5th Thematic CERN School of Computing

Europe/Zurich
MEDILS in Split, Croatia

MEDILS in Split, Croatia

Ivica Puljak (University of Split. Fac.of Elect. Eng., Mech. Eng. and Nav.Arc) , Sebastian Lopienski (CERN)
Description

t CSC 2017
Efficient Parallel Processing of Future Scientific Data

The 5th Thematic CERN School of Computing will be held in Split, Croatia from Monday 5 June to Friday 9 June 2017 evening. Students should arrive on Sunday 4 June and leave on Saturday 10 June.

The school is organized by CERN in collaboration with the University of Split (FESB - www.fesb.hr) and is hosted at the Mediterranean Institute For Life Sciences (MEDILS) Conference Centre (www.medils.org). The Centre is a historical renovated building situated in a wooded and landscaped park located on the Adriatic Sea coast, a few kilometres from the centre of Split.

Applications are now closed.

Important Dates

We have 24 places at the School available. Due to the popularity of our School, only selected applicants may attend. Will you be one of them?

  • Sunday 5 March 2017 (midnight CET/GMT+1). Deadline for application.
  • Monday 27 March - you will be informed about the outcome of the selection by email.
  • Friday 21 April - latest date for School fees payment if you are accepted.
  • Sunday 4 June afternoon/evening - student arrivals at MEDILS.
  • Saturday 10 June morning - departure.
tCSC2017 Virtual Machine
    • 15:00 19:00
      Registration 4h
    • 20:00 21:00
      Dinner at MEDILS 1h
    • 09:00 10:00
      Opening Session
      Conveners: Ivica Puljak (University of Split. Fac.of Elect. Eng., Mech. Eng. and Nav.Arc) , Sebastian Lopienski (CERN) , Sven Gotovac (FESB)
      • 09:00
        Introduction to FESB 20m
        Speaker: Sven Gotovac (University of Split. Fac.of Elect. Eng., Mech. Eng. and Nav.Arc)
      • 09:20
        Introduction to the thematic School of Computing 40m
        Speaker: Sebastian Lopienski (CERN)
    • 10:00 11:00
      Inaugural Scientific Lecture
      Convener: Ozren Polasek (University of Split)
    • 11:00 11:15
      Updates 15m
    • 11:15 11:45
      Coffee 30m
    • 11:45 12:45
      Introduction to Efficient Computing 1h
      Speaker: Andrzej Nowak
    • 12:45 14:00
      Lunch 1h 15m
    • 14:00 15:00
      Intermediate Concepts in Efficient Computing 1h
      Speaker: Andrzej Nowak
    • 15:00 15:30
      Break 30m
    • 15:30 16:30
      Exercise 1 1h
      Speaker: Andrzej Nowak
    • 16:30 17:30
      Exercise 2 1h
      Speaker: Andrzej Nowak
    • 17:30 17:45
      Free time 15m
    • 17:45 18:00
      Transport to Split 15m
    • 18:00 20:15
      Visit to Split 2h 15m Split

      Split

    • 20:15 22:30
      Outside Welcome Dinner 2h 15m Re Di Mare

      Re Di Mare

      http://www.redimare.com/

      http://www.redimare.com/

    • 09:00 10:00
      Scientific software programming: a modern approach 1h

      Lecture 1

      Speaker: Danilo Piparo (CERN)
    • 10:00 11:00
      Data Oriented Design 1h

      Lecture 3

      Speaker: Andrzej Nowak
    • 11:00 11:15
      Updates 15m
    • 11:15 11:45
      Coffee break 30m
    • 11:45 12:45
      Exercise 3 1h
      Speaker: Andrzej Nowak
    • 12:45 13:30
      Lunch 45m
    • 13:30 16:00
      Study time and/or Daily Sports Program 2h 30m
    • 16:00 16:30
      Break 30m
    • 16:30 17:30
      Structuring data for efficient I/O 1h

      Lecture 1

      Speaker: Sebastien Ponce (CERN)
    • 17:30 18:30
      Expressing Parallelism Pragmatically 1h

      Lecture 2

      Speaker: Danilo Piparo (CERN)
    • 18:30 19:30
      Many ways to store data 1h

      Lecture 2

      Speaker: Sebastien Ponce (CERN)
    • 19:30 20:00
      Free time 30m
    • 20:00 21:00
      Dinner at MEDILS 1h
    • 09:00 10:00
      Protection of Resources and Thread Safety 1h

      Lecture 3

      Speaker: Danilo Piparo (CERN)
    • 10:00 11:00
      Ensure Correctness of a Parallel Scientific Application 1h

      Lecture 4

      Speaker: Danilo Piparo (CERN)
    • 11:00 11:15
      Updates 15m
    • 11:15 11:45
      Coffee 30m
    • 11:45 12:45
      Exercise 1 1h
      Speaker: Danilo Piparo (CERN)
    • 12:45 13:45
      Lunch 1h
    • 13:45 19:00
      Sports afternoon 5h 15m Cetina gorge

      Cetina gorge

      Rafting near Omis, Free time in Omis

    • 19:00 21:00
      Outside Dinner 2h Radmanove Mlinice

      Radmanove Mlinice

      http://www.radmanove-mlinice.hr/index.php/en/radman-s-mills
    • 09:00 10:00
      Preserving Data 1h

      Lecture 3

      Speaker: Sebastien Ponce (CERN)
    • 10:00 11:00
      Key Ingredients to achieve effective I/O 1h

      Lecture 4

      Speaker: Sebastien Ponce (CERN)
    • 11:00 11:15
      Updates 15m
    • 11:15 11:45
      Coffee 30m
    • 11:45 12:45
      Exercise 2 1h
      Speaker: Danilo Piparo (CERN)
    • 12:45 13:30
      Lunch 45m
    • 13:30 16:00
      Study time and/or Daily Sports Program 2h 30m
    • 16:00 16:20
      Break 20m
    • 16:20 16:30
      Official School Photo 10m
    • 16:30 17:30
      Exercise 3 1h
      Speaker: Danilo Piparo (CERN)
    • 17:30 18:30
      Exercise 1 1h
      Speaker: Sebastien Ponce (CERN)
    • 18:30 19:30
      Exercise 2 1h
      Speaker: Sebastien Ponce (CERN)
    • 19:30 20:00
      Free time 30m
    • 20:00 21:00
      Dinner at MEDILS 1h
    • 09:00 10:00
      Exercise 3 1h
      Speaker: Sebastien Ponce (CERN)
    • 10:00 11:00
      Spare exercise hour 1h
    • 11:00 11:15
      Updates 15m
    • 11:15 11:45
      Coffee 30m
    • 11:45 12:45
      Summary and Future Technologies Overview 1h

      Lecture 4

      Speaker: Andrzej Nowak
    • 12:45 14:00
      Lunch 1h 15m
    • 14:00 15:00
      Self-Evaluation by students 1h
    • 15:00 15:45
      Closing lecture - Some Aspects of Astroparticle Physics 45m
      Speaker: Nikola Godinovic (University of Split. Fac.of Elect. Eng., Mech. Eng. and Nav.Arc)
    • 15:45 16:00
      Break 15m
    • 16:00 17:30
      Student presentations 1h 30m
      Speakers: Christina Quast (Technische Universitaet Berlin (DE)) , Flavio Pisani (Universita e INFN, Bologna (IT)) , Georgios Voulgarakis (CERN) , Jose Maria Benlloch Rodriguez (Univ. of Valencia and CSIC (ES)) , Michela Paganini (Yale University (US)) , Mykhailo Dalchenko (Texas A & M University (US)) , Nikola Hardi (University of Novi Sad (RS)) , Tea Marasovic , Victoria Tokareva (Joint Institute for Nuclear Research)
      • The nature of neutrinos and the NEXT experiment 10m

        I will discuss one of the fundamental questions about neutrinos: are they their own antiparticles? NEXT (Neutrino EXperiment TPC) is trying to answer this question, we will see how it can be done and what the implications could be.

        Speaker: Jose Maria Benlloch Rodriguez (Univ. of Valencia and CSIC (ES))
      • Data acquisition system on gas-electron multiplier muon detectors in CMS at the LHC 10m

        GEM detectors are among the first detectors employing Zynq CPU coupled by high speed AXI bus to the FPGA in the backend electronic board. This allows to implement register-intensive operations as remote procedure calls, allowing fast detector calibration and convenient monitoring environment.

        Speaker: Mykhailo Dalchenko (Texas A & M University (US))
      • Speeding up C++ code for fun and profit 10m

        Quick overview of the techniques I used to speed up some physics code. Furthermore, introduction to roofline plots.

        Speaker: Christina Quast (Technische Universitaet Berlin (DE))
      • Autopilot for Linac4 H-Ion Source 10m

        A real-time failure-tracking & optimization software for the Linac4 H- Ion Source, making it even more of a black box than it already is!

        Speaker: Georgios Voulgarakis (CERN)
      • Parallel partial wave analysis for the BES-III experiment 10m

        Partial wave analysis is a fundamental technique for extracting hadron spectra and hadron decay properties. It is widely employed in current experiments in particle physics, particularly at BES-III.

        The analysis is typically performed using the event-by-event maximum likelihood method. For the BES-III experiment, fitting the accumulated data (about 1.225 billion J/psi decays) using currently employed software takes a long time, which significantly complicates and sometimes restricts data analysis. Fortunately, computing the likelihood function can be naturally parallelized. Thus the development of new multicore CPU's and GPU's makes using parallel programming technologies natural to decrease the data fitting time.

        The talk will be about the development of highly scalable parallel framework for the partial wave analysis. The framework is intended to accelerate the calculations on modern multicore CPUs and CPU-like coprocessors by employing OpenMP parallel computing technology, high-performance computing optimizations like vectorization or aligned memory access, and offloading computationally intensive parts of the code to massively parallel co-processors.

        Current results will be presented.

        Speaker: Victoria Tokareva (Joint Institute for Nuclear Research)
      • Evolutionary code optimisation 10m

        Compilers optimise programs by transforming the code in such way that it improves performance, produces smaller executables or similar. It is very difficult to predict effects of a single transformation and even more difficult to predict how multiple transformations combine. Sometimes it's worth trying to compile programs many times, apply transformations even if improvement is uncertain and hopefully produce better results. I will present a quick overview of such optimisation techniques in past 40 years.

        Speaker: Nikola Hardi (University of Novi Sad (RS))
      • Simulating EM Calorimeter Showers with Generative Adversarial Networks 10m

        Simulation is a key component of physics analysis in particle physics and nuclear physics. The most computationally expensive simulation step is the detailed modeling of particle showers inside calorimeters. Full detector simulations are too slow to meet the growing demands resulting from large quantities of data; current fast simulations are not precise enough to serve the entire physics program. Therefore, we introduce CaloGAN, a new fast simulation based on generative adversarial neural networks (GANs). We apply the CaloGAN to model electromagnetic showers in a longitudinally segmented calorimeter. This represents a significant stepping stone toward a full neural network-based detector simulation that could save significant computing time and enable many analyses now and in the future. In particular, the CaloGAN achieves speedup factors comparable to or better than existing fast simulation techniques on CPU (100×-1000×) and even faster on GPU (up to ∼105×)) and has the capability of faithfully reproducing many aspects of key shower shape variables for a variety of particle types.

        Speaker: Michela Paganini (Yale University (US))
      • Interconnection networks simulations for computing systems dedicated to scientific applications 10m

        The increasing complexity of high performance computing system imposes careful network design and optimization. To optimize all the various parameters of an interconnection network we have to use highly accurate software simulations. In this presentation we will discuss a particular implementation of an interconnection network simulation and we will compare the performances of different network topologies and routing algorithms.

        Speaker: Flavio Pisani (Universita e INFN, Bologna (IT))
      • Early days of computer games 10m

        In this presentation, we are going to take a short trip down the memory lane and discuss the origins of computer games history.

        Speaker: Tea Marasovic
    • 17:30 18:00
      Closing session 30m
    • 18:00 20:00
      Free time 2h
    • 20:00 22:00
      Outside Closing Dinner 2h Zrno Soli

      Zrno Soli

      http://www.dobri-restorani.hr/restoran/zrno-soli/

      Zrno Soli

    • 09:00 14:00
      Departure 5h
Your browser is out of date!

Update your browser to view this website correctly. Update my browser now

×