BEGIN:VCALENDAR
VERSION:2.0
X-WR-TIMEZONE:America/Chicago
PRODID:-//Apple Inc.//iCal 3.0//EN
CALSCALE:GREGORIAN
X-WR-CALNAME:Load Balance: Correlating Application-Independent Measurements with Application-Semantic Computational Models
METHOD:PUBLISH
BEGIN:VTIMEZONE
TZID:America/Chicago
BEGIN:DAYLIGHT
TZOFFSETFROM:-0600
TZOFFSETTO:-0500
DTSTART:20070311T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU
TZNAME:CDT
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0500
TZOFFSETTO:-0600
DTSTART:20071104T020000
RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU
TZNAME:CST
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
SEQUENCE:2
DTSTART;TZID=America/Chicago:20101116T171500
DESCRIPTION:ABSTRACT: Load balance analysis and optimization is critical to performance as we continue to march towards exascale. Current tools\, however\, are limited since they do not take the applications’ semantics into account\, and hence cannot attribute measurements to phases of the parallel computation. For this purpose\, many applications include custom instrumentation\, but manual instrumentation can omit important aspects of application behavior. We are developing a new generation of load balance tools that correlate application-independent measurements with application-semantic computational models. In this poster\, we present four simple performance models and we show how these models can be correlated with measurements from Libra\, our scalable load-balance measurement tool. We show that our approach can be used both to guide application-level load balance algorithms\, as well as to validate and extend application-semantic models for more accurate measurements.
UID:post188@sc10.supercomputing.org
SUMMARY:Load Balance: Correlating Application-Independent Measurements with Application-Semantic Computational Models
DTEND;TZID=America/Chicago:20101116T190000
LOCATION:Main Lobby
END:VEVENT
END:VCALENDAR