- Notifications
You must be signed in to change notification settings - Fork0
High-Performance Streaming Graph Analytics on GPUs
License
NicolaDes/hornetsnest
Folders and files
Name | Name | Last commit message | Last commit date | |
---|---|---|---|---|
Repository files navigation
The repository provides the graph algorithms implemented on top Hornetdata structure.
For additional information concerning the data structure and its APIs pleaserefer toHornet repository.
The document is organized as follows:
- Requirements
- Quick start
- Hornet Algorithms
- Performance
- Hornet Algorithms Lines of Code
- Code Documentation
- Reporting bugs and contributing
- Publications
- Hornet Developers
- Acknowledgements
- License
- Nvidia Modern GPU(compute capability ≥ 3.0): Kerpler, Maxwell, Pascal, Volta architectures.
- CUDA toolkit 9 or greater.
- GCC orClang host compiler with support for C++14.Note: the compiler must be compatible with the related CUDA toolkit version.For more information seeCUDA Installation Guide.
- CMake v3.6 or greater.
- 64-bit Operating System (Ubuntu 16.04 or above suggested).
The following basic steps are required to build and execute the Hornetalgorithms:
git clone --recursive https://github.com/hornet-gt/hornetsnestcd hornetsnest/buildcmake ..make p
By default, the CUDA compilernvcc
usesgcc/g++
found in the currentexecution search path as host compiler(cc --version
to get the default compiler on the actual system).To force a different host compiler for compiling C++ files (*.cpp
)you need to set the following environment variables:
CC=<path_to_host_C_compiler>CXX=<path_to_host_C++_compiler>
To force a different host compiler for compiling host side code (*.cu
)substitutecmake ..
with
cmake -DCUDAHC=<path_to_host_C++_compiler> ..
Note: host.cpp
compiler and host side.cu
compiler may be different.The host side compiler must be compatible with the current CUDA Toolkitversion installed on the system(seeCUDA Installation Guide).
Algorithm | Static | Dynamic |
---|---|---|
(BFS) Breadth-first Search | yes | on-going |
(SSSP) Single-Source Shortest Path | yes | on-going |
(CC) Connected Components | yes | on-going |
(SCC) Strongly Connected Components | to-do | to-do |
(MST) Minimum Spanning Tree | to-do | to-do |
(BC) Betweeness Centrality | on-going | on-going |
(PG) Page Rank | yes | on-going |
(TC) Triangle Counting | yes | yes |
(KC) Katz Centrality | yes | yes |
(MIS) Maximal Independent Set | on-going | to-do |
(MF) Maximum Flow | to-do | to-do |
(CC) Clustering Coeffient | to-do | to-do |
(ST) St-Connectivity | to-do | to-do |
(TC) Transitive Closure | to-do | to-do |
Community Detection | to-do | to-do |
Temporal Motif Finding | on-going | to-do |
Sparse Vector-Matrix Multiplication | yes | to-do |
Jaccard indices | to-do | to-do |
Energy/Parity Game | on-going | to-do |
Algorithm | CPU1 | GPU2 | Speedup |
---|---|---|---|
(BFS) Breadth-first Search | |||
(SSSP) Single-Source Shortest Path | |||
(CC) Connected Components | |||
(MST) Minimum Spanning Tree | |||
(BC) Betweenness Centrality | |||
(PG) Page Rank | |||
(TC) Triangle Counting | |||
(KC) Katz Centrality |
1 Intel ...
2 NVidia Tesla P100 ..
Algorithm | Static | Dynamic | Speedup |
---|---|---|---|
(BFS) Breadth-first Search | |||
(SSSP) Single-Source Shortest Path | |||
(CC) Connected Components | |||
(MST) Minimum Spanning Tree | |||
(BC) Betweenness Centrality | |||
(PG) Page Rank | |||
(TC) Triangle Counting | |||
(KC) Katz Centrality |
Algorithm | Static (A) | Static (B) | Dynamic (A) |
---|---|---|---|
(BFS) Breadth-first Search | 4 | 6 | |
(SSSP) Single-Source Shortest Path | |||
(CC) Connected Components | |||
(MST) Minimum Spanning Tree | |||
(BC) Betweenness Centrality | |||
(PG) Page Rank | |||
(TC) Triangle Counting | |||
(KC) Katz Centrality |
(A) lines of code required for the algorithm
(B) lines of code required for the operators
The code documentation is located in thedocs
directory of Hornet datastructure directory (doxygen html format).The documentation is also accessible onlinehere.
If you find any bugs please report them by using the repository (githubissues panel).We are also ready to engage in improving and extending the framework if you request new features.
- Oded Green, David A. Bader,"cuSTINGER: Supporting dynamic graph algorithmsfor GPUs",IEEE High Performance Extreme Computing Conference (HPEC), 13-15 September,2016, Waltham, MA, USA, pp. 1-6.link
- Oded Green, James Fox, Euna Kim, Federico Busato, Nicola Bombieri,Kartik Lakhotia, Shijie Zhou, Shreyas Singapura, Hanqing Zeng,Rajgopal Kannan, Viktor Prasanna, David A. Bader,"Quickly Finding a Truss in a Haystack",IEEE/Amazon/DARPA Graph Challenge, *Innovation Awards*.
- Devavret Makkar, David A. Bader, Oded Green,Exact and Parallel Triangle Counting in Streaming Graphs,IEEE Conference on High Performance Computing, Data, and Analytics (HiPC),18-21 December 2017, Jaipur, India, pp. 1-10.
Federico Busato
, Ph.D. Student, University of Verona (Italy)Oded Green
, Researcher, Georgia Institute of Technology
Federico Busato
, Ph.D. Student, University of Verona (Italy)Oded Green
, Researcher, Georgia Institute of TechnologyJames Fox
, Ph.D. Student, Georgia Institute of Technology :Maximal Independent Set,Temporal Motif FindingDevavret Makkar
, Ph.D. Student, Georgia Institute of Technology :Triangle CountingElisabetta Bergamini
, Ph.D. Student, Karlsruhe Institute of Technology (Germany) :Katz CentralityEuna Kim
, Ph.D. Student, Georgia Institute of Technology :Dynamic PageRank- ...
- Grant...
BSD 3-Clause License
Copyright (c) 2017, HornetAll rights reserved.
Redistribution and use in source and binary forms, with or withoutmodification, are permitted provided that the following conditions are met:
Redistributions of source code must retain the above copyright notice, thislist of conditions and the following disclaimer.
Redistributions in binary form must reproduce the above copyright notice,this list of conditions and the following disclaimer in the documentationand/or other materials provided with the distribution.
Neither the name of the copyright holder nor the names of itscontributors may be used to endorse or promote products derived fromthis software without specific prior written permission.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THEIMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE AREDISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLEFOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIALDAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS ORSERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVERCAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USEOF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.