Implementing Lattice QCD Application with XcalableACC Language on Accelerated Cluster
Implementing Lattice QCD Application with XcalableACC Language on Accelerated Cluster
复制标题
在加速集群上使用 XcalableACC 语言实现莱迪思 QCD 应用
DOI：
10.1109/cluster.2017.58
复制
发表时间：
2017
期刊：
影响因子：
--
通讯作者：
M. Sato
中科院分区：
文献类型：
--
作者：
M. Nakao;H. Murai;H. Iwashita;Akihiro Tabuchi;T. Boku;M. Sato
配备加速器的分布式内存系统的加速簇已在各个字段中使用。对于加速群集，程序员通常通过MPI和CUDA（MPI+CUDA）的组合来实施其应用程序。但是，该方法面临编程复杂性问题。本文介绍了XcalableACC（XACC）语言，该语言是XcalableMP（XMP）和OpenACC的混合模型。 XMP是用于分布式内存系统的基于指令的语言，而OpenACC也是一种基于指令的语言。 XACC使程序员可以轻松地开发有关加速群集的应用程序。为了评估XACC性能和生产率水平，我们使用XACC在64个计算节点和256 GPU上实施了晶格量子染色体动力学（晶格QCD）应用，发现其性能与MPI+CUDA的性能几乎相同。此外，我们发现，XACC与MPI+CUDA相比，从串行晶格QCD代码的变化要少得多，以实现并行晶格QCD代码。
Accelerated clusters, which are distributed memory systems equipped with accelerators, have been used in various fields. For accelerated clusters, programmers often implement their applications by a combination of MPI and CUDA (MPI+CUDA). However, the approach faces programming complexity issues. This paper introduces the XcalableACC (XACC) language, which is a hybrid model of XcalableMP (XMP) and OpenACC. While XMP is a directive-based language for distributed memory systems, OpenACC is also a directive-based language for accelerators. XACC enables programmers to develop applications on accelerated clusters with ease. To evaluate XACC performance and productivity levels, we implemented a lattice quantum chromodynamics (Lattice QCD) application using XACC on 64 compute nodes and 256 GPUs and found its performance was almost the same as that of MPI+CUDA. Moreover, we found that XACC requires much less change from the serial Lattice QCD code than MPI+CUDA to implement the parallel Lattice QCD code.
OpenACC 应用程序的基于指令的数据布局自动调整
DOI：
--
发表时间：
2017
期刊：
影响因子：
--
作者：
Tetsuya Hoshino;Naoya Maruyama;Satoshi Matsuoka
通讯作者：
Satoshi Matsuoka