Evaluation of Knight Landing High Bandwidth Memory for HPC Workloads

Evaluation of Knight Landing High Bandwidth Memory for HPC Workloads
复制标题

DOI:
10.1145/3149704.3149766
复制
发表时间:
2017-11
期刊:
Proceedings of the Seventh Workshop on Irregular Applications: Architectures and Algorithms
影响因子:
--
通讯作者:
S. Salehian;Yonghong Yan
S. Salehian;Yonghong Yan
中科院分区:
其他
文献类型:
--
作者:
S. Salehian;Yonghong Yan

文献摘要

相似文献

英特尔 Knight Landing (KNL) 众核芯片包含名为 MCDRAM 的 3D 堆栈内存,也称为高带宽内存 (HBM),适用于需要扩展到高线程数的并行应用。在本文中,我们对使用 DDR4 和 MCDRAM 时的 HPC 代理应用程序(包括 Lulesh、HPCG、AMG 和 Hotspot)的 KNL 进行了定量研究。结果表明,HBM 在 HPCG 中显着提高了内存密集型应用程序的性能,比 DDR4 提高了三倍,而 Lulesh 和 HPCG 分别提高了 40% 和 200%。对于选定的计算密集型应用,MCDRAM 相对于 DDR4 的性能优势从 2% 到 28% 不等。我们还观察到,MCDRAM 开始优于 DDR4 的交叉点约为 8 到 16 个线程。
The Intel Knight Landing (KNL) manycore chip includes 3D-stacked memory named MCDRAM, also known as High Bandwidth Memory (HBM) for parallel applications that needs to scale to high thread count. In this paper, we provide a quantitative study of the KNL for HPC proxy applications including Lulesh, HPCG, AMG, and Hotspot when using DDR4 and MCDRAM. The results indicate that HBM significantly improves the performance of memory intensive applications for as many as three times better than DDR4 in HPCG, and Lulesh and HPCG for as many as 40% and 200%. For the selected compute intensive applications, the performance advantage of MCDRAM over DDR4 varies from 2% to 28%. We also observed that the cross-points, where MCDRAM starts outperforming DDR4, are around 8 to 16 threads.