BEGIN:VCALENDAR
PRODID:-//Microsoft Corporation//Outlook MIMEDIR//EN
VERSION:1.0
BEGIN:VEVENT
DTSTART:20121114T183000Z
DTEND:20121114T190000Z
LOCATION:355-D
DESCRIPTION;ENCODING=QUOTED-PRINTABLE:ABSTRACT: Hierarchical, multi-resolution data representations enable interactive analysis and visualization of large-scale simulations. One promising application of these techniques is to store HPC simulation output in a hierarchical Z (HZ) ordering that translates data from a Cartesian coordinate scheme to a one dimensional=0Aarray ordered by locality at different resolution levels. However, when the dimensions of the simulation data are not an even power of two, parallel HZ-ordering produces sparse memory and network access patterns that inhibit I/O=0Aperformance. This work presents a new technique for parallel HZ-ordering of simulation datasets that restructures simulation data into large power of two blocks to facilitate efficient I/O aggregation. We perform both weak and strong scaling experiments using the S3D combustion application on both Cray-XE6 (65536 cores) and IBM BlueGene/P (131072 cores) platforms. We demonstrate that data can be written in hierarchical, multiresolution format with performance competitive to that of native data ordering methods.
SUMMARY:Efficient Data Restructuring and Aggregation for IO Acceleration in PIDX
PRIORITY:3
END:VEVENT
END:VCALENDAR
BEGIN:VCALENDAR
PRODID:-//Microsoft Corporation//Outlook MIMEDIR//EN
VERSION:1.0
BEGIN:VEVENT
DTSTART:20121114T183000Z
DTEND:20121114T190000Z
LOCATION:355-D
DESCRIPTION;ENCODING=QUOTED-PRINTABLE:ABSTRACT: Hierarchical, multi-resolution data representations enable interactive analysis and visualization of large-scale simulations. One promising application of these techniques is to store HPC simulation output in a hierarchical Z (HZ) ordering that translates data from a Cartesian coordinate scheme to a one dimensional=0Aarray ordered by locality at different resolution levels. However, when the dimensions of the simulation data are not an even power of two, parallel HZ-ordering produces sparse memory and network access patterns that inhibit I/O=0Aperformance. This work presents a new technique for parallel HZ-ordering of simulation datasets that restructures simulation data into large power of two blocks to facilitate efficient I/O aggregation. We perform both weak and strong scaling experiments using the S3D combustion application on both Cray-XE6 (65536 cores) and IBM BlueGene/P (131072 cores) platforms. We demonstrate that data can be written in hierarchical, multiresolution format with performance competitive to that of native data ordering methods.
SUMMARY:Efficient Data Restructuring and Aggregation for IO Acceleration in PIDX
PRIORITY:3
END:VEVENT
END:VCALENDAR