基于GPU的微波器件微放电阈值快速粒子模拟
|
摘要
本文提出了一种基于GPU加速的微波器件微放电阈值三维粒子模拟算法,将微放电粒子模拟软件MSAT中的粒子推进求解算法利用GPU加速执行。为了验证GPU加速程序的正确性,以微波阶梯阻抗变换器为研究对象,采用GPU加速程序模拟其微放电过程,并将计算的结果与原始程序进行比对,结果表明,二者计算的粒子数目曲线存在细微的差异,但其整体变化趋势几乎一致。此外,为了说明GPU加速程序的高效性,采用该程序模拟不同初始加载粒子数目情况下的微放电过程并记录其耗时,结果表明,GPU加速程序的计算效率提高了6倍。
A three-dimensional particle-in-cell(PIC)algorithm based on GPU is proposed for calculating the multipactor threshold.In this code,the module of electron motion is running on the GPU platform.To validate the GPU code,the multipactor processes of impedance transformer at different input powers are studied by the GPU and the CPU code separately.Simulation results from these two codes agree well with each other.In addition,the effect of the number of initial macro-particles on the computational efficiency is studied.The simulation result shows that when the initial number of particles is small,the acceleration performance of the GPU code is not significantly improved compared to that of the CPU code.However,the speed up factor increases with the initial number of particles.When the number of particles is about 640000,the speed up factor increases by 6 times.
A three-dimensional particle-in-cell(PIC)algorithm based on GPU is proposed for calculating the multipactor threshold.In this code,the module of electron motion is running on the GPU platform.To validate the GPU code,the multipactor processes of impedance transformer at different input powers are studied by the GPU and the CPU code separately.Simulation results from these two codes agree well with each other.In addition,the effect of the number of initial macro-particles on the computational efficiency is studied.The simulation result shows that when the initial number of particles is small,the acceleration performance of the GPU code is not significantly improved compared to that of the CPU code.However,the speed up factor increases with the initial number of particles.When the number of particles is about 640000,the speed up factor increases by 6 times.
引文
[1]Vaughan J R M.Multipactor[J].Electron Devices IEEE Transactions on,1988,35(7):1172-1180.
[2]Kishek R A,Lau Y Y,Ang L K,et al.Multipactor Discharge on Metals and Dielectrics:Historical Review and Recent Theories[J].Physics of Plasmas,1998,5(5):2120-2126.
[3]You J W,Wang H G,Zhang J F,et al.Accurate Numerical Method for Multipactor Analysis in Microwave Devices[J].IEEE Transactions on Electron Devices,2014,61(5):1546-1552.
[4]You J W,Wang H G,Zhang J F,et al.Accurate Numerical Analysis of Nonlinearities Caused by Multipactor in Microwave Devices[J].IEEE Microwave&Wireless Components Letters,2014,24(APMC2008):730-732.
[5]Li Y,Cui W Z,Wang H G.Simulation Investigation of Multipactor in Metal Components for Space Application with an Improved Secondary Emission Model[J].Physics of Plasmas,2015,22(5):1172-2126.
[6]王洪广,翟永贵,李记肖,等.基于频域电磁场的微波器件微放电阈值快速粒子模拟[J].物理学报,2016,65(23):275-281.
[7]刘雷,李永东,王瑞,等.微波阶梯阻抗变换器低气压电晕放电粒子模拟[J].物理学报,2013,62(2):25201-025201.
[8]Garland M,Grand SL,Nickolls J,et al.Parallel Computing Experiences with CUDA[J].IEEE Micro,2008,28(4):13-27.
[9]Nickolls J,Buck I,Garland M,et al.Scalable Parallel Programming with CUDA[J].ACM Queue,2008,6(2):40-53.
[2]Kishek R A,Lau Y Y,Ang L K,et al.Multipactor Discharge on Metals and Dielectrics:Historical Review and Recent Theories[J].Physics of Plasmas,1998,5(5):2120-2126.
[3]You J W,Wang H G,Zhang J F,et al.Accurate Numerical Method for Multipactor Analysis in Microwave Devices[J].IEEE Transactions on Electron Devices,2014,61(5):1546-1552.
[4]You J W,Wang H G,Zhang J F,et al.Accurate Numerical Analysis of Nonlinearities Caused by Multipactor in Microwave Devices[J].IEEE Microwave&Wireless Components Letters,2014,24(APMC2008):730-732.
[5]Li Y,Cui W Z,Wang H G.Simulation Investigation of Multipactor in Metal Components for Space Application with an Improved Secondary Emission Model[J].Physics of Plasmas,2015,22(5):1172-2126.
[6]王洪广,翟永贵,李记肖,等.基于频域电磁场的微波器件微放电阈值快速粒子模拟[J].物理学报,2016,65(23):275-281.
[7]刘雷,李永东,王瑞,等.微波阶梯阻抗变换器低气压电晕放电粒子模拟[J].物理学报,2013,62(2):25201-025201.
[8]Garland M,Grand SL,Nickolls J,et al.Parallel Computing Experiences with CUDA[J].IEEE Micro,2008,28(4):13-27.
[9]Nickolls J,Buck I,Garland M,et al.Scalable Parallel Programming with CUDA[J].ACM Queue,2008,6(2):40-53.