首页> 外文会议>Progress in electromagnetics research symposium >Parallel Performance of the Multi-threaded ICCG Solver in Electromagnetic Finite Element Analyses on the Latest Processors
【24h】

Parallel Performance of the Multi-threaded ICCG Solver in Electromagnetic Finite Element Analyses on the Latest Processors

机译:最新处理器在电磁有限元分析中的多线程ICCG解算器并行性能

获取原文

摘要

The performance of multithreaded incomplete Cholesky conjugate gradient solvers in finite element electromagnetic field analysis is analyzed. We compare the parallel performance of the additive Schwarztype incomplete Cholesky preconditioned conjugate gradient (AS-ICCG) solver and the parallelized incomplete Cholesky conjugate gradient solver using algebraic block multicolor ordering (ABMC-ICCG) on the latest processors. A numerical test of the interior-permanent-magnet motor model analysis shows that the ABMC-ICCG solver has faster computation speed than the AS-ICCG solver in 16-core executions on an Intel~® Xeon~® E5 processor, and attains a 7.6-fold speedup relative to the sequential solver. Moreover, we analyze performance on an Intel~® Xeon Phi™coprocessor and attain a 35-fold speedup relative to the sequential solver on the same coprocessor using the ABMC-ICCG solver.
机译:分析了多线程不完全Cholesky共轭梯度求解器在有限元电磁场分析中的性能。我们在最新处理器上比较了使用代数块多色排序(ABMC-ICCG)的加性Schwarztype不完全Cholesky共轭梯度求解器(AS-ICCG)和并行不完全Cholesky共轭梯度求解器的并行性能。对内部永磁电机模型分析的数值测试表明,在英特尔®至强®E5处理器上的16核执行中,ABMC-ICCG求解器的计算速度比AS-ICCG求解器快。 -相对于顺序求解器的加速倍数。而且,我们分析了英特尔®至强融核™协处理器的性能,并使用ABMC-ICCG求解器将同一个协处理器上的顺序求解器的速度提高了35倍。

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