Parallel Optimization Research Of Method Of Characteristics Based On Sunway Bluelight Ⅱ Supercomputer

In recent years,with the continuous development and upgrading of computer technology,it can use multiple computers and storage devices to calculate through the aggregation structure,which is called high performance computing.High performance computing has become a necessity for scientific research in various fields,for example,in physical simulation,engineering design,data analysis,artificial intelligence,and other research areas.With the development of multi-core and many-core processor technology,high performance computers have more hardware computing resources.However,hardware improvement cannot bring the same proportion of software performance improvement.Therefore,parallel computing has become one of the focuses in the field of high-performance computing.In order to improve the efficiency of programs,researchers are committed to implementing the parallel computing of the program,and matching the hardware of the high-performance computer from the software level.Parallel computing research is of great significance to promote the development and progress of science and technology.Sunway Bluelight Ⅱ supercomputer is a high-performance computer developed independently by China,and is also the latest generation of Sunway series supercomputer.Its peak performance can reach 3.13 PFLOPS.It can not only accelerate the execution speed of computing tasks,but also improve the utilization of computing resources,and reduce energy consumption and cost.At present,it is widely used in scientific computing,artificial intelligence,big data analysis and other fields.With the development of nuclear energy technology,reactor physical calculations have higher requirements for calculation accuracy and speed,and it has become a necessary choice to use high-performance computers for reactor simulation calculations.The method of characteristics(MOC)is currently recognized as the preferred method for simulating neutron transport in the nuclear reactor core,because it can handle arbitrary geometric structures.Based on the architecture of Sunway many-core processor and Sunway Bluelight Ⅱ supercomputer,this thesis proposes a fine grained and universal two-level parallelization,including thread-level parallelization and process-level parallelization.In the thread-level parallelization,the methods such as job pipeline optimization,load balancing on CPEs,and I/O optimization are proposed for acceleration.In the process-level parallelization,a mapping method from software to hardware is proposed.This method can make full use of the hardware of Sunway supercomputers,and improve the computing efficiency and data transmission efficiency.For the first time,the Open MOC program is transplanted and parallel optimized on the Sunway supercomputers,which enriched the application ecology of Sunway supercomputers.Compared with the original program,the thread-level parallelization can achieve up to 5.4x speedup,and the process-level parallelization can achieve up to 4.8x speedup.Finally,the two-level parallelization can achieve up to 18.6x speedup.Moreover,our parallelization is capable to run on more than 3,750 processes of Sunway Bluelight Ⅱ supercomputer with good strong and weak scalability.