OpenMPCon 2015 Program

OpenMP 4.1 vs OpenACC 2.0

James Beyer, NVIDIA.

Abstract: As both an OpenMP and OpenACC insider I will present my opinion of the current status of these two directive sets for Programming “accelerators”. Insights into why some decisions were made in initial releases and why … read full abstract

Download Slides

Inside Intel Compilers: Effective OpenMP SIMD Vectorization for Intel Xeon and Xeon Phi Architectures

Xinmin Tian, Intel Corp.

The relentless pace of Moore’s Law will lead to modern multi-core processors, coprocessors and GPU designs with extensive on-die integration of SIMD execution units on CPU and GPU cores to achieve better performance and power efficiency. To make efficient … read full abstract

View Presentation

Enabling Application Portability across HPC Platforms using Open Standards: User-oriented goals for OpenMP

Alice Koniges, Lawrence Berkeley National Lab

To be presented on Alice’s behalf by Tim Mattson.

Abstract: Portability plus performance are key requirements for large-scale scientific simulations on the path to exascale. Users of the high-end computing facilities such as the National Energy Research Scientific Computing Center (NERSC) and the Oak Ridge Leadership Computing Facility (OLCF) are demanding portable standards … read full abstract

Co-authors: Fernanda Foertter (Oak Ridge National Laboratory), Yun He and Richard Gerber (Lawrence Berkeley National Lab), Tim Mattson, Intel

Download Slides

Porting OpenACC 2.0 to OpenMP 4.0: Key Similarities and Differences

Oscar Hernandez, Oak Ridge National Laboratory

The importance of computational accelerator technology for the future of high performance computing is widely acknowledged.  Programmability, performance portability and developer productivity for these systems is a topic of growing importance … read full abstract

Co-authors: Wei Ding, Wayne Joubert, Markus Eisenbach, Christos Kartsaklis, David E Bernholdt (Oak Ridge National Laboratory)

Download Slides  – View Presentation

OpenMP Tips, Tricks and Gotchas

Mark Bull, EPCC, University of Edinburgh

This talk will present a series of practical hints for OpenMP programmers, collected from many years experience of teaching OpenMP, and from answering questions on the OpenMP Forum. I will describe some common pitfalls, and … read full abstract

Download Slides

GCC support to compile OpenMP 4 target constructs for Heterogeneous System Architecture (HSA) accelerators

Martin Jambor, SUSE Linux.

The talk summarizes our experience from ongoing development of a GCC branch which takes OpenMP code and compiles it so that it runs on HSA GPGPUs. We outline the architecture of the process with emphasis on the differences between compilation for the host and HSA accelerators … read full abstract

Download Slides – View Presentation

“OpenMP Does Not Scale”

Ruud van der Pas, Oracle Corporation

Unfortunately it is a very widespread myth that OpenMP Does Not Scale – a myth we intend to dispel in this talk.Every parallel system has its strengths and weaknesses. This is true for clustered systems, but also for shared memory parallel computers.  While nobody in their right mind would consider … read full abstract

 Download Slides  – View Presentation

OpenMP 4.0 vs. OpenCL: Performance comparison

Sergey Vinogradov, Intel

There are many discussions about whether different programming models can achieve high performance and how easy that is for the programmer. For HPC applications, where performance is critical, this question is especially interesting in the context of OpenMP 4.0 and OpenCL which offer different constructs for describing data parallelism … read full abstract

 Download Slides  – View Presentation

Co-authors: Simon McIntosh-Smith, Dan Curran (University of Bristol), Julia Fedorova, James Cownie(Intel)

Advanced OpenMP Tasking

Michael Klemm, Intel Corp. Christian Terboven, RWTH Aachen University

Download Slides

Using OpenMP at NERSC

Helen He, Lawrence Berkeley National Lab.

This presentation will describe how OpenMP is used at NERSC. NERSC is the primary supercomputing facility for Office of Science in the US Depart of Energy (DOE). Our next production system will be an Intel Xeon Phi Knights Landing (KNL) system, with 60+ cores per node and 4 hardware threads per core. The recommended programming model is hybrid MPI/OpenMP, which also … read full abstract

Co-authors: Alice Koniges, Richard Gerber, Katie Antypas (Lawrence Berkeley National Lab)

Download Slides – View Presentation

SIMD Programming with OpenMP

Georg Zitzelsberger, Intel Corp.

If you think OpenMP is merely about threading then you might be interested in the latest features of OpenMP 4.x that exploit the SIMD capabilities of modern processors.   Since processors tend to spend more die space for SIMD, growing with every new generation, the so-called “vectorization” becomes more important … read full abstract

Download Slides

Where Did My Performance Go?

Mark Bull, EPCC, University of Edinburgh

Understanding how and why OpenMP programs lose performance is difficult! In this talk I will attempt to enumerate all the possible ways that OpenMP programs can deliver less than ideal speedup, divided into six main categories: lack of parallelism, load imbalance, synchronisation … read full abstract

Download Slides  – View Presentation

OpenMP 4.x Accelerator Model

Eric Stotzer, Texas Instruments and James Beyer, NVIDIA

OpenMP is the dominant programming model for shared-memory parallelism in C, C++ and Fortran due to its easy-to-use directive-based style, portability and broad support by compiler vendors. Compute-intensive application regions are increasingly being accelerated using specialized heterogeneous devices and a programming model with similar characteristics is needed here. This tutorial will focus on the OpenMP 4.0 accelerator model that provides such a programming model. read full abstract

OpenMP analysis in Intel VTune Amplifier XE – talking to a user about performance in the language a program was written in

Sergey Vinogradov, Intel.

OpenMP directives remain one of the most popular ways to express shared memory multi-threaded parallelism. Fork-join work sharing constructs are simple to use and permit the incremental introduction of data parallelism. As a result the OpenMP language is widely used in … read full abstract

Download Slides

A Success Case using Parallware: The NAS Parallel Benchmark EP

Manuel Arenaz, Appentra Solutions & University of A Coruña

The manual parallelization of existing code is usually a tedious and error-prone task, specially in the case of large projects. Parallware is the first commercial OpenMP-enabling source-to-source compiler that automatically adds OpenMP capabilities in scientific programs. The compiler automatically discovers the parallelism available in sequential codes … read full abstract

Co-authors: Jacobo Lobeiras (Appentra Solutions)

Download Slides