Software

Graph 500

Graph 500 establishes a new set of data-intensive benchmarks and performance metrics for large-scale computer systems and applications.  The benchmarks are needed to guide the design of hardware architectures and software systems intended to support emerging data-intensive applications.  The Graph 500 results are announced each year at SC and ISC.

Graph 500 is backed by a steering committee of over 50 international HPC experts from academia, industry, and national laboratories, including Professors Lumsdaine and Sterling from CREST.  In addition, Professor Lumsdaine serves on the Graph 500 executive committee.

CREST has been heavily involved in the Graph 500 benchmark since its beginning in June 2010.  Jeremiah Willcock from CREST developed the graph generator and the MPI-based Graph 500 reference implementation.  Dr. Willcock also developed optimized implementations for the IBM Blue Gene/P system that placed first on the first two Graph 500 lists, and fifth on the third list.  CREST developed the Graph 500 Web site and has been maintaining it since the initiation of the benchmark.

Tuned MPI Graph 500 Implementation with 2-D Data Distribution

Graph 500 uses a two-dimensional data distribution and several other techniques for scalable performance on large systems and graphs. "Big Data" and other unstructured, irregular applications have become increasingly important in the past few years, and high-performance computing is being applied to these problems. However, standard benchmarks used for supercomputers, particularly LINPACK, do not measure performance on these new types of applications. The Graph 500 benchmark is a community effort to produce a benchmark representative of system performance on graph algorithms and other irregular applications. This software represents the tuned implementation of Graph 500 used for Argonne National Laboratory's Intrepid and Indiana University's Big Red 2 systems; it should be portable to any system with MPI, however. -Jeremiah Willcock


Click on the icons to see a description below.


______________________________________________________

The distribution of open source software is one way that CREST contributes to the wealth of scientific research at Indiana University and throughout the country. The production of open source software significantly enhances the value of scholarly work due to reproducibility andability of scientific research. The open source software created by CREST and its affiliates boosts IU’s competitive footing for federal grant funding, which in turn, brings federal research monies and new high-quality jobs into the state. Also used in the private sector, these open source software products aid in the development of the Indiana economy.

----------