BEGIN:VCALENDAR
PRODID:-//Microsoft Corporation//Outlook MIMEDIR//EN
VERSION:1.0
BEGIN:VEVENT
DTSTART:20111115T223000Z
DTEND:20111115T224500Z
LOCATION:TCC LL1
DESCRIPTION;ENCODING=QUOTED-PRINTABLE:ABSTRACT: With single-core speeds no longer rising, dramatically increased parallelism is now the means of getting more performance from supercomputers. The current generation of algorithms run on these machines will have to adapt to this new landscape. In this talk, we focus on algebraic multigrid (AMG), a popular linear solver with many scientific and engineering applications. AMG has the attractive property of requiring work that is linear in the number of unknowns. However, it also has substantial communication requirements that impede its scalability on emerging architectures. We present an overview of the issues, and discuss algorithmic changes that are underway to address the scalability issues. A notable feature is the resurrection of ideas first proposed before the dramatic rise in single-core speeds seen in the late 1990s through early 2000s; what distinguishes our work is the use of performance modeling to guide our changes to ensure their success.
SUMMARY:Scaling Algebraic Multigrid Solvers to Future Supercomputers
PRIORITY:3
END:VEVENT
END:VCALENDAR