Leveraging Task-Parallelism with OmpSs in ILUPACK's Preconditioned CG Method

Leveraging Task-Parallelism with OmpSs in ILUPACK's Preconditioned CG Method
复制标题

在 ILUPACK 的预处理 CG 方法中利用 OmpS 的任务并行性

DOI:
10.1109/sbac-pad.2014.24
复制
发表时间:
2014
期刊:
2014 IEEE 26th International Symposium on Computer Architecture and High Performance Computing
影响因子:
--
通讯作者:
E. Quintana
E. Quintana
中科院分区:
--
文献类型:
--
作者:
J. Aliaga;Rosa M. Badia;M. Barreda;M. Bollhöfer;E. Quintana

文献摘要

被引文献

相似文献

本文描述了如何利用ILUPACK的多级预处理CG方法在多线程处理器上有效地利用任务并行性求解稀疏线性系统。使用一对数据结构,我们捕获了在该方法中两个最具挑战性的操作(预置条件的计算及其应用程序)中出现的任务依赖关系,并将此信息传递给omps运行时,然后该运行时可以实现整个求解器的正确且有效的调度。在配备英特尔和AMD处理器的高端多核平台上,我们的结果显示了显著的性能提升,表明omps提供了一种高效且接近无缝的方法,可以在像ILUPACK这样的复杂科学代码中利用并发性。
In this paper we describe how to efficiently exploit task parallelism for the solution of sparse linear systems on multithreaded processors via ILUPACK's multi-level preconditioned CG method. Using a pair of data structures, we capture the task dependencies that appear in the two most challenging operations in the method (calculation of the preconditioned and its application), passing this information to the OmpSs runtime which can then implement a correct and efficient schedule of the entire solver. Our results with high-end multicore platforms equipped with Intel and AMD processors report significant performance gains, demonstrating that OmpSs provides an efficient and close-to seamless means to leverage the concurrency in a complex scientific code like ILUPACK.