08:50 - 09:00 Welcome to PMBS Workshop (Room 283,84,85)

09:00 - 09:30 Algebraic Multigrid on a Dragonfly Network: First Experiences on a Cray XC30 [PDF, Talk Slides]

Hormozd Gahvari1, William Gropp2, Kirk Jordan3, Martin Schulz1 and Ulrike Yang1

1Lawrence Livermore National Laboratory, 2University of Illinois at Urbana-Champaign, 3IBM TJ Watson Research Center

09:30 - 10:00 Performance Evaluation of Scientific Applications on POWER8 [PDF, Talk Slides]

Andrew V. Adinetz1, Paul F. Baumeister1, Hans Bottiger3, Thorsten Hater1, Thilo Maurer3, Dirk Pleiter1, Wolfram Schenck4 and Sebastiano Fabio Schifano2

1Julich Supercomputing Centre, 2Universita di Ferrara, 3IBM Deutschland R&D, 4Institute for Advanced Simulation and JARA

10:00 - 10:30 Coffee

10:30 - 11:00 Roofline Model Toolkit: A Practical Tool for Architectural and Program Analysis [PDF, Talk Slides, Project Website]

Yu Jung Lo1, Samuel Williams2, Brian Straalen2, Terry Ligocki2, Matthew Cordery2, Leonid Oliker2, Mary Hall1 and Nicholas Wright2

1University of Utah, 2Lawrence Berkeley National Laboratory

11:00 - 11:30 Insight into Application Performance Using Application-Dependent Characteristics [PDF, Talk Slides]

Waleed Alkohlani1, Jeanine Cook2 and Nafiul Siddique1

1New Mexico State University, 2Sandia National Laboratories

11:30 - 12:00 Energy-Performance Tradeoffs in Multilevel Checkpoint Strategies [PDF, Talk Slides]

Prasanna Balaprakash1, Leonardo Arturo Bautista Gomez1, Mohamed-Slim Bouguerra1, Stefan Wild1, Franck Cappello1,2 and Paul Hovland1

1Argonne National Laboratory, 2University of Illinois at Urbana-Champaign

12:00 - 12:30 Assessing General-Purpose Algorithms to Cope with Fail-stop and Silent Errors [PDF, Talk Slides]

Anne Benoit1, Aurelien Cavelan1, Yves Robert1,2 and Hongyang Sun1

1Ecole Normale Superieure de Lyon, 2University of Tennessee Knoxville

12:30 - 13:30 Lunch

13:30 - 14:00 SPEC ACCEL - A Standard Application Suite for Measuring Hardware Accelerator Performance [PDF, Talk Slides]

Guido Juckeland1, William Brantley2, Sunita Chandrasekaran3, Barbara Chapman3, Shuai Che2, Mathew Colgrove4, Huiyu Feng5, Alexander Grund1, Robert Henschel6, Wen-Mei Hwu7, Huian Li6, Matthias S. Muller8, Maxim Perminov9, Pavel Shelepugin9, Kevin Skadron10, John Stratton7, Alexey Titov2, Ke Wang10, Matthijs van Waveren11, Brian Whitney12, Sandra Wienke8, Rengan Xu3 and Kalyan Kumaran13

1Technische Universitat Dresden, 2AMD, 3University of Houston, 4NVIDIA, 5SGI, 6Indiana University, 7University of Illinois at Urbana-Champaign, 8RWTH Aachen University, 9Intel, 10University of Virginia, 11Compilaflows, 12Oracle, 13Argonne National Laboratory

14:00 - 14:30 A CUDA implementation of the High Performance Conjugate Gradient benchmark [PDF, Talk Slides]

Everett Phillips and Massimiliano Fatica

NVIDIA

14:30 - 15:00 Modeling Stencil Computations on Modern HPC Architectures [PDF, Talk Slides]

Raul de La Cruz1 and Mauricio Araya-Polo2

1Barcelona Supercomputing Centre, 2Shell International Exploration & Production Inc.

15:00 - 15:30 Coffee

15:30 - 16:00 Performance Modeling of the HPCG Benchmark [PDF, Talk Slides]

Vladimir Marjanovic, Jose Gracia and Colin Glass

University of Stuttgart

16:00 - 16:30 Performance Analysis of a High-level Abstractions-based Hydrocode on Future Computing Systems [PDF, Talk Slides]

Gihan Mudalige1, Istvan Reguly1, Michael Giles1, Andrew Mallinson3, Wayne Gaudin2 and Andrew Herdman2

1University of Oxford, 2UK AWE plc., 3University of Warwick

16:30 - 17:00 On the Performance Prediction of BLAS-based Tensor Contractions [PDF, Talk Slides]

Elmar Peise, Diego Fabregat-Traver and Paolo Bientinesi

RWTH Aachen

17:00 - 17:30 Late Breaking Research Papers

17:00 - 17:20 A Case for Epidemic Fault Detection and Group Membership in HPC Storage Systems [PDF, Talk Slides]

Shane Snyder1, Philip Carns1, Jonathan Jenkins1, Kevin Harms1, Robert Ross1, Misbah Mubarak2 and Christopher Carothers2

1Argonne National Laboratory, 2Rensselaer Polytechnic Institute

17:20 - 17:40 On the Energy Proportionality of Distributed NoSQL Data Stores [PDF, Talk Slides]

Balaji Subramaniam and Wu-chun Feng

Virginia Tech