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:20260522T150129Z
LOCATION:C141/143/149
DTSTART;TZID=America/Chicago:20181113T143000
DTEND;TZID=America/Chicago:20181113T150000
UID:submissions.supercomputing.org_SC18_sess209_pap421@linklings.com
SUMMARY:HPL and DGEMM Performance Variability on the Xeon Platinum 8160 Pr
 ocessor
DESCRIPTION:John D. McCalpin (University of Texas, Texas Advanced Computin
 g Center)\n\nDuring initial testing of a large cluster equipped with Xeon 
 Platinum 8160 processors, we observed infrequent, but significant, perform
 ance drops in HPL benchmark results. The variability was seen in both sing
 le node and multi-node runs, with approximately 0.4% of results more than 
 10% slower than the median. We were able to reproduce this behavior with a
  single-socket (24-core) DGEMM benchmark. Performance counter analysis of 
 several thousand DGEMM runs showed that increased DRAM read traffic is the
  primary driver of increased execution time. Increased DRAM traffic in thi
 s benchmark is primarily generated by dramatically elevated snoop filter e
 victions, which arise due to the interaction of high-order (physical) addr
 ess bits with the hash used to map addresses across the 24 coherence agent
 s on the processor. These conflicts (and the associated performance variab
 ility) were effectively eliminated (for both DGEMM and HPL) by using 1 GiB
  large pages.\n\nTag: OpenMP, Performance, Power, Tools\n\nRegistration Ca
 tegory: Tech Program Reg Pass\n\nSession Chair: Felix Wolf (Technical Univ
 ersity of Darmstadt)\n\n
END:VEVENT
END:VCALENDAR