BEGIN:VCALENDAR
PRODID:-//Microsoft Corporation//Outlook MIMEDIR//EN
VERSION:1.0
BEGIN:VEVENT
DTSTART:20121114T223000Z
DTEND:20121115T000000Z
LOCATION:155-F
DESCRIPTION;ENCODING=QUOTED-PRINTABLE:ABSTRACT: Scientific simulations running at extreme=0Ascale on leadership class systems are generating unprecedented amount of=0Adata, which has to be analyzed and understood to enable scientific=0Adiscovery. However, the increasing gap between computation and disk IO=0Aspeeds makes traditional data analytics pipelines based on post-processing=0Acost prohibitive and often infeasible. This research explores hybrid=0Ain-situ/in-transit data staging and online execution of scientific data=0Aprocessing as coupled workflow. Specifically, we investigate the=0Aprogramming support for composing the coupled workflow from heterogeneous=0Acomputation components, and the runtime framework for distributed data=0Asharing and task execution. The framework employs data-centric task=0Aplacement to map workflow computations onto processor cores to reduce=0Anetwork data movement and increase intra-node data reuse. In addition, the=0Aframework implements the shared space programming abstraction with=0Aspecialized one-sided asynchronous data access operators and can be used=0Ato express coordination and data exchanges between the coupled components.
SUMMARY:Programming and Runtime Support for Enabling In-Situ/In-Transit Scientific Data Processing
PRIORITY:3
END:VEVENT
END:VCALENDAR