• NSF Sponsored Workshop on

    Architecture and Software for
    Emerging Applications (WASEA)

    September 11-12, 2016
    -King Solomon Hall-
    Held in conjunction with the 25th International Conference on Parallel Architecture and Compiler Techniques (PACT) in Haifa, Israel.


    Lawrence Rauchwerger, Texas A&M
    Wen-mei Hwu, UIUC
    Nancy Amato, Texas A&M


    High-valued domain applications in areas such as medicine, biology, physics, engineering, and social phenomena demand both fast innovation and high execution speed and require productive development environments for domain experts who may not be computer science experts.  For example, image recognition methods from many fields use deep neural networks that evolve very quickly.  At the same time, better tools for machine learning can grant us access to a wealth of existing data that is not yet used.
    This workshop brings together leading researchers in architecture, compilers and programming languages, and domain experts to discuss and debate potential approaches to accelerating progress in such high-valued domains with an emphasis on developing strategies for exploiting machine learning, including strategies for accelerating learning algorithms through parallelism.  The goal is to stimulate an in-depth discussion of the potential benefits of joint architecture and compiler approaches.


    Sunday 11 Sept., 2016

    10:30 – 10:45 – Yan Solihin – NSF Programs

    10:45 – 12:15 – Session I: Machine Learning (6 short talks)

    • Yale Patt, Machine Learning and Other Sacred Cows, Is Anyone Listening?
    • Kunle Olukotun, Everything You learned About Correct Parallel Computation is Wrong for Analytics!
    • Mattan Erez, Machine-Learning Problems and Near-Memory Processing
    • Gabriel Silberman, Who moved my Artificial Cheese?
    • Xipeng Shen, Can Compilers Generate Better Learning Algorithms than Experts do?
    • David Padua, Dynamic Languages for Parallel and Distributed Computing

    12:15 – 13:50 – Lunch

    13:50 – 15:35 -  Session II: More ML and Some Hardware (7 short talks)

    • Mike Gschwind, Accelerating Cognitive Applications with an Open Ecosystem Approach
    • Jaejin Lee, Scalable Heterogeneous Programming Models for Deep Learning Applications
    • Ronny Ronen, Machine Learning Workloads – Speed them up, take their energy down
    • Ravi Iyer, Architectures and Tools for Continuous Learning
    • Alex Nicolau, Acceleration of Deep Learning Applications
    • Albert Cohen, Building blocks for computational cyberphysical systems
    • Hironori Kasahara, Parallelization and Power Reduction Compiler for Heterogeneous  Multicores for Emerging Applications

    15:35 – 16:00 – Coffee Break

    16:00 – 17:45 – Session III: HW-SW Interaction (7 short talks)

    • Arvind, Are OOO processors relevant for Key-Value stores (KVS) and other Data Center Applications?
    • Kemal Ebcioglu, Memory partitioning in the limit
    • Per Stenstrom, Cutting Through the Transformation Hierarchy Through Cooperative Hardware-Software Cache Management
    • Mark Silberstein, An Accelerator-Centric Operating System: Rethinking the Role of CPUs in Modern Computers
    • Andrew Lenharth, Compiling Graph Algorithms to Distributed, Heterogeneous Systems
    • Arturo Gonzalez, Dealing with parallelism at run-time for new dynamic evolving applications
    • Yoav Etsion, O-structures: Tracking Fine-Grain, Inter-task Data Dependencies in Memory

    Monday 12 Sept., 2016

    09:00 – 11:00 – Panel and Open Discussion


    The workshop will produce a report providing recommendations on:

    • Joint compiler/language and architecture approaches
    • Compiler/language support enabling more aggressive hardware capabilities
    • Architecture support enabling more effective compilers
    • Applications whose development process could benefit by these advances

    The report will identify research opportunities in the interaction between developers, and architecture and language/compiler researchers to enable productive domain application development and highly efficient and scalable implementation on heterogeneous computing systems. The report will outline promising approaches and the research required for these approaches to become usable by the domain application developers.


  • Dates & Deadlines


    • Conference Registration
      Early Bird Registration Cutoff Date - August 15, 2016

    • Student Travel Grants
      Application Deadline: July 25, 2016

    • Main Conference Papers
      Abstract Deadline: March 14, 2016
      Paper Deadline: March 21, 2016
      Rebuttal after Phase I: May 7 -- May 11, 2016
      Rebuttal after Phase II: June 15 -- June 19, 2016
      Author Notification: June 30, 2016
      Camera Ready Final Papers: July 27, 2016

    • Workshops + Tutorials
      Proposal Deadline: April 8, 2016 (23:59 PST)
      Acceptance Notification: April 22, 2016 (23:59 PST)

    • ACM Student Research Competition
      Abstract Deadline: June 24, 2016
      Author Notification: July 12, 2016