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:20260422T000618Z
LOCATION:505
DTSTART;TZID=America/Denver:20231112T090000
DTEND;TZID=America/Denver:20231112T123000
UID:submissions.supercomputing.org_SC23_sess421@linklings.com
SUMMARY:RSDHA: Redefining Scalability for Diversely Heterogeneous Architec
 tures
DESCRIPTION:Accelerator Integration in a Tile-Based SoC:  Lessons Learned 
 with a Hardware Floating Point Compression Engine\n\nHeterogeneous Intelle
 ctual Property (IP) hardware acceleration engines have emerged as a viable
  path forward to improving performance in the waning of Moore’s Law and De
 nnard scaling. In this study, we design, prototype, and evaluate the HPC-s
 pecialized ZHW floating point compression accelerat...\n\n\nXueyang Liu (G
 eorgia Institute of Technology); Patricia Gonzalez-Guerrero (Lawrence Berk
 eley National Laboratory (LBNL)); Ivy (Bo) Peng (KTH Royal Institute of Te
 chnology, Sweden); Ronald Minnich (Samsung Corporation); and Maya Gokhale 
 (Lawrence Livermore National Laboratory (LLNL))\n---------------------\nEv
 aluating Primitives in Deep Neural Network Libraries:  A Case Study with t
 he Softmax Functions\n\nA deep neural network library (DNNL) is an optimiz
 ed library of low-level computational primitives for deep neural networks.
  In this study, we choose the softmax function, a primitive commonly used 
 in new computing models for DNNs, as a case study on evaluating the unique
  programming models adopted ...\n\n\nZheming Jin (Oak Ridge National Labor
 atory (ORNL)) and Jeffrey Vetter (IEEE Computer Society)\n----------------
 -----\nVertical Scaling of Variational Multiscale Modeling for Fluid Dynam
 ics: Successes, Challenges, and Opportunities\n\nWe investigate the vertic
 al scaling of a mixed-precision variational multiscale method. In this met
 hod, the finescales are represented in reduced-precision floating-point fo
 rmat while the coarse scales are represented in double-precision floating-
 point format. We accelerate the solve of the finescal...\n\n\nChristopher 
 Coley (United States Air Force Academy)\n---------------------\nFFTX-IRIS:
   Toward Performance Portability and Heterogeneity for SPIRAL Generated Co
 de\n\nFFTX-IRIS is a dynamic system to efficiently utilize novel heterogen
 eous platforms. This system links two next generation frameworks, FFTX and
  IRIS, to navigate the complexity of different hardware architectures. FFT
 X provides a runtime code generation framework for high performance Fast F
 ourier Tra...\n\n\nSanil Rao and Het Mankad (Carnegie Mellon University), 
 Mohammad Alaul Haque Monil and Jeffrey Vetter (Oak Ridge National Laborato
 ry (ORNL)), and Franz Franchetti (Carnegie Mellon University)\n-----------
 ----------\nInvited Talk: Scaling Computing for Concurrent Data Structures
  Using Near-Memory Processing Architectures\n\nIn recent years, there has 
 been a renewed interest in near-memory processing (NMP) architectures as a
  workaround for the performance and energy issues of frequent and irregula
 r memory accesses.  However, effective use of NMP architectures requires r
 ethinking data structures and their algorithms, esp...\n\n\nIris Bahar (Co
 lorado School of Mines)\n---------------------\nValue-Based Resource Manag
 ement at SoC Scale\n\nValue-based resource management heuristics, which ar
 e traditionally deployed in heterogeneous HPC systems, maximize system pro
 ductivity by assigning resources to each job based on its priority and est
 imated value gain relative to each job's completion time. We investigate t
 he utility of value-based ...\n\n\nSerhan Gener, Sahil Hassan, and Ali Ako
 glu (University of Arizona)\n---------------------\nCHARM-SYCL: New Unifie
 d Programming Environment for Multiple Accelerator Types\n\nAddressing per
 formance portability across diverse accelerator architectures has emerged 
 as a major challenge in the development of application and programming sys
 tems for high-performance computing (HPC) environments. Although the recen
 t performance portability programming systems significantly impr...\n\n\nN
 orihisa Fujita (University of Tsukuba), Beau Johnston (Oak Ridge National 
 Laboratory (ORNL)), Ryohei Kobayashi (University of Tsukuba), Keita Terani
 shi and Seyong Lee (Oak Ridge National Laboratory (ORNL)), Taisuke Boku (U
 niversity of Tsukuba), and Jeffrey Vetter (Oak Ridge National Laboratory (
 ORNL))\n---------------------\nRSDHA – Panel Discussion\n\nSeyong Lee (Oak
  Ridge National Laboratory (ORNL)); Mehmet Belviranli (Colorado School of 
 Mines); Ali Akoglu (University of Arizona); Mohamed Wahib (RIKEN); Ivy Pen
 g (KTH Royal Institute of Technology, Sweden); Ulya Karpuzcu (University o
 f Minnesota, Twin Cities); and Sameer Shende (University of Oregon)\n-----
 ----------------\nRSDHA – Morning Break\n---------------------\nNVMe-Backe
 d GNN Training on GPU Leveraging a Paged UVM Memory System\n\nGraph Neural
  Networks (GNNs) are powerful machine learning models that learn on graph 
 data by extracting embeddings that represent vertex and edge features, as 
 well as graph topology. With graph data scale increasing, and high memory 
 pressure generated from GNN feature data, we turn to out-of-core t...\n\n\
 nBenjamin Wagley (Colorado School of Mines), Pak Markthub (NVIDIA Corporat
 ion), and Bo Wu and Mehmet Belviranli (Colorado School of Mines)\n\nTag: A
 ccelerators, Edge Computing, Heterogeneous Computing\n\nRegistration Categ
 ory: Workshop Reg Pass\n\nSession Chairs: Ali Akoglu (University of Arizon
 a), Mehmet E Belviranli (Colorado School of Mines), and Seyong Lee (Oak Ri
 dge National Laboratory (ORNL))
END:VEVENT
END:VCALENDAR