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:20260422T000714Z
LOCATION:702
DTSTART;TZID=America/Denver:20231113T153000
DTEND;TZID=America/Denver:20231113T155400
UID:submissions.supercomputing.org_SC23_sess449_ws_pawatm105@linklings.com
SUMMARY:symPACK:  A GPU-Capable Fan-Out Sparse Cholesky Solver
DESCRIPTION:Julian Bellavita (Lawrence Berkeley National Laboratory (LBNL)
 ); Mathias Jacquelin (Cerebras Systems); and Esmond Ng, Dan Bonachea, John
 ny Corbino, and Paul H. Hargrove (Lawrence Berkeley National Laboratory (L
 BNL))\n\nSparse symmetric positive definite systems of equations are ubiqu
 itous in scientific workloads and applications. Parallel sparse Cholesky f
 actorization is the method of choice for solving such linear systems. Ther
 efore, the development of parallel sparse Cholesky codes that can efficien
 tly run on today’s large-scale heterogeneous distributed-memory platforms 
 is of vital importance. Modern supercomputers offer nodes that contain a m
 ix of CPUs and GPUs. To fully utilize the computing power of these nodes, 
 scientific codes must be adapted to offload expensive computations to GPUs
 .\n\nWe present symPACK, a GPU-capable parallel sparse Cholesky solver tha
 t uses one-sided communication primitives and remote procedure calls provi
 ded by the UPC++ library. We also utilize the UPC++ "memory kinds" feature
  to enable efficient communication of GPU-resident data. We show that on a
  number of large problems, symPACK outperforms comparable state-of-the-art
  GPU-capable Cholesky factorization codes by up to 14x on the NERSC Perlmu
 tter supercomputer.\n\nTag: Accelerators, Artificial Intelligence/Machine 
 Learning, Algorithms, Applications, Architecture and Networks, Distributed
  Computing, Compilers, Data Analysis, Visualization, and Storage, Exascale
 , Heterogeneous Computing, Linear Algebra, Message Passing, Performance Op
 timization, Programming Frameworks and System Software, Quantum Computing,
  Runtime Systems, Software Engineering, Sustainability, Task Parallelism, 
 Tensors\n\nRegistration Category: Workshop Reg Pass\n\nSession Chairs: Eng
 in Kayraklioglu (Hewlett Packard Enterprise (HPE)), Daniele Lezzi (Barcelo
 na Supercomputing Center (BSC)), Bill Long (Retired), Karla Vanessa Morris
  Wright (Sandia National Laboratories), Irene Moulitsas (Cranfield Univers
 ity), and Elliott Slaughter (SLAC National Accelerator Laboratory)\n\n
END:VEVENT
END:VCALENDAR