Parallel Graph Algorithms on the Xeon Phi Coprocessor

Dennis Felsing
Department of Informatics, Institute of Theoretical Computer Science, Parallel Computing Group, Karlsruhe Institute of Technology
Karlsruhe Institute of Technology, 2015

   title={Parallel Graph Algorithms on the Xeon Phi Coprocessor},

   author={Felsing, Dennis},



Download Download (PDF)   View View   Source Source   



Complex networks have received interest in a wide area of applications, ranging from road networks over hyperlink connections in the world wide web to interactions between people. Advanced algorithms are required for the generation as well as visualization of such graphs. In this work two graph algorithms, one for graph generation, the other for graph visualization, are studied exemplarily. We detail the work of adapting and porting the algorithms to the Intel Xeon Phi coprocessor architecture. Problems in porting real software projects and used libraries are encountered and solved. Memory allocations turned out to be a major problem for the graph generation algorithm. The limited memory of the Xeon Phi forced us to offload chunks of the data from the host system to the Xeon Phi, which impeded performance, eliminating any significant speedup. The data sets consisting of at most 365 000 edges for the graph visualization algorithm fit into the Xeon Phi’s memory easily, which simplified the porting process significantly. We achieve a speedup for sparse graphs over the host system containing two 8-core Intel Xeon (Sandy Bridge) processors. While the hot inner loop by itself can utilize the 512-bit vector instructions of the Xeon Phi, the benefit disappears when embedded in the more complicated full program.
VN:F [1.9.22_1171]
Rating: 0.0/5 (0 votes cast)

* * *

* * *

TwitterAPIExchange Object
    [oauth_access_token:TwitterAPIExchange:private] => 301967669-yDz6MrfyJFFsH1DVvrw5Xb9phx2d0DSOFuLehBGh
    [oauth_access_token_secret:TwitterAPIExchange:private] => o29ji3VLVmB6jASMqY8G7QZDCrdFmoTvCDNNUlb7s
    [consumer_key:TwitterAPIExchange:private] => TdQb63pho0ak9VevwMWpEgXAE
    [consumer_secret:TwitterAPIExchange:private] => Uq4rWz7nUnH1y6ab6uQ9xMk0KLcDrmckneEMdlq6G5E0jlQCFx
    [postfields:TwitterAPIExchange:private] => 
    [getfield:TwitterAPIExchange:private] => ?cursor=-1&screen_name=hgpu&skip_status=true&include_user_entities=false
    [oauth:protected] => Array
            [oauth_consumer_key] => TdQb63pho0ak9VevwMWpEgXAE
            [oauth_nonce] => 1477600394
            [oauth_signature_method] => HMAC-SHA1
            [oauth_token] => 301967669-yDz6MrfyJFFsH1DVvrw5Xb9phx2d0DSOFuLehBGh
            [oauth_timestamp] => 1477600394
            [oauth_version] => 1.0
            [cursor] => -1
            [screen_name] => hgpu
            [skip_status] => true
            [include_user_entities] => false
            [oauth_signature] => kGGtp1J/vgA21isQiABvgAk3QbQ=

    [url] => https://api.twitter.com/1.1/users/show.json
Follow us on Facebook
Follow us on Twitter

HGPU group

2036 peoples are following HGPU @twitter

HGPU group © 2010-2016 hgpu.org

All rights belong to the respective authors

Contact us: