Workshop on HPC services

in conjunction with SYNASC 2012

27-29 September 2012

 

The event is built upon four distinct parts:

HPC workshop – sessions of regular papers, 28 September

The purpose of this workshop is to present and discuss the state-of-the-art in high-performance scientific computing with applications and various fields, like environment, chemistry, physics etc. Thus, the presentations will focus on applications needing high-performance computing, physical modeling of large-scale problems, and the development of scalable algorithms for solving large scale problems on modern parallel and distributed high-performance computing platforms, including multicore architectures, GPGPUs/GPUs, and clusters. Furthermore, during the workshop the participants will have the opportunity to discuss and share the latest research in parallel and distributed high performance computing systems applied to scientific problems. The emphasis of this workshop will be on running complex realistic applications at sustained performance in production-grade HPC environments. Scalability studies of complex computing codes on HPC platforms and the tools and development environments facilitating improved scalability are among the expected contributions.

Topics:

Specific topics for this workshop include, but are not limited to, the following:

  • Multicore/manycore architectures in scientific applications
  • GPU support for applications
  • Cluster, Grid and Cloud Computing in scientific applications
  • Parallelization of compute- or data-intensive tasks in scientific applications
  • Data handling, integration and visualization in scientific applications
  • Distributed infrastructures for scientific applications
  • Programming paradigms for high-performance computing in scientific applications
  • Tools and programming environments supporting high-performance computing in scientific applications
  • Scheduling in high-performance computing for scientific applications
  • Workflow management and remote collaboration in scientific applications
  • System level support for high-performance computing in scientific applications
  • Fault-tolerance of distributed scientific applications
  • Scalability of infrastructures and applications in scientific applications

Chairs:

  • Emil Slusanschi – University “Politehnica” of Bucharest, Romania
  • Ralf-Peter Mundani – Technische Universität München, Germany
  • Ioan Lucian Muntean – Technical University of Cluj-Napoca, Romania

Programme Committee:

  • Tobias Weinzierl – Technische Universität München, Germany
  • Rodica Potolea – Technical University of Cluj-Napoca, Romania
  • Alin Murarasu – Technische Universität München, Germany
  • Ana Gainaru – University of Illinois at Urbana-Champaign, US
  • Alexandru Olteanu – University “Politehnica” of Bucharest, Romania
  • Voichita Iancu – University “Politehnica” of Bucharest, Romania

Deadlines:

  • Call for papers: May 2012
  • Full paper submissions: 20 July 2012 30 July 2012 (firm deadline)
  • Notifications: 15 August 2012 20 August 2012
  • Revised papers and registrations: 1 Sept 2012
  • Event: 28-29 Sept 2012
  • Publication of selected papers in IEEE proceedings of SYNASC 2012: Jan 2013
  • Extended version of selected papers for SCPE: Feb 2013
  • Notifications: March 2013
  • Publication of selected papers in SCPE: April 2013

Submission system:

Up to 8 pages IEEE CS Format – please follow the links from SYNASC Workshops 2012.

HPC training, 27-28 September

The aim of the training is to offer the knowledge and to acquire practical skills for HPC.

Tentative agenda:

  1. Introduction to parallel processing and parallel processors: where does MPI fit into the landscape? Concepts of batch running and scheduling: PBS/Torque scheduler. Practical work (C or Fortran): simple MPI programs.
  2. Point-Point Communication (blocking). Timing programs (and its pitfalls). Practical work (C or Fortran): point-point communication.
  3. Blocking versus non-blocking Point-Point. Collective Communication & Reductions.  Practical work (C or Fortran): collective communication. Various parallel libraries.
  4. Cartesian Topologies. Practical work (C or Fortran): topologies. User-defined MPI types and operators.  Practical work on these.

Trainer:

  • James Davenport, University of Bath, UK

 

Grid tutorial – 29 September

Grids have become a common workplace for computational scientists. The Globus Toolkit (GT), the de-facto standard middleware for grids, is a mature grid technology with world-wide outreach. The tutorial will provide first-hand practical information about Globus Toolkit v.5, valuable to resource providers, computing site administrators, end-users of the grid.

The tutorial targets both grid administrators and users of GT, offering/providing first-hand information and experience sharing from European Globus specialists. The focus of the tutorial is set on the latest version 5 of the toolkit: Authentication and Authorization, GridFTP, GRAM5, MyProxy etc. Globus services included in the Unified Middleware Distribution of EGI will be presented as well. Furthermore, the novel GlobusOnline cloud service for reliable file transfer will be demonstrated.

Virtual machines for the tutorial participants and remote resources for the training will be provided by the IGE project as part of the IGE Testbed (the IGE project is a consortium of Globus experts that coordinates Globus activities and developments in Europe).

Tentative agenda:

  • Session 1: lectures/presentations with examples
  • Session 2: hands-on exercises: setup of a grid from scratch; common user scenarios: job submission, file transfer; introduction to the GlobusOnline file transfer service

 Trainers:

  • Ioan Lucian Muntean, Technical University of Cluj-Napoca, Romania
  • Marius Joldos, Technical University of Cluj-Napoca, Romania

 

Affiliated workshops

The topics of interest in relationship with the HOST activities are encountered in the affiliated workshops organized in the same time with the HPC tracks :

  • Cloud computing: MICAS 2012 – Workshop on Management of Resources and Services in Cloud and Sky Computing
  • Artificial intelligence: ACSys 2012 – Workshop on Agents for Complex Systems and NCA 2012 – Natural Computing and Applications
  • Numerics: IAFP 2012 – Workshop on Iterative Approximation of Fixed Points

 

Organization

General chairs:

  • Dana Petcu – West University of Timisoara, Romania
  • Krzysztof Kurowski – Poznan Supercomputing and Networking Center, Poland

Local organizers:

  • Marc Frincu, West University of Timisoara, Romania
  • Silviu Panica,  West University of Timisoara, Romania