BEGIN:VCALENDAR
PRODID:-//Microsoft Corporation//Outlook MIMEDIR//EN
VERSION:2.0
BEGIN:VEVENT
DTSTART:20151117T231500Z
DTEND:20151118T010000Z
LOCATION:Level 4 - Lobby
DESCRIPTION;ENCODING=QUOTED-PRINTABLE:ABSTRACT: In this work we present a library, LIBXSMM, that provides a high performance=0Aimplementation of small sparse and dense matrix multiplications on latest Intel architectures. Such operations are important=0Abuilding blocks in modern scientific applications and general math libraries are normally tuned for all dimensions=0Abeing large. LIBXSMM follows a matrix multiplication code generation approach specifically matching the applications' needs. By=0Aproviding several interfaces, the replacement of BLAS calls is simple and straightforward.=0AWe show that  depending on the application's=0Acharacteristics, LIBXSMM can either leverage the entire DRAM bandwidth or reaches close to the processor's =0Acomputational peak performance.=0AOur performance results of CP2K and SeisSol=0Atherefore demonstrate that using LIBXSMM as a highly-efficient computational=0Abackend, leads to speed-ups of greater than two compared to compiler=0Agenerated inlined code or calling highly-optimized vendor math libraries.
SUMMARY:LIBXSMM: A High Performance Library for Small Matrix Multiplications
PRIORITY:3
END:VEVENT
END:VCALENDAR