CLSAC
  • Home
  • Events
    • CLSAC'12
    • CLSAC'13
    • CLSAC'14
    • CLSAC'15
    • CLSAC'16
    • CLSAC'17
    • CLSAC'18
    • CLSAC'19
    • CLSAC'20
    • CLSAC'21
    • CLSAC'22
    • CLSAC'23
  • George Cotter Award
  • Contact
Picture
SNL researcher Darby Smith with an Intel Loihi neuromorphic computer.

CLSAC 2021

The Greening of Systems and Analytics


October 5 and 6, 2021
Virtual Meeting




Theme

Increasingly, complex control systems require energy-efficient computing, from the edge to large computer centers--including public and private clouds.  Successfully harnessing such large collections of sensors and systems are reshaping core elements of our economy, enabling services like autonomous transportation, smart cities, smart agriculture and smart manufacturing.  However, the opportunity space that must be considered to build such services is increasingly rich, complex, and diverse.  As scale and distribution pushes “The Edge” further out, the need for energy aware analytics and efficient compute, storage and networking resources only grows.  Some might say that this is the time for “Green Energy and Green Analytics.”


To build services like smart agriculture while adhering to tight energy restrictions, we must consider all options in analytics, algorithms, compute, storage, and communication.  Only by considering all of them can we provide the understanding required on vast data volumes arriving at ever-higher velocities within shrinking time-windows.  Fortunately, low-power AI accelerators are emerging across a wide spectrum of devices enabling a transition from large-scale, centralized systems to widely-dispersed computing environments for analytics.  

However, the build out of large-scale, geographically dispersed systems presents a multitude of challenges.  First, we must manage tens of billions of endpoints capable of collecting new and growing sources of data.  To do so, we must consider what mix of resources to devote to compute, storage and network in a wide array of elements: sensors, edge processors, mobile devices, on-premise servers, Smart NICs, Computational Storage, Clouds, and even HPC systems.  These elements have diverse power and operating constraints that, in turn, create significant performance, cost, and software development complexities.  Poor decisions on where to collect, process, and analyze data will result in unacceptable latencies, erroneous conclusions, lost opportunities to leverage data, and low recall.  In system-level terms, we might fail to deliver the intended capability.

In addition, we face other challenges across the system.  Despite today’s AI systems providing increasingly energy-efficient solutions per watt, the explosion of data and the demand for its understanding has simultaneously increased the percentage of world-wide power devoted to AI.   Pushing analytics to the edge attempts to address this issue by moving compute closer to data.  Time-critical decisions and data analytics traditionally computed on centralized systems can be pushed closer to the edge, enabling faster decisions at lower power.

In CLSAC 2021 we explore the greening of computing systems, analytics, and their application across a breadth of core services in modern society.   We’ll cover the role energy use is playing in the design, architecture and operations of large-scale distributed systems, energy-optimized diverse hardware platforms, and the pursuit of ‘greenness’ in large-scale analytics themselves.

This year’s conference will consider:  
  • Opportunities/challenges to improve overall system/analytic performance and efficiency
  • Integration of IoT, Cloud, HPC and Edge computing
  • Managing the complexity of a highly dispersed and heterogeneous computing environment
  • Middleware and Frameworks role in performance and efficiency
  • Green Computing Devices
  • Green Analytics 
  • Failure Rates and Resilience
  • Vulnerabilities

Organizing Committee:
Jim Ang, Pacific Northwest National Laboratory
John Feo, Pacific Northwest National Laboratory
David Haglin, Trovares, Inc.
Ron Oldfield, Sandia National Laboratories
Richard Murphy, Gem State Informatics, Inc.
Almadena Chtchelkanova, National Science Foundation
Brad Spiers, Committee Advisor
Candace Culhane, Los Alamos National Laboratory
Nick Rogers, Department of Defense
TC Tuan, Department of Defense
Steve Pritchard, Committee Advisor
2021 Sponsors
Picture
Picture
Picture
Picture
Picture
Picture
Picture
Picture
Picture
Picture
Picture
Picture


Virtual Agenda (Eastern Standard Time)


Tuesday, October 5
10:30 -- 11:00 am EST
Welcome to CLSAC
Session 1: Innovative System Designs for Green Analytics
Host: Richard Murphy
11:00--11:45 am EST
Keynote: Domain-Specific Accelerators
Bill Dally, NVIDIA
11:45 -- 12:00 pm EST
Student Flash Talk: Causal Discovery for Climate Science and the Energy Exascale Earth System Model
Jake Nichol, University of New Mexico
12:00 -- 12:30 pm EST
Accelerating artificial intelligence with light
Darius Bunandar, Lightmatter
12:30 -- 12:45 pm EST
Break
12:45 -- 1:45 pm EST
Panel Session: Venture Capital       Host: Brad Spiers
  • Insik Rhee, Ventex Ventures
  • Sarah Catanzaro, Amplify Partners
  • Yan Zheng, InQTel
1:45 -- 2:00 pm EST
Break
Session 2: Energy Efficient Hardware Architectures    
Host: Jim Ang
2:00 -- 2:15 pm EST
Student Flash Talk: Biologically Inspired Sparse Coding on a Neuromorphic Device
Kyle Henke, University of New Mexico
2:15 -- 2:30 pm EST
Student Flash Talk: FPGA-Accelerated Ripples
Reece Neff, North Carolina State University

2:30 -- 3:00 pm EST
Exploring Computational Storage for mixed HPC simulation and AI/ML workloads at LANL
Gary Grider, LANL

3:00 -- 3:30 pm EST
Neuromorphic Computing: An Energy Efficient AI Platform for the Future of Computing
Catherine Schuman, ORNL

3:30 -- 3:45 pm EST
Break
3:45 -- 4:45 pm EST
 Panel Session:  Configurability in Future Data Analytic Architectures    Host: Nick Rogers
  • Raghu Prabhakar, SambaNova
  • Kermin (Elliott) Fleming, Intel
  • Derek Chiou, UT Austin
4:45 pm EST
Adjourn

Wednesday, October 6
Random Access
Host: John Feo
11:00 -- 11:10 am EST
Graph Analytics in Arkouda
David Bader, New Jersey Institute of Technology
11:10 -- 11:20 am EST
Performance Portability of an SpMV Kernel Across Scientific Computing and Data Science Applications
Stephen Olivier, SNL
11:20 -- 11:30 am EST
Sustainable Scientific Supercomputing
Jeremy Kepner, MIT Lincoln Lab
11:30 -- 11:40 am EST
Compressing Trained Neural Networks with Tensor Decompositions
Dimitri Leggas, Reservoir Labs
11:40 -- 11:50 am EST
Fast and Ethical AI Computing

Vijay Gadepally, MIT Lincoln Lab

11:50 -- 12:00 pm EST
SODALITE: Software Defined Accelerators from Learning Tools Environment

Antonino Tumeo, PNNL

12:00 -- 12:10 pm EST
State of RAPIDS cuGraph – going big

Brad Rees, NVIDIA

12:10 -- 12:20 pm EST
The Coming Sixth Epoch of Distributed Computing

Timothy Mattson, Intel

12:20 -- 12:35 pm EST
Break
12:35 -- 12:45 pm EST
Analyzing data movement through storage, network, and memory
Nathan Tallent, PNNL
12:45 -- 12:55 pm EST
Wafer-scale processors for HPC

Rob Schreiber, Cerebras

12:55 -- 1:05 pm EST
Continuous analytics for operator readiness using commercial hardware

Grant Gillary, Amplio AI

1:05 -- 1:15 pm EST
ExaFlop and new PetaFlop systems in China
David Kahaner, ATIP
1:15 -- 1:25 pm EST
Accelerating Graph Algorithms with NVSHMEM
Oded Green, NVIDIA
1:25 -- 1:35 pm EST
1:35 -- 1:45 pm EST
Big Data Threat– Russian Federation Campaign Solar Winds
George Cottter
1:45 -- 2:00 pm EST
Break
Session 3: Applications
Host: David Haglin
2:00 -- 2:45 EST
Keynote: The Green Global 5G Network as the World’s Largest Cloud Computer
Steve Papa, Parallel Wireless
2:45 -- 3:00 pm EST
Student Flash Talk: POSIX number representations for computing
Alexandra Poulos, Clemson University
3:00 -- 3:30 pm EST
Analog Processing and Analytics:  the "Outer Edge"
Dave Robertson, Analog Devices
3:30 -- 3:45 pm EST
Break
3:45 -- 4:45 pm EST
Panel Session: Greening of Applications
  • Chandra Krintz, UCSB
  • Xiaoyuan Fan, PNNL
  • Wu Feng, Virginia Tech
Host: Almadena Chtchelkanova
4:45 pm EST
Planning for 2022 CLSAC and Adjourn
Committee

Thank you for attending CLSAC.  See you next year in Annapolis (hopefully)!  Please complete the feedback survey.
The CLSAC web site is sponsored by the Association for High-Speed Computing.