Distributed Memory Graph Coloring Algorithms for Multiple GPUs

被引:3
|
作者
Bogle, Ian [1 ]
Boman, Erik G. [2 ]
Devine, Karen [2 ]
Rajamanickam, Sivasankaran [2 ]
Slota, George M. [1 ]
机构
[1] Rensselaer Polytech Inst, Troy, NY 12181 USA
[2] Sandia Natl Labs, POB 5800, Albuquerque, NM 87185 USA
来源
PROCEEDINGS OF IA3 2020: 2020 IEEE/ACM 10TH WORKSHOP ON IRREGULAR APPLICATIONS: ARCHITECTURES AND ALGORITHMS (IA3) | 2020年
基金
美国国家科学基金会;
关键词
graph coloring; distributed algorithms; GPU;
D O I
10.1109/IA351965.2020.00013
中图分类号
TP3 [计算技术、计算机技术];
学科分类号
0812 ;
摘要
Graph coloring is often used in parallelizing scientific computations that run in distributed and multi-GPU environments; it identifies sets of independent data that can be updated in parallel. Many algorithms exist for graph coloring on a single GPU or in distributed memory, but hybrid MPI+GPU algorithms have been unexplored until this work, to the best of our knowledge. We present several MPI+GPU coloring approaches that use implementations of the distributed coloring algorithms of Gebremedhin et al. and the shared-memory algorithms of Deveci et al. The on-node parallel coloring uses implementations in KokkosKernels, which provide parallelization for both multicore CPUs and GPUs. We further extend our approaches to solve for distance-2 coloring, giving the first known distributed and multi-GPU algorithm for this problem. In addition, we propose novel methods to reduce communication in distributed graph coloring. Our experiments show that our approaches operate efficiently on inputs too large to fit on a single GPU and scale up to graphs with 76.7 billion edges running on 128 GPUs.
引用
收藏
页码:54 / 62
页数:9
相关论文
共 50 条
  • [41] Location-Oblivious Distributed Unit Disk Graph Coloring
    Michel Barbeau
    Prosenjit Bose
    Paz Carmi
    Mathieu Couture
    Evangelos Kranakis
    Algorithmica, 2011, 60 : 236 - 249
  • [42] Location-Oblivious Distributed Unit Disk Graph Coloring
    Barbeau, Michel
    Bose, Prosenjit
    Carmi, Paz
    Couture, Mathieu
    Kranakis, Evangelos
    ALGORITHMICA, 2011, 60 (02) : 236 - 249
  • [43] Distributed-Memory k-mer Counting on GPUs
    Nisa, Israt
    Pandey, Prashant
    Ellis, Marquita
    Oliker, Leonid
    Buluc, Aydin
    Yelick, Katherine
    2021 IEEE 35TH INTERNATIONAL PARALLEL AND DISTRIBUTED PROCESSING SYMPOSIUM (IPDPS), 2021, : 527 - 536
  • [44] Graph coloring algorithms for multi-core and massively multithreaded architectures
    Catalyuerek, Uemit V.
    Feo, John
    Gebremedhin, Assefaw H.
    Halappanavar, Mahantesh
    Pothen, Alex
    PARALLEL COMPUTING, 2012, 38 (10-11) : 576 - 594
  • [45] Acyclic orientation graph coloring for software-managed memory allocation
    Li Wang
    JingLing Xue
    XueJun Yang
    Science China Information Sciences, 2014, 57 : 1 - 18
  • [46] Acyclic orientation graph coloring for software-managed memory allocation
    WANG Li
    XUE JingLing
    YANG XueJun
    ScienceChina(InformationSciences), 2014, 57 (09) : 40 - 57
  • [47] Acyclic orientation graph coloring for software-managed memory allocation
    Wang Li
    Xue JingLing
    Yang XueJun
    SCIENCE CHINA-INFORMATION SCIENCES, 2014, 57 (09) : 1 - 18
  • [48] Compiler-Directed Scratchpad Memory Management via Graph Coloring
    Li, Lian
    Feng, Hui
    Xue, Jingling
    ACM TRANSACTIONS ON ARCHITECTURE AND CODE OPTIMIZATION, 2009, 6 (03)
  • [49] Morph Algorithms on GPUs
    Nasre, Rupesh
    Burtscher, Martin
    Pingali, Keshav
    ACM SIGPLAN NOTICES, 2013, 48 (08) : 147 - 156
  • [50] The Graph Coloring Problem-Review of Algorithms & Neural Networks and a New Proposal
    Ansari, Mohd. Samar
    2013 INTERNATIONAL CONFERENCE ON MULTIMEDIA, SIGNAL PROCESSING AND COMMUNICATION TECHNOLOGIES (IMPACT), 2013, : 310 - 314