BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:America/Denver
X-LIC-LOCATION:America/Denver
BEGIN:DAYLIGHT
TZOFFSETFROM:-0700
TZOFFSETTO:-0600
TZNAME:MDT
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0600
TZOFFSETTO:-0700
TZNAME:MST
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20240116T191702Z
LOCATION:401-402
DTSTART;TZID=America/Denver:20231116T113000
DTEND;TZID=America/Denver:20231116T120000
UID:submissions.supercomputing.org_SC23_sess158_pap376@linklings.com
SUMMARY:ReFloat: Low-Cost Floating-Point Processing in ReRAM for Accelerat
ing Iterative Linear Solvers
DESCRIPTION:Paper\n\nLinghao Song (University of California Los Angeles (U
CLA)), Fan Chen (Indiana University), and Hai Li and Yiran Chen (Duke Univ
ersity)\n\nResistive random access memory (ReRAM) is a promising technolog
y that can perform low-cost and in-situ matrix-vector multiplication (MVM)
in analog domain. Scientific computing requires high-precision floating-p
oint (FP) processing. However, performing floating-point computation in Re
RAM is challenging because of high hardware cost and execution time due to
the large FP value range. In this work we present ReFloat, a data format
and an accelerator architecture, for low-cost and high-performance floatin
g-point processing in ReRAM for iterative linear solvers. ReFloat matches
the ReRAM crossbar hardware and represents a block of FP values with reduc
ed bits and an optimized exponent base for a high range of dynamic represe
ntation. Thus, ReFloat achieves less ReRAM crossbar consumption and fewer
processing cycles and overcomes the noncovergence issue in a prior work. T
he evaluation on the SuiteSparse matrices shows that ReFloat achieves 5.02
x to 84.28x improvement in terms of solver time compared to a state-of-the
-art ReRAM based accelerator.\n\nTag: Algorithms, Linear Algebra, Post-Moo
re Computing\n\nRegistration Category: Tech Program Reg Pass\n\nReproducib
ility Badges: Artifact Available, Artifact Functional, Results Reproduced\
n\nSession Chair: Julien Langou (University of Colorado, Denver; Departmen
t of Mathematical and Statistical Sciences)
END:VEVENT
END:VCALENDAR