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:20260422T000612Z
LOCATION:401-402
DTSTART;TZID=America/Denver:20231117T083000
DTEND;TZID=America/Denver:20231117T120000
UID:submissions.supercomputing.org_SC23_sess460@linklings.com
SUMMARY:Future Is Sparse: Methods and Tools for Sparse Computations
DESCRIPTION:The Future of Machine Learning is Sparse\n\nAs machine learnin
 g models become prohibitively expensive to train and use, sparsity is incr
 easingly essential for neural networks such as large language models. If p
 erformed correctly, sparsity does not come at the cost of performance, and
  pruned deep neural networks have shown benefits in both thr...\n\n\nTal B
 en-Nun (Lawrence Livermore National Laboratory (LLNL))\n------------------
 ---\nMorning Break\n---------------------\nConsider Studying All-Pairs Sho
 rtest Paths\n\nThere is a well-known connection between the all-pairs shor
 test paths (APSP) problem and Gaussian elimination. In the context of this
  workshop, that makes APSP a good candidate to study because, for an input
  graph that is sparse but fully connected, the computation lies at the bou
 ndary between "spar...\n\n\nRichard Vuduc (Georgia Institute of Technology
 )\n---------------------\nDynamic Data Structures on the GPU\n\nIf I had t
 o sum up the purpose of data structures in three words, I'd say they "orga
 nize sparse data". And as we know, the future is sparse. Dynamic data stru
 ctures allow updates to data structures without having to rebuild them com
 pletely. In my talk, I'll discuss recent progress in designing and i...\n\
 n\nJohn Owens (University of California, Davis)\n---------------------\nTe
 nsor Cores for Matrix Multiplication Are on the Rise – Is There Any Hope f
 or Sparse Operations?\n\nIn response to the demands from the machine learn
 ing community, an increasing number of hardware architectures feature tens
 or cores for high performance dense matrix multiplication in low precision
 . This is increasing the peak performance and power draw - and the expecta
 tion that applications achiev...\n\n\nHartwig Anzt\n---------------------\
 nWelcome and Introduction by SparCity\n\nDidem Unat\n---------------------
 \nFuture Is Sparse Panel\n\nThis panel will assemble workshop speakers to 
 explore the challenges and opportunities surrounding sparse computation in
  the domains of HPC and AI. We anticipate that both the moderator and the 
 audience will have excellent and intriguing questions to present to the pa
 nelists.\n\n\nXing Cai (Simula); Hartwig Anzt; Tal Ben-Nun (Lawrence Liver
 more National Laboratory (LLNL)); John Owens (University of California, Da
 vis); and Richard Vuduc (Georgia Institute of Technology)\n\nTag: Graph Al
 gorithms and Frameworks, Linear Algebra, Programming Frameworks and System
  Software, State of the Practice\n\nRegistration Category: Workshop Reg Pa
 ss\n\nSession Chairs: Xing Cai (University of Oslo, Norway; Simula Researc
 h Laboratory/University of Oslo); Aleksandar Ilic (INESC-ID, IST, Universi
 ty of Lisbon; University of Lisbon, Portugal); Kamer Kaya (Sabancı Univers
 ity, Turkey); and Didem Unat (Koç University, Turkey)
END:VEVENT
END:VCALENDAR