XBFS: eXploring Runtime Optimizations for Breadth-First Search on GPUs

被引:25
|
作者
Gaihre, Anil [1 ]
Wu, Zhenlin [1 ]
Yao, Fan [2 ]
Liu, Hang [1 ]
机构
[1] UMass Lowell, Lowell, MA 01854 USA
[2] Univ Cent Florida, Orlando, FL 32816 USA
来源
HPDC'19: PROCEEDINGS OF THE 28TH INTERNATIONAL SYMPOSIUM ON HIGH-PERFORMANCE PARALLEL AND DISTRIBUTED COMPUTING | 2019年
基金
美国国家科学基金会;
关键词
D O I
10.1145/3307681.3326606
中图分类号
TP301 [理论、方法];
学科分类号
081202 ;
摘要
Attracted by the enormous potentials of Graphics Processing Units (GPUs), an array of efforts has surged to deploy Breadth-First Search (BFS) on GPUs, which, however, often exploits the static mechanisms to address the challenges that are dynamic in nature. Such a mismatch prevents us from achieving the optimal performance for offloading graph traversal on GPUs. To this end, we propose XBFS that leverages the runtime optimizations atop GPUs to cope with the nondeterministic characteristics of BFS with the following three techniques: First, XBFS adaptively exploits four either new or optimized frontier queue generation designs to accommodate various BFS levels that present dissimilar features. Second, inspired by the observation that the workload associated with each vertex is not proportional to its degree in bottom-up, we design three new strategies to better balance the workload. Third, XBFS introduces the first truly asynchronous bottom-up traversal which allows BFS to visit vertices for multiple levels at a single iteration with both theoretical soundness and practical benefits. Taken together, XBFS is, on average, 3.5x, 4.9x, 11.2x and 6.1x faster than the state-of-the-art Enterprise, Tigr, Gunrock on a Quadro P6000 GPU and Ligra on a 24-core Intel Xeon Platinum 8175M CPU. Note, the CPU used for Ligra is more expensive than the GPU for XBFS.
引用
收藏
页码:121 / 131
页数:11
相关论文
共 50 条
  • [1] 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
  • [2] Exploring FPGA Optimizations in OpenCL for Breadth-First Search on Sparse Graph Datasets
    Gondhalekar, Atharva
    Feng, Wu-Chun
    2020 30TH INTERNATIONAL CONFERENCE ON FIELD-PROGRAMMABLE LOGIC AND APPLICATIONS (FPL), 2020, : 133 - 137
  • [3] Load-Balanced Breadth-First Search on GPUs
    Zhu, Zhe
    Li, Jianjun
    Li, Guohui
    WEB-AGE INFORMATION MANAGEMENT, WAIM 2014, 2014, 8485 : 435 - 447
  • [4] Breadth-first search
    Swaine, M
    DR DOBBS JOURNAL, 2000, 25 (06): : 100 - +
  • [5] Breadth-first heuristic search
    Zhou, R
    Hansen, EA
    ARTIFICIAL INTELLIGENCE, 2006, 170 (4-5) : 385 - 408
  • [6] Enterprise: Breadth-First Graph Traversal on GPUs
    Liu, Hang
    Huang, H. Howie
    PROCEEDINGS OF SC15: THE INTERNATIONAL CONFERENCE FOR HIGH PERFORMANCE COMPUTING, NETWORKING, STORAGE AND ANALYSIS, 2015,
  • [7] SURF: Direction-Optimizing Breadth-First Search Using Workload State on GPUs
    Yoon, Daegun
    Oh, Sangyoon
    SENSORS, 2022, 22 (13)
  • [8] Efficient Breadth-First Reduct Search
    Boonjing, Veera
    Chanvarasuth, Pisit
    MATHEMATICS, 2020, 8 (05)
  • [9] A compressed breadth-first search for satisfiability
    Motter, DRB
    Markov, IL
    ALGORITHM ENGINEERING AND EXPERIMENTS, 2002, 2409 : 29 - 42
  • [10] WAVE: designing a heuristics-based three-way breadth-first search on GPUs
    Yoon, Daegun
    Jeong, Minjoong
    Oh, Sangyoon
    JOURNAL OF SUPERCOMPUTING, 2023, 79 (06): : 6889 - 6917