BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:America/New_York
X-LIC-LOCATION:America/New_York
BEGIN:DAYLIGHT
TZOFFSETFROM:-0500
TZOFFSETTO:-0400
TZNAME:EDT
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0400
TZOFFSETTO:-0500
TZNAME:EST
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20210402T160551Z
LOCATION:Track 8
DTSTART;TZID=America/New_York:20201111T161000
DTEND;TZID=America/New_York:20201111T163000
UID:submissions.supercomputing.org_SC20_sess198_ws_whpc109@linklings.com
SUMMARY:High-Performance Sparse Tensor Algebra Compiler
DESCRIPTION:Workshop\n\nHigh-Performance Sparse Tensor Algebra Compiler\n\
 nTian, Li, Ren, Kestor\n\nTensor algebra computation is an important compu
 tation kernel used in many applications in different areas. Tensors repres
 ented real-world data are usually large and sparse. Sparse tensors are usu
 ally stored in a compressed way.  Lot of efforts are focusing on improving
  the performance. There are many challenges in improving the performance. 
 First, there are many storage formats to store the sparse tensors and no o
 ne format is good in all cases. Users need to choose the proper format acc
 ording to the feature of the sparse tensors. Secondly, optimizing sparse c
 omputation is difficult. Sparse computation contains many indirect memory 
 accesses and write dependencies. Besides this, the computation kernels are
  also variant with different tensor expressions and different storage form
 ats. It is necessary to use different optimizations in different computati
 ons kernels. Thirdly, there are many hardware platforms at the back end no
 wadays. Different hardware platforms require different code optimizations 
 for high-performance. \n\nTo handle some of the challenges, we propose a c
 ompiler-based approach by building our sparse tensor compiler based on the
  multi-level Intermediate Representation (MLIR) framework. By building our
  sparse tensor compiler based on MLIR infrastructure, our compiler support
 s different hardware platforms.\n\nOur sparse tensor compiler supports sev
 eral formats, such as COO, CSF and so on. Based on the proposed internal t
 ensor storage, we propose an automatic code generation algorithm to genera
 te the computation kernel code. We apply various code optimizations to gua
 rantee the performance. We believe our approach is a promising way to supp
 ort high-performance in a more general and flexible way.\n\nTag: Education
 , Training and Outreach, Professional Development, Workforce Development\n
 \nRegistration Category: Workshop Reg Pass
END:VEVENT
END:VCALENDAR