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:C146
DTSTART;TZID=America/Chicago:20181113T133000
DTEND;TZID=America/Chicago:20181113T140000
UID:submissions.supercomputing.org_SC18_sess178_pap171@linklings.com
SUMMARY:Large-Scale Hierarchical K-Means for Heterogeneous Many-Core Super
 computers
DESCRIPTION:Paper\nAlgorithms, Architectures, Data Analytics, Deep Learnin
 g, Networks, Scientific Computing, Visualization, Tech Program Reg Pass\n\
 nLarge-Scale Hierarchical K-Means for Heterogeneous Many-Core Supercompute
 rs\n\nLi, Yu, Zhao, Fu, Wang...\n\nThis paper presents a novel design and 
 implementation of k-means clustering algorithm targeting the Sunway TaihuL
 ight supercomputer. We introduce a multi-level parallel partition approach
  that not only partitions by dataflow and centroid, but also by dimension.
  Our multi-level (nkd) approach unlocks the potential of the hierarchical 
 parallelism in the SW26010 heterogeneous many-core processor and the syste
 m architecture of the supercomputer. \n\nOur design is able to process lar
 ge-scale clustering problems with up to 196,608 dimensions and over 160,00
 0 targeting centroids, while maintaining high performance and high scalabi
 lity, significantly improving the capability of k-means over previous appr
 oaches. The evaluation shows our implementation achieves performance of le
 ss than 18 seconds per iteration for a large-scale clustering case with 19
 6,608 data dimensions and 2,000 centroids by applying 4,096 nodes (1,064,4
 96 cores) in parallel, making k-means a more feasible solution for complex
  scenarios.
URL:https://sc18.supercomputing.org/presentation/?id=pap171&sess=sess178
END:VEVENT
END:VCALENDAR