BEGIN:VCALENDAR
PRODID:-//Microsoft Corporation//Outlook MIMEDIR//EN
VERSION:1.0
BEGIN:VEVENT
DTSTART:20101117T173000Z
DTEND:20101117T180000Z
LOCATION:393
DESCRIPTION;ENCODING=QUOTED-PRINTABLE:ABSTRACT: Many important problems in computational sciences, social network=0A  analysis, security, and business analytics, are data-intensive and=0A  lend themselves to graph-theoretical analyses. In this paper we=0A  investigate the challenges involved in exploring very large graphs=0A  by designing a breadth-first search (BFS) algorithm for advanced=0A  multi-core processors that are likely to become the building blocks=0A  of future exascale systems. Our new methodology for large-scale=0A  graph analytics combines a high-level algorithmic design that=0A  captures the machine-independent aspects, to guarantee portability=0A  with performance to future processors, with an implementation that=0A  embeds processor-specific optimizations. We present an experimental=0A  study that uses state-of-the-art Intel Nehalem EP and EX processors=0A  and up to 64 threads in a single system. Our performance on several=0A  benchmark problems representative of the power-law graphs found in=0A  real-world problems reaches processing rates that are competitive=0A  with supercomputing results in the recent literature.
SUMMARY:Scalable Graph Exploration on Multicore Processors
PRIORITY:3
END:VEVENT
END:VCALENDAR