ISC High Performance 2018 Workshop


 

 

Workshop on Sustainable Ultrascale Computing Systems.

ISC High Performance 2018

Thursday, June 28, 2018, Marriott Frankfurt Hotel

 

This workshop is aimed to present high quality research topics related to Sustainable Ultrascale Computing Systems. (UCS) are envisioned as large-scale complex systems joining parallel and distributed computing systems, maybe located at multiple sites that cooperate to provide solutions to the users, that will be two to three orders of magnitude larger that today s systems. Sustainability is understood not only in terms of energy, but also related to programmability, usability, and maintenance to facilitate the adoption of UCS.

The workshop is aimed  present the major results of the COST Action IC1305, “Network from Sustainable Ultrascale Computing Systems (NESUS)”. Ultrascale computing systems, included the Research Roadmap proposed. But it is also open for external speakers proposing realistic solutions addressing major challenges of building sustainable ultrascale computing systems, as well as developing collaborative activities among the involved research groups to target cross-layer design issues to offer a unified view of ultrascale platforms.

 

Topics

 

The topics of the workshop are shown below.

  • New programming paradigms, runtimes, and middlewares: techniques to increase productivity, scalability, and reliability of parallel and distributed programming; new programming models suitable for UCS; adaptive runtimes for parallel and distributed systems ; and malleable applications that can adapt their resource usage at runtime.
  • New approaches of continuous running in the presence of failures: resilient schedulers; reactive and proactive error handling; and monitoring and assessment of failures.
  • Sustainable data management proposals for addressing the predicted exponential growth of digital information. synergistic approaches for the whole data life cycle in parallel and distributed systems; new proposal to enhance the Input/Output (I/O) stack; proposal for advanced predictive and adaptive data management; and techniques to improve data locality.
  • Energy efficiency of ultrascale systems: novel metrics for energy monitoring, profiling and modeling in ultrascale components and applications; energy-aware resource management; and hardware/software codesign techniques to enhance energy.
  • Applications: high-level algorithms, and services amenable to UCS; redesign and reprogramming techniques for applications to efficiently exploit ultrascale platforms; energy-aware algorithms.

 

Preliminary Program

 

 

9:15 – 9:30 Welcome

Jesús Carretero, University Carlos III of Madrid

9:30 – 10:15 Invited Speaker 1

Rosa Badía, Barcelona Supercomputing Center, Spain

Title: Programmability-performance trade-off  with a task-based approach

10:15-11:00 Invited Speaker 2

Thomas Ludwig, DKRZ, Germany

Title: Reproducibility in Computer Science & Climate Science in the Era of Exascale Systems

11:00 – 11:30 Coffee break
11:30 – 13:00 Papers presentation session

 

Vladimir Voevodin, Lomonosov MSU, Russia.

  • Using the AlgoWiki Project to Compare Any Supercomputers on Any Algorithms

Radu Prodan, Klagenfurt University, Austria.

  • Multi-objective workflow scheduling on clouds

Alexey Lastovetsky, UCD, Ireland.

  • How pre-multicore methods and algorithms perform in multicore era.

Silvina Caíno, Uc3M, Spain.

  • An architecture for scaling Spark in HPC systems.
13:00 – 14:30 Lunch
14:30-16:00 Papers presentation session

Leonel Sousa, INESC, Portugal & Estefania Serrano, UC3M, Spain

  • Cache-aware Roofline Model and Medical Imaging Optimization

Nina Popova, Lomonosov MSU, Russia.

  • A Practical Approach for Ultrascale Computing Education at Moscow State University

Gabriel Pedraza Ferreira, UIS, Colombia.

  •  Towards an Efficient Software Model Supporting UltrasScale Applications

Jose Luis Gonzalez-Compean, CINVESTAV, Mexico.

  • Application composition by using containerized building blocks.
16:00 – 16:30 Coffee break
16:30 – 17:45  

Panel. “Landscape of Sustainable Ultrascale/Exascale Computing Systems in the EU programs”

Panel Chair: Jesus Carretero, University Carlos III of Madrid

Speakers:

  • Xavier VIGOUROUX – Bull/Atos
  • Marcin Ostasz, ETP4HPC
  • Tomi Ilijas, ARCTURC
  • Thomas Farhinger, University of Innsbruck, Austria
17:45 – 18:00 Closing

 

Invited Talks

 

 
Rosa M. Badía. Barcelona Supercomputing Center, Spain
Title: Programmability-performance trade-off with a task-based approach

Abstract:
Programming large-scale systems poses several challenges to scientific application developers. Systems are growing in complexity, and the concurrency is not the only difficulty that programmers face, but also the distribution of memory spaces, complex architectures, new storage technology, etc. Task-based programming models have proven to be a good alternative to traditional programming models for HPC. PyCOMPSs is a pioneer approach on task-based programming in Python that enables simple sequential codes to be executed in distributed computing platforms. The talk will present an overview of PyCOMPSs and show how it offers a good trade-off between programmability and performance in HPC.

Bio:
Rosa M. Badia holds a PhD on Computer Science (1994) from the Technical University of Catalonia (UPC). She is the manager of the Workflows and Distributed Computing research group at the Barcelona Supercomputing Center (BSC).
Her current research interest are programming models for complex platforms (from multicore, GPUs to Cloud). The group lead by Dr. Badia has been developing StarSs programming model for more than 10 years, with a high success in adoption by application developers. Currently the group focuses its efforts in PyCOMPSs/COMPSs, an instance of the programming model for distributed computing including Cloud.
Dr Badia has published near 200 papers in international conferences and journals in the topics of her research. Her group is very active in projects funded by the European Commission and in contracts with industry.

 
Thomas Ludwig, DKRZ, Germany
Title: Reproducibility in Computer Science & Climate Science in the Era of Exascale Systems

Abstract:
Reproducibility of scientific results became a heavily debated issue in the last recent years. Looking closer into that field we find many different levels of concern. In computational sciences where we conduct experiments in silico and receive results from computer simulations we are faced with special aspects of reproducibility. Can we run third party code and reproduce the result that formed the basis for a publication? Following the rules of good scientific practice: can we even reproduce our own results after a certain time with perhaps a modified computer infrastructure? Finally, if we give our best, what is the influence of the hardware and the processor itself? Can we achieve a bitwise reproduction of a single program run? And if so, how is this connected to reproducibility of findings in science? The talk will discuss several of these critical issues and concentrate on aspects of programming and hardware issues.

Bio:

Thomas Ludwig is a professor for scientific computing at the Universität Hamburg, Hamburg, Germany. At the same time he is the director of the German Climate Computing Centre (DKRZ) in Hamburg. His research focus as a professor is on high performance storage systems (parallel I/O, parallel file systems) and energy efficiency with high-performance computing. As director of DKRZ he takes the responsibility to provide the climate research community with high computational power, considerable storage space and sophisticated services. Thomas Ludwig received his doctoral degree from Technische Universität München, Germany, in 1992. He received the habilitation degree in 1998 at the same university. From 2001 to 2009 he held a professor position at the Universtät Heidelberg, Germany. Thomas Ludwig works in HPC since 1987 and has published more than 180 papers. In 2010 he launched the international conference series on Energy-Aware High Performance Computing.

 

Workshop Organization

Workshop Chair

Jesus Carretero,  University Carlos III of Madrid, Spain

Program Committee
Carlos Jaime Barrios, Universidad Industrial de Santander, Colombia
Angelos Bilas, ICS FORTH, Greece
Harold Castr, Universidad de los Andes, Colombia
F. Javier García-Blas, University Carlos III of Madrid, Spain
J. Daniel Garcia,  University Carlos III of Madrid, Spain
Alexey Lastoveysky, University College, Dublin, Ireland
Laurent Lefevre, Ecole Normale Superiour Lyon, France
Svetozar Margenov, Bulgarian Academy of Science, Bulgaria
Radu Prodan, Alpen-Adria University of Klagenfurt, Austria
Gudula Ruenger, Technical University Chemnitz, Germany
Leonel Sousa  INESC, Lisbon, Portugal
Xavier Vigoroux, ATOS-BULL, France
Vladimir Voevodin,  Lomonosov Moscow State University, Russia
Zhiyi Huang, University of Otago, New Zealand