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:20210402T160558Z
LOCATION:Track 2
DTSTART;TZID=America/New_York:20201113T172500
DTEND;TZID=America/New_York:20201113T175500
UID:submissions.supercomputing.org_SC20_sess221_ws_p3hpc104@linklings.com
SUMMARY:Cross-Platform Performance Portability of DNN Models Using SYCL
DESCRIPTION:Workshop\n\nCross-Platform Performance Portability of DNN Mode
 ls Using SYCL\n\nGoli, Narasimhan, Reyes, Tracy, Soutar...\n\nThe incoming
  deployment of Exascale platforms with a myriad of different architectures
  and co-processors have prompted the need to provide a software ecosystem 
 based on open standards that can simplify maintaining HPC applications on 
 different hardware.  Applications written for a particular platform should
  be portable to a different one, ensuring performance is as close to the p
 eak as possible.  However, it is not expected that key performance routine
 s on relevant HPC applications will be performance portable as is, especia
 lly for common building blocks such as BLAS or DNN.  The oneAPI the initia
 tive aims to tackle this problem by combining a programming model, SYCL, w
 ith a set of interfaces for common building blocks that can be optimized f
 or different hardware vendors.  In particular, oneAPI includes the oneDNN 
 performance library, which contains building blocks for deep learning appl
 ications and frameworks.  \n\nBy using the SYCL programming model, it can 
 integrate easily with existing SYCL and C++ applications, sharing data and
  executing collaboratively on devices with the rest of the application.  I
 n this paper, we introduce a cuDNN backend for oneDNN, which allows runnin
 g oneAPI applications on NVIDIA hardware taking advantage of existing buil
 ding blocks from the CUDA ecosystem.  We implement relevant neural network
 s (ResNet-50 and VGG-16) on native CUDA and also a version of oneAPI with 
 a CUDA backend, and demonstrate that performance portability can be achiev
 ed by leveraging existing building blocks for the target hardware.\n\nRegi
 stration Category: Workshop Reg Pass
END:VEVENT
END:VCALENDAR