Performance of Coupled-Cluster Singles and Doubles on Modern Stream Processing Architectures.,Journal of Chemical Theory and Computation

当前位置： X-MOL 学术 › J. Chem. Theory Comput. › 论文详情

Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)

Performance of Coupled-Cluster Singles and Doubles on Modern Stream Processing Architectures.
Journal of Chemical Theory and Computation ( IF 5.7 ) Pub Date : 2020-06-20 , DOI: 10.1021/acs.jctc.0c00336
B Scott Fales _{1,

2} , Ethan R Curtis _{1,

2} , K Grace Johnson _{1,

2} , Dean Lahana _{1,

2} , Stefan Seritan _{1,

2} , Yuanheng Wang _{1,

2} , Hayley Weir _{1,

2} , Todd J Martínez _{1,

2} , Edward G Hohenstein _{1,

2}

Affiliation

We develop a new implementation of coupled-cluster singles and doubles (CCSD) optimized for the most recent graphical processing unit (GPU) hardware. We find that a single node with 8 NVIDIA V100 GPUs is capable of performing CCSD computations on roughly 100 atoms and 1300 basis functions in less than 1 day. Comparisons against massively parallel implementations of CCSD suggest that more than 64 CPU-based nodes (each with 16 cores) are required to match this performance.

中文翻译：

耦合集群单打和双打在现代流处理架构上的性能。

我们开发了针对最新图形处理单元（GPU）硬件进行了优化的耦合集群单双打（CCSD）的新实现。我们发现具有8个NVIDIA V100 GPU的单个节点能够在不到1天的时间内对大约100个原子和1300个基础函数执行CCSD计算。与CCSD大规模并行实现的比较表明，要实现此性能，需要超过64个基于CPU的节点（每个节点具有16个内核）。

更新日期：2020-07-14

点击分享查看原文

点击收藏

阅读更多本刊最新论文本刊介绍/投稿指南11