Weekly outline

  • General

    The seminar covers optimization and parallelization techniques for modern multi- and manycore systems. The topics are chosen from interesting contemporary problems in High Performance Computing on modern hardware like multicore processors, accelerators (e.g., GPGPUs or Xeon Phi), and clusters.

    Lecturer: Prof. G. Wellein (gerhard.wellein@fau.de), Martensstr. 1, Room 01.131. Phone -28136
    Location: 2.037 (e-Studio), Martensstr. 1 (RRZE), 2nd floor
    Time: Tuesday 8:30-10:00

    Either 2.5 or 5 ECTS credits will be granted, depending on whether the student gives one or two talks. In either case, a written seminar report is mandatory.

    Possible topics can be found in the intro talk (see below).


  • 12 October - 18 October

    First seminar: Tuesday, October 13, 8:30, RRZE eStudio

    G. Wellein: Intro, presentation of seminar topics

    Dominik Ernst: "First results for tall skinny MMM on Nvidia GPGPUs"

  • 19 October - 25 October

    Guest talk (KONWIHR)

    Markus Kiunke, Universität Würzburg: "Monte Carlo X-ray Scattering Studies in the MeV Regime".

    Abstract: In recent years the photon energy for X-ray scanning tasks in the non-destructive testing field has risen up to peak energies of 9 MeV. Since the usual Monte Carlo (MC) simulations are very time consuming, deterministic X-ray simulations are developed to provide a fast and cheap estimate of the high energy X-ray images. The talk will focus on preliminary MC studies to evaluate the influence of energy, material and length on the different scattering effects in sample objects and the issues to be addressed within the KONWIHR grant.

    • 26 October - 1 November

      Guest talk

      Danilo Guerrera, Antonio Maffia   (University of Basel): "Achieve Reproducible Research with PROVA!:
      Performance Reproduction of Various Applications" Slides

      Preparation talk for PMBS15 workshop

      Julian Hammer: "Automatic Loop Kernel Analysis and Performance Modeling With Kerncraft"

      • 2 November - 8 November

        Seminar shifted to October 27 (talk by Julian Hammer)

        • 9 November - 15 November

          Guest talk:

          Dr. Karel de Vogeleer, TELECOM ParisTech: "The energy-frequency convexity rule" (cancelled due to airline strike)

          Marco Heisig: Common Lisp in High Performance Computing

          High Performance Computing (HPC) is about utilizing parallel computers that can execute up to several peta-operations per second. But as the compute power increases from year to year, the complexity of programming such systems grows almost at the same rate. Traditional HPC programming languages like C, C++ and Fortran allow to write fast numerical codes, but offer no good solution on how to deal with the complexity of massively parallel computing. The Lisp family of languages on the other hand was developed as a tool for artificial intelligence and is very well suited for such tasks. We investigate whether Lisp allows more robust and productive use of modern parallel computers. In this talk I give a gentle introduction to Lisp and present our current research with respect to Lisp and parallel computing.

        • 23 November - 29 November

          Master Thesis status presentation (8:30):

          Dominik Thoennes:  Symmetric sparse matrix-vector multiplication on multicore CPUs (cancelled - moved to next week)

          Guest talk (08:45 am):

          Dr. Karel de Vogeleer, TELECOM ParisTech:  Energy Optimization of Computer Systems via Frequency Scaling


          Energy efficiency is an important aspect for energy-critical computer systems, e.g., super computers or battery-driven devices. In this talk we look at the trade-off between energy consumption and performance of a computer system including a microprocessor, which we refer to as the Energy/Frequency Convexity Rule. We discuss under what circumstances this convexity rule can be exploited, and when other methods are more effective, to improve the energy efficiency of the computer system. Among the parameters that affect the convexity rule are; the microprocessor's thermal behavior, the overhead of the system, the background power requirements of the system, and others.

          • 30 November - 6 December

            Master Thesis status presentation (8:30):

            Dominik Thoennes:  Symmetric sparse matrix-vector multiplication on multicore CPUs

            • 7 December - 13 December

              Guest talk:

              Christopher Bross: Implementing the HPCG benchmark in HPX.

            • 14 December - 20 December

              Jacob Snoeijer: Benchmarking the semi-stencil algorithm. (1st seminar talk) Slides

              Guest talk:

              Dominik Ernst: Report on Intel code optimization contest

              • 21 December - 27 December

                Sebastian Friedemann: Validating the multicore power model (1st seminar talk)

                Guest talk:

                Dr. Harald Koestler (LSS): Code generation for stencils

                • 11 January - 17 January

                  8:45: First seminar talk by Viktor Haag: Implementing ILBDC on GPGPUs.

                  • 18 January - 24 January

                    Guest talk:

                    09:00: Hartwig Anzt (Innovative Computing Lab, University of Tennessee, Knoxville): Parallel Threshold ILU

                    • 25 January - 31 January

                      No seminar today