Enterprise: Breadth-First Graph Traversal on GPUs

被引:96
|
作者
Liu, Hang [1 ]
Huang, H. Howie [1 ]
机构
[1] George Washington Univ, Dept Elect & Comp Engn, Washington, DC 20052 USA
来源
PROCEEDINGS OF SC15: THE INTERNATIONAL CONFERENCE FOR HIGH PERFORMANCE COMPUTING, NETWORKING, STORAGE AND ANALYSIS | 2015年
关键词
INTERNET;
D O I
10.1145/2807591.2807594
中图分类号
TP301 [理论、方法];
学科分类号
081202 ;
摘要
The Breadth-First Search (BFS) algorithm serves as the foundation for many graph-processing applications and analytics workloads. While Graphics Processing Unit (GPU) offers massive parallelism, achieving high-performance BFS on GPUs entails efficient scheduling of a large number of GPU threads and effective utilization of GPU memory hierarchy. In this paper, we present Enterprise, a new GPU-based BFS system that combines three techniques to remove potential performance bottlenecks: (1) streamlined GPU threads scheduling through constructing a frontier queue without contention from concurrent threads, yet containing no duplicated frontiers and optimized for both top-down and bottomup BFS. (2) GPU workload balancing that classifies the frontiers based on different out-degrees to utilize the full spectrum of GPU parallel granularity, which significantly increases thread-level parallelism; and (3) GPU based BFS direction optimization quantifies the effect of hub vertices on direction-switching and selectively caches a small set of critical hub vertices in the limited GPU shared memory to reduce expensive random data accesses. We have evaluated Enterprise on a large variety of graphs with different GPU devices. Enterprise achieves up to 76 billion traversed edges per second (TEPS) on a single NVIDIA Kepler K40, and up to 122 billion TEPS on two GPUs that ranks No. 45 in the Graph 500 on November 2014. Enterprise is also very energy-efficient as No. 1 in the GreenGraph 500 (small data category), delivering 446 million TEPS per watt.
引用
收藏
页数:12
相关论文
共 50 条
  • [1] Specialization or Generalization: A Study on Breadth-First Graph Traversal on GPUs
    Zhong, Wenyong
    Cao, Yanxin
    Li, Jiawen
    Sun, Jianhua
    Chen, Hao
    PROCEEDINGS OF 2017 IEEE INTERNATIONAL CONFERENCE ON PROGRESS IN INFORMATICS AND COMPUTING (PIC 2017), 2017, : 294 - 301
  • [2] Breadth-First Traversal via Staging
    Gibbons, Jeremy
    Kidney, Donnacha Oisin
    Schrijvers, Tom
    Wu, Nicolas
    MATHEMATICS OF PROGRAM CONSTRUCTION (MPC 2022, 2022, 13544 : 1 - 33
  • [3] Mesh Analysis via Breadth-first Traversal
    O'Gwynn, David
    Johnstone, John
    PROCEEDINGS OF THE 48TH ANNUAL SOUTHEAST REGIONAL CONFERENCE (ACM SE 10), 2010, : 345 - 350
  • [4] iBFS: Concurrent Breadth-First Search on GPUs
    Liu, Hang
    Huang, H. Howie
    Hu, Yang
    SIGMOD'16: PROCEEDINGS OF THE 2016 INTERNATIONAL CONFERENCE ON MANAGEMENT OF DATA, 2016, : 403 - 416
  • [5] Load-Balanced Breadth-First Search on GPUs
    Zhu, Zhe
    Li, Jianjun
    Li, Guohui
    WEB-AGE INFORMATION MANAGEMENT, WAIM 2014, 2014, 8485 : 435 - 447
  • [6] Block Modeling Technology Research Based on Breadth-first Traversal
    Liu, Xinlei
    Yao, Changli
    Zhen, Yuanman
    NEAR-SURFACE GEOPHYSICS AND GEOHAZARDS, 2014, : 617 - 623
  • [7] XBFS: eXploring Runtime Optimizations for Breadth-First Search on GPUs
    Gaihre, Anil
    Wu, Zhenlin
    Yao, Fan
    Liu, Hang
    HPDC'19: PROCEEDINGS OF THE 28TH INTERNATIONAL SYMPOSIUM ON HIGH-PERFORMANCE PARALLEL AND DISTRIBUTED COMPUTING, 2019, : 121 - 131
  • [8] A new method for business process retrieval using breadth-first traversal
    Tan, Wenan
    Xie, Na
    Zhao, Lu
    Xu, Lida
    Sun, Yong
    ENTERPRISE INFORMATION SYSTEMS, 2020, 14 (01) : 83 - 109
  • [9] Research of FTP File Traversal Method based on Breadth-First Search
    Yan, Lei
    Ma, Hong-lin
    Kang, Li
    INTERNATIONAL CONFERENCE ON GRAPHIC AND IMAGE PROCESSING (ICGIP 2012), 2013, 8768
  • [10] SlimSell: A Vectorizable Graph Representation for Breadth-First Search
    Besta, Maciej
    Marending, Florian
    Solomonik, Edgar
    Hoefler, Torsten
    2017 31ST IEEE INTERNATIONAL PARALLEL AND DISTRIBUTED PROCESSING SYMPOSIUM (IPDPS), 2017, : 32 - 41