BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:America/Chicago
X-LIC-LOCATION:America/Chicago
BEGIN:DAYLIGHT
TZOFFSETFROM:-0600
TZOFFSETTO:-0500
TZNAME:CDT
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0500
TZOFFSETTO:-0600
TZNAME:CST
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20181221T160729Z
LOCATION:C141/143/149
DTSTART;TZID=America/Chicago:20181113T103000
DTEND;TZID=America/Chicago:20181113T110000
UID:submissions.supercomputing.org_SC18_sess203_pap262@linklings.com
SUMMARY:GPU Age-Aware Scheduling to Improve the Reliability of Leadership 
 Jobs on Titan
DESCRIPTION:Paper\nGPUs, Resiliency, State of the Practice, System Softwar
 e, Tech Program Reg Pass\n\nGPU Age-Aware Scheduling to Improve the Reliab
 ility of Leadership Jobs on Titan\n\nZimmer, Maxwell, McNally, Atchley, Va
 zhkudai\n\nThe increasing rate of failures on the Oak Ridge Leadership Com
 puting Facility's (OLCF) Titan supercomputer, resulted in the replacement 
 of 50% of its GPUs between 2015 and 2017. The largest jobs, also known as 
 "leadership jobs'', continued to experience increased application failures
 . These jobs contained significant amounts of low-failure rate and high-fa
 ilure rate GPUs. The impacts of these failures were felt more by leadershi
 p jobs due to longer wait times, runtimes, and higher charge rates. In thi
 s work, we have designed techniques to increase the use of low-failure GPU
 s in leadership jobs through targeted resource allocation. This employed t
 wo complementary techniques, updating both the system ordering and the all
 ocation mechanisms. In simulation, the application of these techniques res
 ulted in a 33% increase in low-failure GPU hours being assigned to leaders
 hip jobs. Our GPU Age-Aware Scheduling has been used in production on Tita
 n since July of 2017.
URL:https://sc18.supercomputing.org/presentation/?id=pap262&sess=sess203
END:VEVENT
END:VCALENDAR