GPU Day 2016

The Future of Many-Core Computing in Science

2-3 06 2016

About the GPU Day

GPU Day 2016 - THE FUTURE OF MANY-CORE COMPUTING IN SCIENCE

The 6th in the conference series organized by the Wigner Research Centre for Physics of the Hungarian Academy of Sciences is dedicated to the field of many-core computing in scientific and industrial applications.

For the earlier events see: 2015, 2014

This years event is sponsored by AdasWorks Ltd.

TOPICS INCLUDE:

  • Current status and near-future of many-core hardware and software
  • Many-core computing in physics and other fields of science
  • Medical applications of parallel technologies
  • Machine Learning, neural networks, feature recognition
  • Image processing, computer vision and reconstruction
  • Industrial applications of many-core computing
  • Many-core computing in education
  • Mobile and embedded parallel computing in science
  • Emerging accelerator platforms

CONFIRMED KEY TALKS:

Vulkan Fast Paths
Lukasz Migas, AMD

Radeon Open Compute stack overview
Dmitry Kozlov, AMD

SYCL  building blocks for C++ libraries
Maria Rovatsou, CODEPLAY

Polgraw-allsky: GPU-accelerated all-sky blind search for periodic gravitational waves
Michał Bejger, Nicolaus Copernicus Astronomical Center, Warsaw

Line Segment Maps: Fast and Deterministic Line Detection on the GPU
Dr-Ing. Gernot Ziegler - Geofront e.U.

Turning software into hardware – Hastlayer
Zoltán Lehóczky, Lombiq Ltd.

Multi everything: parallelism in all shapes and sizes in HoloVizio lightfield display systems
Attila Barsi, Holografika Ltd.

Ring artifact removal method using GPU-based FDK reconstruction for cone beam CT
Dr. Zsolt Balogh - Mediso Ltd.

AdasWorks: artificial intelligence for self-driving cars
Árpád Takács - AdasWorks Ltd.

DEADLINES:

  • Extended abstract submission: 29 April 2016
  • Decision on accepted presentations: 9 May 2016
  • Registration deadline: 29 May 2016
  • Registration is closed

Please send your abstracts and questions to:

Dániel BERÉNYI
{berenyi} . {daniel} {at} {wigner} . {mta} . {hu}

ORGANIZERS:
Dániel BERÉNYI (Wigner RCP)
Máté Ferenc NAGY-EGRI (Wigner RCP)
Gábor Cseh (Wigner RCP)
Gergely Gábor BARNAFÖLDI (Wigner RCP)

Schedule

First Day

  • 9:00 - 9:10
    Opening
  • 9:10 - 9:35
    NIIF HPC for research and education accelerated with GPUs
  • 9:35 - 10:25
    Radeon Open Compute stack overview
  • 10:25 - 10:45
    Coffee Break
  • 10:45 - 11:35
    SYCL building blocks for C++ libraries
  • 11:35 - 12:15
    Turning software into hardware – Hastlayer
  • 12:15 - 12:40
    Parallel evaluation of general vector-arithmetic trees
  • 12:40 - 14:10
    Lunch
  • 14:10 - 14:50
    Polgraw-allsky: GPU-accelerated all-sky blind search for periodic gravitational waves
  • 14:50 - 15:15
    Simulation of Surface Growth on GPU supercomputers
  • 15:15 - 15:40
    A GPU cluster optimized multigrid scheme for computing unsteady incompressible fluid flow
  • 15:40 - 16:05
    HIJING++, OOP concepts and simple parallelization
  • 16:05 - 16:25
    Coffee Break
  • 16:25 - 16:50
    Applications of GPU-accelerated deep neural networks to analyze galaxy spectra
  • 16:50 - 17:15
    Light curve modeling of eclipsing binary stars
  • 17:15 - 17:40
    Fast determination of some orbital elements of eclipsing binary stars on GPU
  • 17:40 - 18:05
    Eclipse Timing Variation analysis of binary stars with CUDA
  • 18:05 - 18:30
    The PYCBC package and its use in the search of the gravitational waves detection and its application on GPUs

Second day

  • 9:00 - 9:50
    Vulkan Fast Paths
  • 9:50 - 10:30
    Multi everything: parallelism in all shapes and sizes in HoloVizio light field display systems
  • 10:30 - 22:50
    Coffee Break
  • 10:50 - 11:30
    AdasWorks: artificial intelligence for self-driving cars
  • 11:30 - 12:10
    Line Segment Maps: Fast and Deterministic Line Detection on the GPU
  • 12:10 - 12:35
    Hierarchical Volumetric Fusion of Depth Images
  • 12:35 - 14:05
    Lunch
  • 14:05 - 14:45
    Ring artifact removal method using GPU-based FDK reconstruction for cone beam CT
  • 14:45 - 15:10
    Photorealistic 3D CT visualization supported by HPC GPU and CPU coprocessors
  • 15:10 - 15:35
    A high-throughput stream-oriented GPU environment for simultaneous processing and visualization of EEG measurements
  • 15:35 - 15:45
    Closing

Speakers

Speakers

Gergely Gábor Barnaföldi

Gergely Gábor BarnaföldiWigner RCP

Zoltán Kiss

Zoltán KissNational Information Infrastructure Development Institute

Zoltán Lehóczky

Zoltán LehóczkyLombiq Technologies Ltd., Budapest, Hungary

Dmitry Kozlov

Dmitry KozlovAMD

Maria Rovatsou

Maria RovatsouCODEPLAY

Zoltán Lehóczky

Zoltán LehóczkyLombiq Ltd.

András Leitereg

András LeiteregEötvös University

Michał Bejger

Michał BejgerNicolaus Copernicus Astronomical Center, Warsaw

Géza Ódor

Géza ÓdorCentre of Energy Research, Institute for Technical Physics and Materials Science

György Tegze

György TegzeWigner Research Centre for Physics, Institute for Solid State Physics and Optics

Szilveszter Harangozó

Szilveszter HarangozóWigner Research Centre for Physics, Institute for Particle and Nuclear Physics

László Dobos

László DobosEötvös University, Department of Physics of Complex Systems

János Sztakovics

János SztakovicsEötvös University

Gábor Marschalkó

Gábor MarschalkóSzeged University, Baja Observatory

Tamás Hajdu

Tamás HajduEötvös University

Balázs Kacskovics

Balázs KacskovicsWigner Research Centre for Physics, Institute for Particle and Nuclear Physics

László Somlai

László SomlaiWigner Research Centre for Physics, Institute for Particle and Nuclear Physics

Lukasz Migas

Lukasz MigasAMD

Attila Barsi

Attila BarsiHolografika Ltd.

Árpád Takács

Árpád TakácsAdasWorks Ltd.

Gernot Ziegler

Gernot ZieglerGeofront e.U.

Milán Magdics

Milán MagdicsBudapest University of Technology and Economics, Department of Control Engineering and Information Technology

Zsolt Balogh

Zsolt BaloghMediso Ltd.

László Kovács

László KovácsUniversity of Debrecen, Faculty of Informatics

Zoltán Juhász

Zoltán JuhászUniversity of Pannonia, Department of Electrical Engineering and Information Systems