PMAW: The Programming Models and Algorithms Workshop

Held in Conjunction with IPDPS'18

Workshop Overview   Workshop Program   Workshop Scope   Workshop Organization   Important Dates   Submission   Proceedings   More Information

Workshop Overview

Programming models are widespread, and have gained vast popularity due to a variety of optimization mechanisms and abstractions, their transparency and other runtime features. Examples of these features are automatic accelerator offloading, directive-based compiler optimizations and parallelization, integration into high-level programming languages, locality enhancements, work stealing and pushing, overlapping communication and computation, providing a wide range of dynamic scheduling techniques, scaling and enabling the execution of large scale applications for distributed computing, leveraging and exposing interfaces that enable a fine grained control of underlying computing and network hardware, among many other features. Despite this long list of major advancements, there still exists a wide gap between the algorithmic design, specification and implementation, and these programming models.

To exacerbate the problem at hand, it is widely known that to achieve the expected levels of performance and power cap of Exascale and future computing eras, tremendous efforts will have to be invested and will require the convergence of several traditionally disjoint fields. Programming models excel at raising the level of abstraction from a concrete target platform, architecture or even the underlying network. In other words, their primary role is to hide the complexities of some lower level of the hardware and software stack, thereby greatly simplifying and speeding the development process of scientific software.

Complicating even more the matter, new technologies such as upcoming programming models and parallel runtimes are not fully exploited. To a large extent, this is caused by the unfamiliarity of domain experts and computational scientists with new runtimes and software developments, or conversely, because of the lack of knowledge of the runtime or compiler expert with the fundamental properties of a computational algorithm. Moreover, simple mechanical porting of traditional and classic algorithms, many of which were devised 40 or more years ago, considered much more generic and abstract models such as the P-RAM model, the Bulk-Synchronous Parallel (BSP) model, LogP, or even the traditional von Neumann model. More precisely, this process has been target agnostic, and did not envision any specific hardware or runtime. This approach often results in suboptimal performance, loses important information in translation or incorrectly exploits the algorithmic intrinsics.

Workshop Agenda (May 25th, 2018)

  • 9am - 10am: Keynote
    Speaker: Hal Finkel (Argonne National Laboratory)
    Title: Parallel Programming Models and Compiler Optimization
    Abstract: Parallel programming models tend to be designed in one of two categories: Either they're completely compiler driven (e.g., OpenMP) or they're library solutions opaque to the compiler. In this talk, we'll explore potential middle ground by looking at recent work on optimizing parallelism constructs in LLVM, recent advances in parallel programming models in C++, and how the two might fit together. In the future, parallel programming models that are easy to use, amenable to abstraction, and understood by the compiler's optimizer should be possible to create. As we approach the exascale era and hardware diversifies, this might be the only high-productivity option.
  • 10am - 10:30am : Break
  • 10:30 - 11am: Talk #1
    Title: A Software Abstraction Method for Efficient 2D Grid Computations in Heterogeneous HPC Environments
    Authors: Ryan Marshall, Sheikh Ghafoor and Michael Rogers
  • 11:00 - 11:30am: Talk #2
    Title: Space Consistency for Distributed Memory Systems
    Authors: Khaled Ibrahim
  • 11:30am - 12: Talk #3
    Title: Loop Scheduling in OpenMP
    Authors: Vivek Kale
  • 12 - 12:30: Talk #4
    Title: Adaptation of sparse-matrix multiplication primitives for efficient triangle counting in graphs
    Authors: Vineeth Thumma, Aravind Sukumaran-Rajam, and P. Sadayappan
  • 12:30 - 1:00pm: Talk #5
    Title: High-performance tensor contraction on GPUs
    Authors: Jinsung Kim, Aravind Sukumaran-Rajam, and P. Sadayappan

Workshop Scope and Topics of Interest

The purpose of this workshop is to bring together experts of different areas to bridge the gap between programming models and algorithms in general. The ultimate goal is to foster the interdisciplinary relationships among domain scientists, and compiler, language and runtime (CLR) experts in order to re-design and rethink the next generation of algorithms that will be used during the forthcoming decades. Therefore, the intended audience and participants of the proposed workshop involves experts, practitioners, students, professors and researchers of the mentioned fields, of the broader computer science, applied mathematics and computational science fields. In the shorter term, the objective of this workshop is to enable and retarget core algorithmic building blocks that leverage cutting-edge data-flow, task and data-parallel runtimes, and gradually promote the absorption of undeniable performance, energy and other execution properties that an algorithm should embody in order to fully exploit future hardware and software computing infrastructure.

We welcome ongoing and preliminary work, experience reports showing a negative result or a gap between the two communities. In general, PMAW topics lie between fields of domain scientists and CLR experts. For instance, domain experts who are attempting to map their application to some specific platform or runtime, as well as CLR experts wishing to identify and learn about strong and impactful applications and benchmarks suitable for some new software technology.

All papers must be original and not simultaneously submitted to another journal or conference. The following is a non-exhaustive list of topics of interest:

  • Implementation of algorithms using data-flow models and runtimes
  • Domain specific languages, compilers and tools for data-flow, data and task parallelism
  • New parallel runtime features that facilitate algorithmic description
  • Rethinking of classical algorithms that incorporate parallelism, energy and other performance properties
  • Communication minimization schemes
  • Network aware computing
  • Locality and Communication abstractions for algorithmic specification
  • Orchestrating runtimes and heterogeneous runtime scheduling
  • Experiences in porting or implementing scientific frameworks over one or more programming models
  • A negative result demonstrating general poor support for some application or class of computation and potential solutions or workaround
  • Application and runtime demos

Organizers and Committee


  • Martin Kong (Assistant Computational Scientist, Brookhaven National Laboratory, USA)
  • Zoran Budimlic (Senior Research Scientist, Rice University, USA)
Program Committee:
  • Albert Cohen (Professor, Inria, France)
  • Barbara Chapman (Professor, Stony Brook University, USA)
  • Jun Shirako (Research Scientist, Georgia Institute of Technology, USA)
  • Meifeng Lin (Associate Scientist, Brookhaven National Laboratory, USA)
  • Kamer Kaya (Assistant Professor, Sabanci University, Turkey)
  • Erdem Sariyuce (Assistant Professor, SUNY University at Buffalo, USA)
  • Antoniu Pop (Assistant Professor, University of Manchester, UK)
  • Didem Unat (Assistant Professor, Koc ̧ University, Turkey)
  • Oscar Hernandez (Oak Ridge National Laboratory, USA)
  • George Slota (Assistant Professor, Rensselaer Polytechnic Institute, USA)
  • Mehmet Deveci (Post-doctoral Researcher, Sandia National Laboratories, USA)
  • Florina Ciorba (Assistant Professor, University of Basel, Switzerland)
  • Ye Luo (Researcher, Argonne National Laboratory, USA)

Important Dates

The main dates for this workshop will be the following:

  • Paper submission start: November 6, 2017
  • Abstract due: February 9, 2018
  • Abstract due: February 19, 2018
  • Paper submission due : February 16, 2018
  • Paper submission due : February 28, 2018
  • Notification of acceptance : April 2, 2018
  • Camera ready papers: April 23, 2018
  • Workshop date: May 25, 2018


  • Please submit your paper here:
  • Papers must use the ACM standard LATEX or Word templates.
  • As this workshop encourages works-in-progress, papers should be 5-7 pages in length, including references and appendices


PMAW will not have formal proceedings. However, accepted papers will be posted online so as to allow authors to re-submit their work to other venues and using the feedback provided in the workshop.

More Information

For more information regarding this workshop, please contact Martin Kong