International Workshop on Sustainable High Performance Computing 2016 (SHPC2016)

In Collaboration with the 46th Annual International Conference on Parallel Processing (ICPP2016) 

About SHPC2016

Sustainable high performance computing put dependable and autonomic computing in context. Without self-healing, self-configuration and self-monitoring, a high performance computing application will not be able to complete in extreme scales. Without dependability, high performance computing applications are to suffer rapidly decreasing MTBF (mean time between failures) as we expand the size of computing clusters. Security is another dimension that mission critical high performance applications must take into account, or unintended consequences may cause irreparable harms.

While the above discussion made sense at the conceptual level, current research communities are narrowly divided into specific subareas.  In fact, all these topics concern how to handle volatile resources in terms of programming, monitoring, configuration, load distribution/optimization and management. As computing projects are fast become expected services, holistic views are necessary for re-defining the research scopes for the potential applications. In this sense, flawless high performance computing has everything to do with mission critical applications and services.  SHPC2016 Workshop aims to bring researchers in these related areas together to exchange ideas and concepts via the success and failure stories in all relevant areas for the benefit of sustainable high performance computing -- the non-lethal large scale typical distributed computing applications.

Unlike traditional conferences that favor successful novel technologies, SHPC2016 encourages submissions of credible in-depth stories that may be success or failure in dependable computing, autonomic computing, information security and sustainable high performance computing efforts. In fact, we can learn more from failures than successes. For example, in 2015, we have learned that based on Alan Fekete's 1993 impossibility result, the tightly coupled programming APIs (application programming interfaces) are directly responsible for applications scalability dilemma.

In particular, data intensive distributed computing efforts are highly encouraged since they are more difficult to secure, to practice dependability and be made autonomically computable.  

SHPC2016 provides a platform for academic architecture researchers, industry practitioners and government agencies to exchange evolving distributed mission critical computing requirements, ideas and preliminary results.

As computational services are to become integral parts of human societies, performance, dependability, sustainability and security of these services directly impact future social and economic prosperity. SHPC2016 will be a checkpoint of dependable computing, autonomic computing and sustainable HPC efforts. The results may prove to be of fundamentally importance towards the building of 21st century software engineering principles. 

Topics of Interest:

We invite speakers to give talks on the theoretical foundation of programming for volatile resources. Topics include but not limited to the following areas:

  • Theoretical foundation of programming volatile resources.
  • Theoretical architecture concepts for volatile resources.
  • Application scalability analysis using volatile resources.
  • Investigative reports on delivered performance for computation intensive and data intensive applications with failure
  • Investigative results on delivered cloud performances for computation intensive and data intensive applications
  • Theoretical models and experiences in non-conventional HPC programming paradigms
  • Experiences in using auction-based HPC cloud resources
  • Experiences in virtualized GPU for HPC applications
  • Experiences in virtualized network for HPC applications
  • Innovative failure prevention and recovery methods
  • HPC security considerations using cloud resources
  • Communication infrastructure virtualization experiences
  • Private cloud implementation experiences
  • Innovative cloud auction pricing models

    Date: 8/16/2016

    Location: Room 608, Temple University Center City


    8:15 AM: "OSI Standards and the Top Fallacy in Distributed Computing,"  Justin Y. Shi, Temple University 

    9:15 AM:  "Evaluating Online Global Recovery with Fenix using Application-aware In-memory Checkpointing Techniques, " Marc Gamell, Rutgers University 

    Workshop Organizers:

    • Justin Y. Shi, Temple University (Workshop Chair), USA
      • Krishna Kant, Temple University (Workshop Vice Chair), USA
      • Avinash Srinivasan, Temple University, USA
      • Sen Chiao, San Jose State University, USA
      • Dave Yuen, Earth Sciences, University of Minnesota, USA
      • Kitrick Sheets, Cray Inc., USA
      • Joseph Jupin, Temple University, USA
      • Gang Wang, Nankai University, China
      • Xiaoming Liu, Nankai University, China
      • Eagles Song, Parallel Computers Technology Inc., USA

    Workshop Contact: Justin Y. Shi,