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:20260522T150110Z
LOCATION:C2/3/4 Ballroom
DTSTART;TZID=America/Chicago:20181114T083000
DTEND;TZID=America/Chicago:20181114T170000
UID:submissions.supercomputing.org_SC18_sess342_drs105@linklings.com
SUMMARY:In-Memory Accelerator Architectures for Machine Learning and Bioin
formatics
DESCRIPTION:Roman Kaplan and Ran Ginosar (Israel Institute of Technology)\
n\nMost contemporary accelerators are von Neumann machines. With the incre
asing sizes of gathered and then processed data, memory bandwidth is the
main limiting of performance. One approach to mitigate the bandwidth const
raint is to bring the processing units closer to the data. This approach i
s known as near-data processing (NDP). However, NDP architecture
(e.g., Hybrid Memory Cube) are still inherently limited because they are b
ased on replicating the von Neumann architecture in memory.
My resea
rch proposes two new processing-in-storage architectures, where each bitce
ll can both store information and perform computation. The main building b
lock of the architectures are memristors, an emerging memory technolgy. The first architecture I propose, PRinS, was applied to accelerate mac
hine learning and large-scale DNA sequence alignment. Using Associative Pr
ocessing, PRinS achieves massive parallelism. The second, RASSA, accelerat
es DNA long read mapping, with approximate Hamming distance computation an
d quantifying mismatch of a pattern to voltage level.\n\nRegistration Cate
gory: Workshop Reg Pass, Tutorial Reg Pass, Tech Program Reg Pass, Exhibit
s Reg Pass, Exhibits - Exhibit Hall Only Reg Pass\n\n
END:VEVENT
END:VCALENDAR