lect22-parallel-io

Lect22-parallel-io - Notes Group Project presentations next Tuesday/Thursday email sent with(approximate schedule final report due Wed Dec 14 CMSC

Info iconThis preview shows pages 1–2. Sign up to view the full content.

View Full Document Right Arrow Icon
CMSC 714 Lecture 22 Parallel I/O Alan Sussman 2 Notes ! Group Project presentations next Tuesday/Thursday – email sent with (approximate) schedule – final report due Wed., Dec. 14 ! Course evaluation web site open – http://www.CourseEvalUM.umd.edu 3 IBM GPFS ! Designed to support high throughput parallel applications, including multimedia – well suited for scientific computations – still used in many of Top 500 supercomputers ! Main idea is to use parallel I/O to increase performance and scale to large configurations – increase bandwidth by spreading reads and writes (even to a single file) across multiple disks, especially for sequential access – avoid the “one file per parallel process” model, or sending all I/O through one node – use internal high performance switch, plus separate I/O nodes, for I/O from parallel processes running on nodes – files can be both striped across multiple I/O nodes, and across multiple disks in each I/O node 4 IBM GPFS !
Background image of page 1

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Image of page 2
This is the end of the preview. Sign up to access the rest of the document.

This note was uploaded on 01/12/2012 for the course CMSC 714 taught by Professor Staff during the Fall '07 term at Maryland.

Page1 / 2

Lect22-parallel-io - Notes Group Project presentations next Tuesday/Thursday email sent with(approximate schedule final report due Wed Dec 14 CMSC

This preview shows document pages 1 - 2. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online