When the latest version of the Graph 500 list was released Nov. 16 at the SC16 conference, there were two new entries in the top 10, both contributed by Khaled Ibrahim of Berkeley Lab’s Computational Research Division. “Ibrahim explains that such workloads, known as communication-bound applications are typically the most difficult to scale on HPC systems. But finding a way to scale up their performance can have a big payoff by reducing the computational “expense,” or amount of computing time needed to solve a problem.”
George Slota presented this talk at the Blue Waters Symposium. “In recent years, many graph processing frameworks have been introduced with the goal to simplify analysis of real-world graphs on commodity hardware. However, these popular frameworks lack scalability to modern massive-scale datasets. This work introduces a methodology for graph processing on distributed HPC systems that is simple to implement, generalizable to broad classes of graph algorithms, and scales to systems with hundreds of thousands of cores and graphs of billions of vertices and trillions of edges.”
The Graph500 executive committee is soliciting submissions to the 8th Graph500 and Green Graph500 lists, which will be unveiled in June at ISC’14 in Leipzig.