BEGIN:VCALENDAR
PRODID:-//Microsoft Corporation//Outlook MIMEDIR//EN
VERSION:1.0
BEGIN:VEVENT
DTSTART:20111116T011500Z
DTEND:20111116T030000Z
LOCATION:WSCC North Galleria 2nd/3rd Floors
DESCRIPTION;ENCODING=QUOTED-PRINTABLE:ABSTRACT: With the increasing complexity of compute nodes, it is helpful to exploit multiple levels of parallelism even within a single compute node.=0AWe present ShyLU, a "hybrid-hybrid" solver for general sparse linear systems=0Athat is hybrid in two ways: First, it combines direct and iterative methods. The iterative method=0Ais based on approximate Schur complements. Second, the solver uses two=0Alevels of parallelism via hybrid programming (MPI+threads).=0AOur solver is useful both in shared-memory environments and on large parallel computers with distributed memory (as a subdomain solver).=0AWe show the robustness of ShyLU against other algebraic preconditioners. ShyLU scales well up to 192 cores for a given problem size. We compare flat MPI performance=0Aof ShyLU against a hybrid implementation. We conclude that on present multicore nodes flat MPI is better. However, for future manycore machines (48 or more cores)=0Ahybrid/ hierarchical algorithms and implementations are important for sustained performance.
SUMMARY:A Hybrid-Hybrid Solver for Manycore Platforms
PRIORITY:3
END:VEVENT
END:VCALENDAR