Random Walks on Huge Graphs at Cache Efficiency

被引:15
|
作者
Yang, Ke [1 ,2 ,3 ]
Ma, Xiaosong [2 ]
Thirumuruganathan, Saravanan [2 ]
Chen, Kang [1 ,3 ]
Wu, Yongwei [1 ,3 ]
机构
[1] Tsinghua Univ, Dept Comp Sci & Technol, Beijing Natl Res Ctr Informat Sci & Technol BNRis, Beijing, Peoples R China
[2] Hamad Bin Khalifa Univ, Qatar Comp Res Inst, Ar Rayyan, Qatar
[3] Beijing HaiZhi XingTu Technol Co Ltd, Beijing, Peoples R China
来源
PROCEEDINGS OF THE 28TH ACM SYMPOSIUM ON OPERATING SYSTEMS PRINCIPLES, SOSP 2021 | 2021年
关键词
graph computing; random walk; memory; cache; ANALYTICS; NETWORKS; SYSTEM;
D O I
10.1145/3477132.3483575
中图分类号
TP31 [计算机软件];
学科分类号
081202 ; 0835 ;
摘要
Data-intensive applications dominated by random accesses to large working sets fail to utilize the computing power of modern processors. Graph random walk, an indispensable workhorse for many important graph processing and learning applications, is one prominent case of such applications. Existing graph random walk systems are currently unable to match the GPU-side node embedding training speed. This work reveals that existing approaches fail to effectively utilize the modern CPU memory hierarchy, due to the widely held assumption that the inherent randomness in random walks and the skewed nature of graphs render most memory accesses random. We demonstrate that there is actually plenty of spatial and temporal locality to harvest, by careful partitioning, rearranging, and batching of operations. The resulting system, FlashMob, improves both cache and memory bandwidth utilization by making memory accesses more sequential and regular. We also found that a classical combinatorial optimization problem (and its exact pseudo-polynomial solution) can be applied to complex decision making, for accurate yet efficient data/task partitioning. Our comprehensive experiments over diverse graphs show that our system achieves an order of magnitude performance improvement over the fastest existing system. It processes a 58GB real graph at higher per-step speed than the existing system on a 600KB toy graph fitting in the L2 cache.
引用
收藏
页码:311 / 326
页数:16
相关论文
共 50 条
  • [41] Local Limit Theorems for Sequences of Simple Random Walks on Graphs
    Croydon, D. A.
    Hambly, B. M.
    POTENTIAL ANALYSIS, 2008, 29 (04) : 351 - 389
  • [42] Asymptotic capacity of the range of random walks on free products of graphs
    Gilch, Lorenz A.
    ELECTRONIC JOURNAL OF PROBABILITY, 2024, 29
  • [43] Local Limit Theorems for Sequences of Simple Random Walks on Graphs
    D. A. Croydon
    B. M. Hambly
    Potential Analysis, 2008, 29 : 351 - 389
  • [44] Hitting Times of Random Walks on Edge Corona Product Graphs
    Zhu, Mingzhe
    Xu, Wanyue
    Li, Wei
    Zhang, Zhongzhi
    Kan, Haibin
    COMPUTER JOURNAL, 2024, 67 (02) : 485 - 497
  • [45] Convergence of mixing times for sequences of random walks on finite graphs
    Croydon, D. A.
    Hambly, B. M.
    Kumagai, T.
    ELECTRONIC JOURNAL OF PROBABILITY, 2012, 17 : 1 - 32
  • [46] ON THE TRACE OF RANDOM WALKS ON RANDOM GRAPHS USING POISSON (Λ)-GALTON-WATSON TREE
    Vijayaseetha, N.
    Malliga, P.
    Kothandaraman, M.
    Malini, N.
    ADVANCES AND APPLICATIONS IN MATHEMATICAL SCIENCES, 2021, 21 (02): : 973 - 984
  • [47] Analytical results for the distribution of first return times of random walks on random regular graphs
    Tishby, Ido
    Biham, Ofer
    Katzav, Eytan
    JOURNAL OF PHYSICS A-MATHEMATICAL AND THEORETICAL, 2021, 54 (32)
  • [48] Analytical results for the distribution of first hitting times of random walks on random regular graphs
    Tishby, Ido
    Biham, Ofer
    Katzav, Eytan
    JOURNAL OF PHYSICS A-MATHEMATICAL AND THEORETICAL, 2021, 54 (14)
  • [49] RANDOM WALKS AND DIMENSIONS OF RANDOM TREES
    Konsowa, Mokhtar H.
    INFINITE DIMENSIONAL ANALYSIS QUANTUM PROBABILITY AND RELATED TOPICS, 2010, 13 (04) : 677 - 689
  • [50] One-mode interacting Fock spaces and random walks on graphs
    Obata, Nobuaki
    STOCHASTICS-AN INTERNATIONAL JOURNAL OF PROBABILITY AND STOCHASTIC PROCESSES, 2012, 84 (2-3) : 383 - 392