SC14: International Conference for High Performance Computing, Networking, Storage and Analysis最新文献_第2页

DISC: A Domain-Interaction Based Programming Model with Support for Heterogeneous Execution 支持异构执行的基于域交互的编程模型

SC14: International Conference for High Performance Computing, Networking, Storage and Analysis Pub Date : 2014-11-16 DOI: 10.1109/SC.2014.76

Mehmet Can Kurt, G. Agrawal

引用次数: 5

CYPRESS: Combining Static and Dynamic Analysis for Top-Down Communication Trace Compression CYPRESS:结合静态和动态分析的自顶向下通信跟踪压缩

SC14: International Conference for High Performance Computing, Networking, Storage and Analysis Pub Date : 2014-11-16 DOI: 10.1109/SC.2014.17

Jidong Zhai, Jianfei Hu, Xiongchao Tang, Xiaosong Ma, Wenguang Chen

{"title":"CYPRESS: Combining Static and Dynamic Analysis for Top-Down Communication Trace Compression","authors":"Jidong Zhai, Jianfei Hu, Xiongchao Tang, Xiaosong Ma, Wenguang Chen","doi":"10.1109/SC.2014.17","DOIUrl":"https://doi.org/10.1109/SC.2014.17","url":null,"abstract":"Communication traces are increasingly important, both for parallel applications' performance analysis/optimization, and for designing next-generation HPC systems. Meanwhile, the problem size and the execution scale on supercomputers keep growing, producing prohibitive volume of communication traces. To reduce the size of communication traces, existing dynamic compression methods introduce large compression overhead with the job scale. We propose a hybrid static-dynamic method that leverages information acquired from static analysis to facilitate more effective and efficient dynamic trace compression. Our proposed scheme, Cypress, extracts a program communication structure tree at compile time using inter-procedural analysis. This tree naturally contains crucial iterative computing features such as the loop structure, allowing subsequent runtime compression to \"fill in\", in a \"top-down\" manner, event details into the known communication template. Results show that Cypress reduces intra-process and inter-process compression overhead up to 5× and 9× respectively over state-of-the-art dynamic methods, while only introducing very low compiling overhead.","PeriodicalId":275261,"journal":{"name":"SC14: International Conference for High Performance Computing, Networking, Storage and Analysis","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2014-11-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131329974","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 23

Fail-in-Place Network Design: Interaction Between Topology, Routing Algorithm and Failures 故障就地网络设计:拓扑、路由算法和故障之间的交互作用

SC14: International Conference for High Performance Computing, Networking, Storage and Analysis Pub Date : 2014-11-16 DOI: 10.1109/SC.2014.54

Jens Domke, T. Hoefler, S. Matsuoka

引用次数: 31

Compiler Techniques for Massively Scalable Implicit Task Parallelism 大规模可扩展隐式任务并行的编译器技术

SC14: International Conference for High Performance Computing, Networking, Storage and Analysis Pub Date : 2014-11-16 DOI: 10.1109/SC.2014.30

Timothy G. Armstrong, J. Wozniak, M. Wilde, Ian T Foster

{"title":"Compiler Techniques for Massively Scalable Implicit Task Parallelism","authors":"Timothy G. Armstrong, J. Wozniak, M. Wilde, Ian T Foster","doi":"10.1109/SC.2014.30","DOIUrl":"https://doi.org/10.1109/SC.2014.30","url":null,"abstract":"Swift/T is a high-level language for writing concise, deterministic scripts that compose serial or parallel codes implemented in lower-level programming models into large-scale parallel applications. It executes using a data-driven task parallel execution model that is capable of orchestrating millions of concurrently executing asynchronous tasks on homogeneous or heterogeneous resources. Producing code that executes efficiently at this scale requires sophisticated compiler transformations: poorly optimized code inhibits scaling with excessive synchronization and communication. We present a comprehensive set of compiler techniques for data-driven task parallelism, including novel compiler optimizations and intermediate representations. We report application benchmark studies, including unbalanced tree search and simulated annealing, and demonstrate that our techniques greatly reduce communication overhead and enable extreme scalability, distributing up to 612 million dynamically load balanced tasks per second at scales of up to 262,144 cores without explicit parallelism, synchronization, or load balancing in application code.","PeriodicalId":275261,"journal":{"name":"SC14: International Conference for High Performance Computing, Networking, Storage and Analysis","volume":"8 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2014-11-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116861617","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 50

Nonblocking Epochs in MPI One-Sided Communication MPI单侧通信中的非阻塞时代

SC14: International Conference for High Performance Computing, Networking, Storage and Analysis Pub Date : 2014-11-16 DOI: 10.1109/SC.2014.44

Judicael A. Zounmevo, Xin Zhao, P. Balaji, W. Gropp, A. Afsahi

引用次数: 6

A Computation- and Communication-Optimal Parallel Direct 3-Body Algorithm 一种计算和通信最优的并行直接三体算法

SC14: International Conference for High Performance Computing, Networking, Storage and Analysis Pub Date : 2014-11-16 DOI: 10.1109/SC.2014.35

Penporn Koanantakool, K. Yelick

引用次数: 11

Two-Choice Randomized Dynamic I/O Scheduler for Object Storage Systems 对象存储系统的双选择随机动态I/O调度程序

SC14: International Conference for High Performance Computing, Networking, Storage and Analysis Pub Date : 2014-11-16 DOI: 10.1109/SC.2014.57

Dong Dai, Yong Chen, D. Kimpe, R. Ross

{"title":"Two-Choice Randomized Dynamic I/O Scheduler for Object Storage Systems","authors":"Dong Dai, Yong Chen, D. Kimpe, R. Ross","doi":"10.1109/SC.2014.57","DOIUrl":"https://doi.org/10.1109/SC.2014.57","url":null,"abstract":"Object storage is considered a promising solution for next-generation (exascale) high-performance computing platform because of its flexible and high-performance object interface. However, delivering high burst-write throughput is still a critical challenge. Although deploying more storage servers can potentially provide higher throughput, it can be ineffective because the burst-write throughput can be limited by a small number of stragglers (storage servers that are occasionally slower than others). In this paper, we propose a two-choice randomized dynamic I/O scheduler that schedules the concurrent burst-write operations in a balanced way to avoid stragglers and hence achieve high throughput. The contributions in this study are threefold. First, we propose a two-choice randomized dynamic I/O scheduler with collaborative probe and preassign strategies. Second, we design and implement a redirect table and metadata maintainer to address the metadata management challenge introduced by dynamic I/O scheduling. Third, we evaluate the proposed scheduler with both simulation tests and experimental tests in an HPC cluster. The evaluation results confirm the scalability and performance benefits of the proposed I/O scheduler.","PeriodicalId":275261,"journal":{"name":"SC14: International Conference for High Performance Computing, Networking, Storage and Analysis","volume":"251 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2014-11-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121341660","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 18

The DRIHM Project: A Flexible Approach to Integrate HPC, Grid and Cloud Resources for Hydro-Meteorological Research DRIHM项目:整合HPC、网格和云资源用于水文气象研究的灵活方法

SC14: International Conference for High Performance Computing, Networking, Storage and Analysis Pub Date : 2014-11-16 DOI: 10.1109/SC.2014.49

D. D'Agostino, A. Clematis, A. Galizia, A. Quarati, E. Danovaro, Luca Roverelli, Gabriele Zereik, D. Kranzlmüller, Michael Schiffers, N. Felde, Christian Straube, Olivier Caumontz, E. Richard, L. Garrote, Quillon Harphamk, H.R.A. Jagers, V. Dimitrijevic, L. Dekic, Elisabetta Fiorizz, F. Delogu, A. Parodi

{"title":"The DRIHM Project: A Flexible Approach to Integrate HPC, Grid and Cloud Resources for Hydro-Meteorological Research","authors":"D. D'Agostino, A. Clematis, A. Galizia, A. Quarati, E. Danovaro, Luca Roverelli, Gabriele Zereik, D. Kranzlmüller, Michael Schiffers, N. Felde, Christian Straube, Olivier Caumontz, E. Richard, L. Garrote, Quillon Harphamk, H.R.A. Jagers, V. Dimitrijevic, L. Dekic, Elisabetta Fiorizz, F. Delogu, A. Parodi","doi":"10.1109/SC.2014.49","DOIUrl":"https://doi.org/10.1109/SC.2014.49","url":null,"abstract":"The distributed research infrastructure for hydrometeorology (DRIHM) project focuses on the development of an e-Science infrastructure to provide end-to-end hydro meteorological research (HMR) services (models, data, and post processing tools) by exploiting HPC, Grid and Cloud facilities. In particular, the DRIHM infrastructure supports the execution and analysis of high-resolution simulations through the definition of workflows composed by heterogeneous HMR models in a scalable and interoperable way, while hiding all the low level complexities. This contribution gives insights into best practices adopted to satisfy the requirements of an emerging multidisciplinary scientific community composed of earth and atmospheric scientists. To this end, DRIHM supplies innovative services leveraging high performance and distributed computing resources. Hydro meteorological requirements shape this IT infrastructure through an iterative \"learning-by-doing\" approach that permits tight interactions between the application community and computer scientists, leading to the development of a flexible, extensible, and interoperable framework.","PeriodicalId":275261,"journal":{"name":"SC14: International Conference for High Performance Computing, Networking, Storage and Analysis","volume":"172 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2014-11-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121558063","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 23

Fast Parallel Computation of Longest Common Prefixes 最长公共前缀的快速并行计算

SC14: International Conference for High Performance Computing, Networking, Storage and Analysis Pub Date : 2014-11-16 DOI: 10.1109/SC.2014.37

Julian Shun

引用次数: 21

Lattice QCD with Domain Decomposition on Intel® Xeon Phi Co-Processors 基于Intel®Xeon Phi协处理器的点阵QCD域分解

SC14: International Conference for High Performance Computing, Networking, Storage and Analysis Pub Date : 2014-11-16 DOI: 10.1109/SC.2014.11

S. Heybrock, B. Joó, Dhiraj D. Kalamkar, M. Smelyanskiy, K. Vaidyanathan, T. Wettig, P. Dubey

引用次数: 36