ComplexHPC Events Wiki Blog
HPCCE: High-Performance Computing on Complex Environments

COST Action

HPCCE: High-Performance Computing on Complex Environments

Heraklion, Crete, Greece
Date: 20 - 24 September, 2010
Program

Schedule and Submissions

Submitted manuscripts should be formatted according to IEEE Cluster 2010 proceedings guidelines.

Session 1: WG1: Numerical analysis for hierarchical, heterogeneous and multi-core systems

Chair: Anne Elster

Efficient and adaptable numerical algorithms are essential to achieving high performance computing on complex environments. As computing systems become more hierarchical and heterogeneous (e.g., a cluster of multiprocessor nodes using multi-core processors), they pose several interesting challenges to numerical algorithms. One of the central challenges is to create models that allow high flexibility for experimenting with techniques (e.g., addressing vectorization and asynchrony, dynamic/adaptive out of order execution, hiding memory latency, blocked storage, heterogeneity, etc,) and consequently, provides a framework for discovering optimal solutions.

The papers that are expected to be presented in this session will be addressing the challenges related to numerical algorithms for complex environments ranging from linear algebra libraries to self-adapting numerical solution of ordinary differential equations (ODEs) or partial differential equations (PDEs). Submissions that cover at least two levels of the system hierarchy, e.g. multiple cache levels/multi-core, distributed memory systems and grids, are highly encouraged.

Subjects: Numerically related algorithms and methods that address one or more of the following:

Programme Committee

Session 2: WG2: Multi-core, Clouds, low-level infrastructure, and green computing

Chair: Mark Baker

New hardware and software technologies are emerging all the time, and there is a never-ending issue of what choices should be used to support both legacy and evolving applications and their algorithms. An example of one of these issues is the use of multi-core processors. The producers of these processors seem to think that a shared-memory model is adequate, but in reality, the HPC market-place will want to use large systems of multi-core processors, where there will be a need for intra and inter communication between the processors and cores. This type of programming can be overcome using MPI/OpenMP/threads, but in reality a programming paradigm that encompasses this type of architecture is needed. Indeed, the way Intel, AMD, IBM Cell, and GPUs processors are designed means that optimisation of code on these systems will take a lot of time and effort. Another area that is changing rapidly is the move away from Grid computing, towards Clouds and Virtualisation, which are innovative, but potentially do not meet the needs of some HPC users. In addition, the need for Green IT computing that optimises application performance and potentially saves power is a very important aspect of computing in the future.

The papers submitted to this session should include:

Programme Committee

Session 3: WG3: Algorithms and tools for mapping and executing applications onto distributed and heterogeneous systems.

Chair: Frederic Suter

Scheduling and resource management are key concepts to achieve High Performance Computing in complex environments. Issues related to these concepts occur at each layer of currently deployed complex infrastructures. Applications usually plan their execution and deployment. Resource brokering and match making are needed at a grid level. Batch schedulers are common at a cluster level. Finally, Operating System's schedulers are now in charge to deal with many cores and hardware accelerators.

The papers expected for this session should ideally present novel work on scheduling or resource management algorithms or services. We strongly encourage submissions that covers at least two levels. Papers on other classical scheduling topics will also be appreciated.

Subjects:

Programme Committee

  • Emmanuel Jeannot, INRIA Research Scientist, Charge de Recherche INRIA,
  • Ozcan Ozturk, Bilkent University, Turkey
  • Rizos Sakellariou, Manchester University,
  • Alexandru Iosup, TU Delft, Netherlands ,
  • Alexey Lastovetsky, School of Computer Science and Informatics, University Colleage Dublin,
  • Helen Karatza, Aristotle University of Thessaloniki, Greece,
  • Olivier Beaumont, INRIA Bordeaux Sud-Ouest, France,
  • Pierre Kuonen, Telecommunications and Computer Science Department, University of Applied Sciences of Western Switzerland, Fribourg, Switzerland,
  • Leonel Sousa, INESC-ID/IST TULisbon, Portugal,
  • Thomas Rauber, University Bayreuth, Germany,
  • Suleyman Tosun, Ankara University, Turkey.
  • Session 4: WG4: Applications of hierarchical-heterogeneous systems

    Chair: Antonio Plaza

    In this session, organized in the framework of the Open European Network for High-Performance Computing on Complex Environments (http://complexHPC.org), the focus will be on applications requiring a large number of processing units to process and analyze scientific data sets. Specifically, papers will be solicited in the area of efficient execution of scientific applications onto complex high performance computing (HPC) infrastructures. Due to the nature of the target infrastructures commonly required in large-scale scientific applications (hierarchical and heterogeneous), addressing this challenge requires solving a lot of issues. In this session, we will pay specific attention to applications that can benefit from the use of a variety of HPC hardware, for example, due to the availability of parallelism at a wide range of grain sizes that can be exploited at different hardware layers (cluster, node, core, or other levels). This session is also intended to support the activities related with the creation of a "catalog of applications" (understood as a comprehensive description of different applications suitable for being addressed as implementation case studies) in the framework of the aforementioned network. Although members of the Applications Working Group of the ComplexHPC network will be invited to submit their work for consideration in the session, the session will remain open to the wide scientific community using HPC techniques and practices to address scientific applications. In this regard, contributions coming from external groups to the ComplexHPC network but which are relevant to the session are also welcome and will be considered in the special session. Only top rated contributions, selected after a peer-review process, will be presented in the session.

    Subjects:

    Programme Committee