首页> 外文期刊>Applied Computational Electromagnetics Society journal >Parallel Higher-Order Method of Moments with Efficient Out-of-GPU Memory Schemes for Solving Electromagnetic Problems
【24h】

Parallel Higher-Order Method of Moments with Efficient Out-of-GPU Memory Schemes for Solving Electromagnetic Problems

机译:具有有效的GPU外内存方案的并行矩高阶方法,可以解决电磁问题

获取原文
获取原文并翻译 | 示例
           

摘要

A distributed parallel Higher-order Method of Moments (HoMoM) for solving electromagnetic problems on CPU/GPU clusters is presented. An MPI/OpenMP/CUDA parallel framework based on the GPU context technique is designed. An out-of-GPU memory scheme is employed to break the limitation of the GPU memory. To improve the performance of data transferring between main memory and GPU memory, an overlapping scheme based on asynchronous technique and CUDA streams is adopted. In comparison with the parallel CPU version only, numerical results including a metallic airplane and an airborne array with dielectric structures demonstrate the high performance of the proposed method.
机译:提出了一种分布式并行高阶矩量法(HoMoM),用于解决CPU / GPU集群上的电磁问题。设计了基于GPU上下文技术的MPI / OpenMP / CUDA并行框架。采用GPU外内存方案可以打破GPU内存的限制。为了提高主内存和GPU内存之间数据传输的性能,采用了基于异步技术和CUDA流的重叠方案。与仅并行CPU版本相比,包括金属飞机和带有电介质结构的机载阵列的数值结果证明了该方法的高性能。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号