BEGIN:VCALENDAR
PRODID:-//Microsoft Corporation//Outlook MIMEDIR//EN
VERSION:1.0
BEGIN:VEVENT
DTSTART:20141118T010000Z
DTEND:20141118T030000Z
LOCATION:Show Floor - 233 Exhibits
DESCRIPTION;ENCODING=QUOTED-PRINTABLE:ABSTRACT: Exascale performance requires a level of energy efficiency only achievable with specialized hardware. Hence, the trend is clearly going towards heterogeneous architectures. However, in order to fully exploit the advantages of such heterogeneous hardware, there is a need for appropriate programming models that do not increase the complexity of the applications.=0AIn the DEEP project, we have extended the OmpSs programming model to offload MPI kernels dynamically by utilizing MPI's spawning feature, here namely provided by ParaStation MPI as the underlying infrastructure.=0AThat way, offloading to specialized hardware gets dramatically simplified while hiding the low-level and error prone MPI spawn calls.=0AThese facts together with its portability make this concept definitely interesting for a broader audience of the HPC community.
SUMMARY:DEEP Collective Offload
PRIORITY:3
END:VEVENT
END:VCALENDAR