BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Revisiting the I/O Bottleneck - University of Cambridge Computer L
 aboratory
DTSTART:20080527T120000Z
DTEND:20080527T130000Z
UID:TALK12354@talks.cam.ac.uk
CONTACT:Henry Robinson
DESCRIPTION:As commodity computers become more powerful with ever increasi
 ng\namounts of RAM and now 4 CPU cores as standard (with further increases
 \nexpected in the near future)\, is it the I/O subsystem that is the\nbott
 leneck?  And if it is\, then given the abundance of RAM and CPU\npower ava
 ilable could we use them to help alleviate the problem?\n\nWhen writing Li
 nux programs today programmers are faced with a\nplethora of choices of ho
 w to do I/O\, and the "wrong" choice could\nmean their program takes twice
  as long to execute.  In this talk I\npresent results from experimenting w
 ith the Linux I/O APIs and show\nthat performance can differ substantially
  depending on which and how\nthe API is used.  Clearly educating the progr
 ammers and re-writing all\nthe existing code would be a solution\, but wit
 h the abundance of RAM\nand CPU power available could other techniques all
 ow the programmer to\nalways use the same API\, but with the kernel giving
  it the performance\nequal to if it had used the "fastest" API.
LOCATION:Computer Laboratory\, William Gates Building\, Room FW11
END:VEVENT
END:VCALENDAR