BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:America/Chicago
X-LIC-LOCATION:America/Chicago
BEGIN:DAYLIGHT
TZOFFSETFROM:-0600
TZOFFSETTO:-0500
TZNAME:CDT
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0500
TZOFFSETTO:-0600
TZNAME:CST
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20211207T055410Z
LOCATION:229
DTSTART;TZID=America/Chicago:20211119T083000
DTEND;TZID=America/Chicago:20211119T120000
UID:submissions.supercomputing.org_SC21_sess504@linklings.com
SUMMARY:RSDHA: Redefining Scalability for Diversely Heterogeneous Architec
 tures
DESCRIPTION:Workshop\n\nInvited Talk:  Mary Hall (University of Utah)\n\nH
 all\n\n---------------------\nRSDHA:  Redefining Scalability for Diversely
  Heterogeneous Architectures\n\nBelviranli, Lee, Song\n\nScalable computin
 g has governed another dimension. Contrary to the traditional HPC definiti
 on of total number of processors in a system, the newly rising dimension s
 cales with the number of different type of processing units sharing the co
 mputational load. HPC systems are expected to employ a more d...\n\n------
 ---------------\nELIXR: Eliminating Computation Redundancy in CNN-Based Vi
 deo Processing\n\nSchmerge, Mawhirter, Holmes, McClurg, Wu\n\nVideo proces
 sing frequently relies on applying convolutional neural networks (CNNs) fo
 r various tasks, including object tracking, real-time action classificatio
 n, and image recognition. Due to complicated network design, processing ev
 en a single frame requires many operations, leading to low through...\n\n-
 --------------------\nPlatform Agnostic Streaming Data Application Perform
 ance Models\n\nFaber, Plano, Kodali, Xiao, Dwaraki...\n\nThe mapping of co
 mputational needs onto execution resources is, by and large, a manual task
 , and users are frequently guided simply by intuition and past experiences
 . We present a queueing theory based performance model for streaming data 
 applications that takes steps towards a better understanding ...\n\n------
 ---------------\nRSDHA:  Morning Break (10:10-10:30)\n\n\n\n--------------
 -------\nDistributed Training for High Resolution Images:  A Domain and Sp
 atial Decomposition Approach\n\nTsaris, Hinkle, Lunga, Dias\n\nIn this wor
 k we developed two Pytorch libraries using the PyTorch RPC interface for d
 istributed deep learning approaches on high resolution images. The spatial
  decomposition library allows for distributed training on very large image
 s, which otherwise won’t be possible on a single GPU. The domain pa...\n\n
 ---------------------\nComparing LLC-Memory Traffic between CPU and GPU Ar
 chitectures\n\nMonil, Lee, Vetter, Malony\n\nThe cache hierarchy in modern
  CPUs and GPUs is becoming complex. The introduction of additional complex
 ities by manufacturers makes understanding the handshake between the memor
 y access pattern and the cache hierarchy difficult. Moreover, the details 
 of different cache policies are not publicly avai...\n\n------------------
 ---\nExploration of Multi-Accelerator Neural Network Inference in Diversel
 y Heterogeneous Embedded Systems\n\nDagli, Belviranli\n\nNeural network in
 ference (NNI) is commonly used in mobile and autonomous systems for latenc
 y-sensitive critical operations such as obstacle detection and avoidance. 
 In addition to latency, energy consumption is also an important factor in 
 such workloads, since battery is a shared and limited resourc...\n\n------
 ---------------\nEnergy Efficient Task Graph Execution Using Compute Unit 
 Masking in GPUs\n\nChow, Ranganath, Lerias Jr., Carodan, Wong\n\nThe front
 iers of Supercomputers are pushed by novel discrete accelerators. Accelera
 tors such as GPUs are employed to enable faster execution of Machine Learn
 ing, Scientific and High-Performance Computing applications. However, it h
 as been harder to gain increased parallelism in traditional workloads...\n
 \n---------------------\nInvited Talk: Jeffrey Vetter (Oak Ridge National 
 Laboratory)\n\nVetter\n\n---------------------\nPanel Discussion\n\nSong, 
 Chamberlain, Malony, Moore, Schuman...\n\nThe panel discussion at RSDHA wi
 ll seek answers for two primary questions:\n\n-- How could the traditional
  HPC applications adopt the architectural, programming and runtime approac
 hes employed by the state-of-the-art diversely heterogeneous systems?\n\n-
 - How could the diversely heterogeneous architectu...\n\n\nTag: Architectu
 res, Extreme Scale Computing, Heterogeneous Systems\n\nRegistration Catego
 ry: Workshop Reg Pass
END:VEVENT
END:VCALENDAR