BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:Europe/Stockholm
X-LIC-LOCATION:Europe/Stockholm
BEGIN:DAYLIGHT
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
TZNAME:CEST
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=-1SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
TZNAME:CET
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=-1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20220812T074334Z
LOCATION:Singapore Room
DTSTART;TZID=Europe/Stockholm:20220627T133000
DTEND;TZID=Europe/Stockholm:20220627T140000
UID:submissions.pasc-conference.org_PASC22_sess118_msa198@linklings.com
SUMMARY:Reducing IO Overhead with Compression and In-Situ Techniques
DESCRIPTION:Minisymposium\n\nReducing IO Overhead with Compression and In-
 Situ Techniques\n\nJu, Perez, Laure, Schlatter, Markidis\n\nCompared to th
 e peak computational performance, I/O performance grows relatively slowly 
 on High Performance Computing (HPC) systems, which also have limited stora
 ge capacity. Applications that aim to leverage the full power of Exascale 
 systems will produce large amounts of data that need to be stored through 
 the I/O subsystem efficiently, and current I/O performance and storage cap
 acity might lead to bottlenecks in performance and scientific discovery. D
 ata compression before storage combined with In-Situ techniques promises t
 o be an effective approach to tackle these problems. In this research, we 
 used the Computational Fluid Dynamics (CFD) solver Nek5000 as an example t
 o study in-situ lossless and lossy compression. We used the Adaptable Inpu
 t Output System (ADIOS2) to perform lossless compression and runtime confi
 guration of the in-situ execution. Moreover, we exploited known spectral p
 roperties of turbulent flows to discard data for lossy compression. We exp
 lored the effects of performing compression synchronously and asynchronous
 ly on a subset of the available resources. We observed that, through in-si
 tu compression, it is possible to reduce the I/O time and storage requirem
 ents, while preserving good accuracy in the data.\n\nDomain: Computer Scie
 nce and Applied Mathematics, Engineering
END:VEVENT
END:VCALENDAR