Several faculty and students from the Department of Computer Science and Engineering at Texas A&M University participated in the 2014 International Conference for High Performance Computing, Networking, Storage and Analysis (SC) held in New Orleans, Nov. 16-21. SC, the largest and most diverse conference in the computing community, is sponsored by the Association for Computing Machinery and the IEEE Computer Society. It brought together more than 10,000 attendees.
One of the 12 Invited Talks, which is a preeminent SC feature, was a presentation by Dr. Valerie E. Taylor, "MuMMI: A Modeling Infrastructure for Exploring Power and Execution Time Tradeoffs." Taylor described the MuMMI (Multiple Metrics Modeling Infrastructure) framework and gave examples of how MuMMI is used to improve the energy efficiency of parallel applications.
Taylor is the senior associate dean for academic affairs in the College of Engineering and Regents Professor and Wisenbaker Professor in computer science and engineering. She is executive director of the Center for Minorities and People with Disabilities in IT (CMD-IT), an organization with a vision of contributing to the national need for an effective workforce in computing and IT through synergistic activities related to minorities and people with disabilities.
"Designing Dense Linear Algebra Libraries for High Performance Computing" was the subject of Dr. Jack Dongarra's talk for the HPC Interconnections program portion of the conference. He previewed software and algorithm challenges that are faced when designing numerical libraries at scale.
Dongarra is a newly-named Texas A&M University Institute for Advanced Study (TIAS) 2014-2015 Faculty Fellow and TEES Research Professor in computer science and engineering. He is the University Distinguished Professor of Computer Science at the University of Tennessee and the director of the Innovative Computing Laboratory, which coordinates and facilitates IT research efforts at the university. Dongarra holds the title of distinguished research staff in the computer science and mathematics division at Oak Ridge National Laboratory.
Another hallmark of SC is the Birds-of-a-Feather (BOF) sessions, that provide attendees with a venue to openly discuss current topics in high performance computing. Dr. Lawrence Rauchwerger was a panelist for the BOF session on "Asynchronous Many-Task Programming Models for Next Generation Platforms." The discourse revolved around the many variants being proposed for the next generation of platform architectures. Participants were encouraged to rethink AMT programming to influence the direction this body of work takes.
Rauchwerger is noted for his research in parallel computing, compilers, and computer architecture. He is the Eppright Professor in the Department of Computer Science and Engineering and co-director of the Parasol Lab, managing the lab's software and systems group.
Dr. Roger Pearce presented "Faster Parallel Traversal of Scale Free Graphs at Extreme Scale with Vertex Delegates" during the Technical Papers program. The SC14 technical paper committee accepted 84 papers from a total of 394 submissions. Pearce, who received his doctoral degree in computer science from Texas A&M last year, is a computer scientist in the Center for Applied Scientific Computing (CASC) at Lawrence Livermore National Laboratory. The paper was co-authored by Pearce's adviser, Dr. Nancy M. Amato, Unocal Professor in the Department of Computer Science and Engineering, and Dr. Maya Gokhale, distinguished member of technical staff at the Lawrence Livermore National Laboratory. It covers techniques to distribute storage, computation, and communication hubs for extreme scale graphs in distributed memory supercomputers.
Olga Pearce, who is a Ph.D. candidate graduating this December and is also advised by Amato, was one of 16 students selected from a field of 44 to present her research in the Doctoral Showcase. Pearce presented "Load Balancing Scientific Applications" in which she addressed how to evaluate load imbalance quickly and make its correction affordable by devising a framework for decoupling the load balancer from the application, enabling asynchronous load balancing.
Next year's conference, SC15, will be held in Austin, Texas. The conference dates are Nov. 15-20. Of note is the new Student Program, which includes a Student Cluster Competition, Student Volunteers, Mentor-Protégé Program, and Student Job Fair.