BEGIN:VCALENDAR
PRODID:-//Microsoft Corporation//Outlook MIMEDIR//EN
VERSION:1.0
BEGIN:VEVENT
DTSTART:20121114T173000Z
DTEND:20121114T180000Z
LOCATION:355-D
DESCRIPTION;ENCODING=QUOTED-PRINTABLE:ABSTRACT: I/O bottlenecks in HPC applications are becoming a more pressing=0Aproblem as compute capabilities continue to outpace I/O capabilities.=0AWhile double-precision simulation data often must be stored losslessly, the loss=0Aof some of the fractional component may introduce acceptably small errors to=0Amany types of scientific analyses.=0A=0AGiven this observation, we develop a precision level of detail (APLOD)=0Alibrary, which partitions double-precision datasets along user-defined=0Abyte boundaries. APLOD parameterizes the analysis accuracy-I/O=0Aperformance tradeoff, bounds maximum relative error, maintains I/O access=0Apatterns compared to full precision, and operates with=0Alow overhead. Using ADIOS as an I/O use-case, we show proportional reduction in=0Adisk access time to the degree of precision. Finally, we show the effects of=0Apartial precision analysis on accuracy for operations such as k-means and=0AFourier analysis, finding a strong applicability for the use of varying degrees=0Aof precision to reduce the cost of analyzing extreme-scale data.
SUMMARY:Byte-Precision Level of Detail Processing for Variable Precision Analytics
PRIORITY:3
END:VEVENT
END:VCALENDAR
BEGIN:VCALENDAR
PRODID:-//Microsoft Corporation//Outlook MIMEDIR//EN
VERSION:1.0
BEGIN:VEVENT
DTSTART:20121114T173000Z
DTEND:20121114T180000Z
LOCATION:355-D
DESCRIPTION;ENCODING=QUOTED-PRINTABLE:ABSTRACT: I/O bottlenecks in HPC applications are becoming a more pressing=0Aproblem as compute capabilities continue to outpace I/O capabilities.=0AWhile double-precision simulation data often must be stored losslessly, the loss=0Aof some of the fractional component may introduce acceptably small errors to=0Amany types of scientific analyses.=0A=0AGiven this observation, we develop a precision level of detail (APLOD)=0Alibrary, which partitions double-precision datasets along user-defined=0Abyte boundaries. APLOD parameterizes the analysis accuracy-I/O=0Aperformance tradeoff, bounds maximum relative error, maintains I/O access=0Apatterns compared to full precision, and operates with=0Alow overhead. Using ADIOS as an I/O use-case, we show proportional reduction in=0Adisk access time to the degree of precision. Finally, we show the effects of=0Apartial precision analysis on accuracy for operations such as k-means and=0AFourier analysis, finding a strong applicability for the use of varying degrees=0Aof precision to reduce the cost of analyzing extreme-scale data.
SUMMARY:Byte-Precision Level of Detail Processing for Variable Precision Analytics
PRIORITY:3
END:VEVENT
END:VCALENDAR