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:20181221T160730Z
LOCATION:C141/143/149
DTSTART;TZID=America/Chicago:20181114T110000
DTEND;TZID=America/Chicago:20181114T113000
UID:submissions.supercomputing.org_SC18_sess205_pap194@linklings.com
SUMMARY:DRAGON: Breaking GPU Memory Capacity Limits with Direct NVM Access
DESCRIPTION:Paper\nGPUs, Memory, NVRAM, Performance, System Software, Tool
 s, Tech Program Reg Pass\n\nDRAGON: Breaking GPU Memory Capacity Limits wi
 th Direct NVM Access\n\nMarkthub, Belviranli, Lee, Vetter, Matsuoka\n\nHet
 erogeneous computing with accelerators is growing in importance in high pe
 rformance computing (HPC). Recently, application datasets have expanded be
 yond the memory capacity of these accelerators, and often beyond the capac
 ity of their hosts. Meanwhile, nonvolatile memory (NVM) storage has emerge
 d as a pervasive component in HPC systems because NVM provides massive amo
 unts of memory capacity at affordable cost. Currently, for accelerator app
 lications to use NVM, they must manually orchestrate data movement across 
 multiple memories and this approach only performs well for applications wi
 th simple access behaviors. To address this issue, we developed DRAGON, a 
 solution that enables all classes of GP-GPU applications to transparently 
 compute on terabyte datasets residing in NVM. DRAGON leverages the page-fa
 ulting mechanism on the recent NVIDIA GPUs by extending capabilities of CU
 DA Unified Memory (UM). Our experimental results show that DRAGON transpar
 ently expands memory capacity and obtain additional speedups via automated
  I/O and data transfer overlapping.
URL:https://sc18.supercomputing.org/presentation/?id=pap194&sess=sess205
END:VEVENT
END:VCALENDAR