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:20260422T000713Z
LOCATION:401-402
DTSTART;TZID=America/Denver:20231115T153000
DTEND;TZID=America/Denver:20231115T160000
UID:submissions.supercomputing.org_SC23_sess174_pap507@linklings.com
SUMMARY:Automatic Generation of Distributed-Memory Mappings for Tensor Com
 putations
DESCRIPTION:Martin Kong, Raneem Abu Yosef, and Atanas Rountev (Ohio State 
 University) and P. Sadayappan (University of Utah)\n\nWhile considerable r
 esearch has been directed at automatic parallelization for shared-memory p
 latforms, little progress has been made in automatic parallelization schem
 es for distributed-memory systems. We introduce an innovative approach to 
 automatically produce distributed-memory parallel code for an important su
 b-class of affine tensor computations common to Coupled Cluster (CC) elect
 ronic structure methods, neuro-imaging applications, and deep learning mod
 els.\n\nWe propose a novel systematic approach to modeling the relations a
 nd trade-offs of mapping computations and data onto multi-dimensional grid
 s of homogeneous nodes.  Our formulation explores the space of computation
  and data distributions across processor grids. Tensor programs are modele
 d as a non-linear symbolic formulation accounting for the volume of data c
 ommunication and per-node capacity constraints induced under specific mapp
 ings. Solutions are found, iteratively, using the Z3 SMT solver, and used 
 to automatically generate efficient MPI code.  Our evaluation demonstrates
  the effectiveness of our approach over Distributed-Memory Pluto and the C
 yclops Tensor Framework.\n\nTag: Artificial Intelligence/Machine Learning,
  Compilers, Performance Measurement, Modeling, and Tools, Performance Opti
 mization, Programming Frameworks and System Software, Tensors\n\nRegistrat
 ion Category: Tech Program Reg Pass\n\nReproducibility Badges: Artifact Av
 ailable, Artifact Functional, Results Reproduced\n\nSession Chair: Kazem C
 heshmi (McMaster University, Ontario, Canada)\n\n
END:VEVENT
END:VCALENDAR