BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:America/New_York
X-LIC-LOCATION:America/New_York
BEGIN:DAYLIGHT
TZOFFSETFROM:-0500
TZOFFSETTO:-0400
TZNAME:EDT
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0400
TZOFFSETTO:-0500
TZNAME:EST
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20260422T143139Z
LOCATION:B302
DTSTART;TZID=America/New_York:20241117T164500
DTEND;TZID=America/New_York:20241117T171000
UID:submissions.supercomputing.org_SC24_sess744_ws_memo103@linklings.com
SUMMARY:UpDown: Combining Scalable Address Translation with Locality Contr
 ol
DESCRIPTION:Yuqing Wang (University of Chicago); Swann Perarnau (Argonne N
 ational Laboratory (ANL)); and Andrew A. Chien (University of Chicago, Arg
 onne National Laboratory (ANL))\n\nGlobal shared memories of petabytes are
  increasingly useful for applications, but traditional page-based techniqu
 es do not scale (limit reach), and scalable techniques such as segments fa
 il to provide needed locality control. We propose a novel two-level transl
 ation scheme, UpDown, that provides compact, efficient access control to d
 istributed segments of arbitrary size and data layout control, solving pro
 blems of reach and data locality. UpDown's novel two-level structure separ
 ates access control from data layout, allowing applications to manage loca
 lity cheaply, without privileged operations. \n\nWe evaluate UpDown agains
 t page-based systems, using big data computations. Our results show that U
 pDown is scalable and provides effective global data locality management. 
 UpDown's translation states are an average of ∼620 billion times smaller t
 otal. UpDown's local translation state is ∼130 billion times smaller. Simu
 lations with synthetic traces show that the two-level translation scheme e
 nables fast, user-level management of data locality in a scalable parallel
  machine.\n\nTag: Data Movement and Memory, Emerging Technologies\n\nRegis
 tration Category: Workshop Reg Pass\n\nSession Chairs: Ron Brightwell (San
 dia National Laboratories), Maya Gokhale (Lawrence Livermore National Labo
 ratory (LLNL)), Kyle Hale (Oregon State University), and Ivy Peng (KTH Roy
 al Institute of Technology)\n\n
END:VEVENT
END:VCALENDAR