The Commonly Used Algorithm Optimization Of The Differential Equation’s Based On The Multi-core CPU And The General Purpose GPU In The MATLAB

With a variety of new architecture constantly emerging,high performancesupercomputer architecture is reforming.Using Multi core Central Processing Unit andGeneral-purpose Graphic Processing Unit computing on graphics processing unitscombining heterogeneous mixed structure computing platform is considered a feasibleway.In heterogeneous parallel operation with general GPU and Multi core CPU, to complete oflarge quantities of high performance computing task with high speed and coordinating，CPU and GPU shall perform their functions respectively.In addition, except for logisticmanagement, CPU should also undertake part in scientific calculation task.The main task of this research is designed by using MATLAB,to discuss thedifferential equation (Group) common algorithms, the corresponding algorithmoptimization problems with general GPU and Multi core CPU.The article choose four differential equation (Group) common algorithms todiscuss,finite difference method, Runge-Kutta method, Cronk-Nicholson method andshooting method.The algorithm optimization problems with general GPU and Multi coreCPU, in order to give a general algorithm optimization guidance, so as to improve theresearch efficiency of the super computer in science and engineering when using a lot ofdifferential equation (Group) in the different structures.The framework of the platform needs to compare three times,serial arithmetic withsingle-core CPU;parallel operation with multi-core CPU;parallel operation with generalGPU and Multi core CPU.and compares the experimental results.With the CPU, GPU and Advanced Mathematics continuous development,people’swork and the scientific research is having a great change, the efficiency is improved,longcalculation time couse by the huge operations is slowly reduced.