并采用快速读写文件的方式替代了MPI相关数据通信函数,极大地提高了计算效率。并研究了不同规模计算实例进程数、加速比、效率之间的关系,确定了最大加速进程数及并行效率最高时的进程数,为科研工作者在计算资源和计算效率之间选择最优计算方案提供了科学依据。MPI在GMT中的成功应用使计算资源得到了充分、高效的利用,极大地提高了计算效率,解决了蒙特卡罗方法中大规模事件模拟计算时间长、计算不稳定等问题,在散裂靶大规模扫描计算中发挥了重要的作用。
For the research and design of the concept of particle target of ADS, the Monte Carlo simulation software GMT has been independently developed by Institute of Modern Physics, Chinese Academy of Sciences. In order to improve the computational efficiency of GMT program, this paper studies the application and development of MPI in GMT, realizes the random distribution of large-scale random numbers in the process, replaces the MPI-related data communication functions by means of fast read and write files, Increased computational efficiency. And studied the relationship between the number of processes, speedups and efficiencies of different scale calculation examples, determined the maximum number of acceleration processes and the number of processes with the highest parallel efficiency, and selected the optimal calculation for scientific researchers between computing resources and computing efficiency Program provides a scientific basis. The successful application of MPI in GMT has made full and efficient use of computing resources, which has greatly improved the computational efficiency and solved the problems of long time-consuming and computationally-large-scale simulation of large-scale events in Monte Carlo methods. In the spallation target Large-scale scanning calculation has played an important role.