BEGIN:VCALENDAR
PRODID:-//Microsoft Corporation//Outlook MIMEDIR//EN
VERSION:1.0
BEGIN:VEVENT
DTSTART:20131121T223000Z
DTEND:20131122T000000Z
LOCATION:601/603
DESCRIPTION;ENCODING=QUOTED-PRINTABLE:ABSTRACT: We conduct performance studies of the High Performance ParalleX (HPX) thread scheduling policies using a task scheduling micro benchmark, the Unbalanced Tree Search (UTS) benchmark, and an adaptive mesh refinement (AMR) application, on a variety of platforms to determine metrics that will aid in improving scheduling policies. Our objective is to determine if an adaptive management of scheduling policies can be used dynamically for task management among hardware threads on a node to improve performance of scaling impaired parallel applications. Once the metrics have been determined, we will apply results to dynamically adjust factors of the thread scheduling policy for improved performance and efficiency. Future work includes using similar performance studies to assess the possibility of using adaptive techniques on distributed processes in the Active Global Address Space (AGAS) module of HPX.
SUMMARY:Performance Analysis Towards Adaptive Thread Scheduling Policy
PRIORITY:3
END:VEVENT
END:VCALENDAR