High-performance external stencil computations using user-controllable I/O

被引:0
|
作者
Lee, JS [1 ]
Kim, MJ [1 ]
机构
[1] ETRI, Comp & Software Technol Labs, Data Engn Dept, Yusong Gu, Taejon 305350, South Korea
关键词
D O I
暂无
中图分类号
TP3 [计算技术、计算机技术];
学科分类号
0812 ;
摘要
The UPIO (User-controllable Parallel I/O) we proposed in [10] extends the abstraction of a linear file model into an n-dimensional file model and removes the limitations of the linear file model, allowing users to control the layout of data blocks across disks and aggregating disk bandwidth through UPIO's interfaces. This enables users to do plan I/Os, computations, communications, and data reuse effectively in many scientific and engineering applications. These applications can be characterized by the corresponding stencil and implemented using stencil algorithms. In this paper we show how well UPIO produces high-performance external stencil codes by designing I/O and communication-efficient external Laplace equation solver algorithms and exploring the effects of UPIO with the codes.
引用
收藏
页码:748 / 755
页数:8
相关论文
共 50 条
  • [1] High-performance external computations using user-controllable I/O
    Lee, JS
    Ko, S
    Ranka, S
    Min, BE
    FIRST MERGED INTERNATIONAL PARALLEL PROCESSING SYMPOSIUM & SYMPOSIUM ON PARALLEL AND DISTRIBUTED PROCESSING, 1998, : 303 - 307
  • [2] User-controllable parallel I/O for scientific applications
    Lee, JS
    Song, Y
    1998 INTERNATIONAL CONFERENCE ON PARALLEL AND DISTRIBUTED SYSTEMS, PROCEEDINGS, 1998, : 792 - 799
  • [3] Logical disks: User-Controllable I/O for scientific applications
    Lee, JS
    Kim, JM
    Berra, PB
    Ranka, S
    EIGHTH IEEE SYMPOSIUM ON PARALLEL AND DISTRIBUTED PROCESSING, PROCEEDINGS, 1996, : 340 - 347
  • [4] User-controllable coherence for high performance shared memory multiprocessors
    McCurdy, C
    Fischer, C
    ACM SIGPLAN NOTICES, 2003, 38 (10) : 73 - 83
  • [5] I/O and memory-efficient matrix multiplication with user-controllable parallel I/O
    Lee, JS
    Park, SY
    Berra, PB
    Ranka, S
    1997 INTERNATIONAL CONFERENCE ON PARALLEL AND DISTRIBUTED SYSTEMS, PROCEEDINGS, 1997, : 59 - 66
  • [6] Guest Editors' Note Special Issue On High-Performance Stencil Computations
    Grosslinger, Armin
    Kostler, Harald
    PARALLEL PROCESSING LETTERS, 2014, 24 (03)
  • [7] User-controllable mesh segmentation using shape harmonic signature
    Miao, Yongwei
    Feng, Jieqing
    Wang, Jinrong
    Jin, Xiaogang
    PROGRESS IN NATURAL SCIENCE-MATERIALS INTERNATIONAL, 2009, 19 (04) : 471 - 478
  • [8] User-controllable mesh segmentation using shape harmonic signature
    Yongwei Miao ab Jieqing Fenga Jinrong Wangac Xiaogang Jin a a State Key Laboratory of CAD CG Zhejiang University Hangzhou China b College of Science Zhejiang University of Technology Hangzhou China c Information Science Engineering School Hangzhou Normal University Hangzhou China
    ProgressinNaturalScience, 2009, 19 (04) : 471 - 478
  • [9] Domain-Specific Optimization and Generation of High-Performance GPU Code for Stencil Computations
    Rawat, Prashant Singh
    Vaidya, Miheer
    Sukumaran-Rajam, Aravind
    Ravishankar, Mahesh
    Grover, Vinod
    Rountev, Atanas
    Pouchet, Louis-Noel
    Sadayappan, P.
    PROCEEDINGS OF THE IEEE, 2018, 106 (11) : 1902 - 1920
  • [10] High-Performance Derivative Computations using CoDiPack
    Sagebaum, Max
    Albring, Tim
    Gauger, Nicolas R.
    ACM TRANSACTIONS ON MATHEMATICAL SOFTWARE, 2019, 45 (04):