NUWEST: NNSA-University Workshop on Exascale Simulation Technologies

Thursday January 18, 2024 in Albuquerque at the Crowne Plaza

The NNSA-University Workshop on Exascale Simulation Technologies (NUWEST) will be held on Thursday January 18, 2024 in Albuquerque at the Crowne Plaza. The overarching goal of NUWEST is to share ideas on technologies for facilitating exascale predictive science, by showcasing available technologies, identifying challenges, and initiating further collaborations with NNSA laboratory efforts.

The workshop will highlight technologies from PSAAPIII centers. The schedule is below.

The workshop is designed to encourage hands-on demonstration of technologies used in and employed across the centers in their predictive simulations, with a goal of encouraging wider adoption of the key concepts in the technology itself or direct use of the underlying software.

Lunch is provided, along with a breakfast and coffee throughout the day. If you have questions, then please contact Luke Olson (lukeo at illinois.edu) or Courtney McLearin (cmcleari at illinois.edu).

Quicklinks:

Schedule

Time Title/Speaker Room Links
0700 - 0800 Breakfast Garden
0800 – 0815 Introduction
Luke Olson, University of Illinois Urbana-Champaign
Garden nuwest.pdf
0815 – 0900 Keynotes Garden
Success through Community Building – A Kokkos Story
Christian Trott, Sandia National Laboratories
[abstract]

What makes a successful Open Source project – and what does it take to achieve wide spread adoption? This talk will explore these questions based on the experience with Kokkos. Now with users at more than 150 institutions, Kokkos represents one of the success stories of DOE’s recently concluded Exascale Computing Project. But technical excellence alone wasn’t enough – a critical aspect of Kokkos adoption progress was the proactive effort of the Kokkos team to widen the contributor base, build a user community, and develop trust. The talk will give an overview of the community efforts Kokkos has undertaken and how trust was build, as well as provide a view into the future of Kokkos as part of the High Performance Software Foundation.

Getting to adoption: Lessons from MPI and PETSc
Bill Gropp, University of Illinois Urbana-Champaign
[abstract]

Many software projects measure their success by the number of users that adopt that software for their work. Few succeed in getting significant adoption. What do these two successful projects tell us about getting an HPC tool to adoption by the community? This talk will review the history of two different projects: PETSc, a software library designed to support the development of applications to solve PDEs in parallel, and MPI, a specification (not a library) for communicating between processes. The importance of both the design and implementation of the projects as well as the marketing and support are discusses, and lessons for other HPC projects are discussed.

slides (pdf)
0900 – 0950 Conceptual Overviews Garden
CUnumeric and Legion
Charlelie Laurent, Stanford University
overview (pptx)
Parsl - Python based workflow management
Daniel S. Katz, Doug Friedel, University of Illinois Urbana-Champaign
overview (google slide)
Scalable and portable HPC in Python using Parla and PyKokkos
George Biros, University of Texas at Austin
overview (pdf)
Pragmatic performance-portable solids and fluids with Ratel, libCEED, and PETSc
Jed Brown, University of Colorado Boulder
overview (pdf)
0950 - 1000 Break
1000 - 1200 Code-alongs
CUnumeric and Legion
Charlelie Laurent, Stanford University
Arizona github.com/CharlelieLrt/Nuwest_cunumeric_demo
Parsl - Python based workflow management
Doug Friedel, Daniel S. Katz, University of Illinois Urbana-Champaign
Nevada github.com/astro-friedel/Parsl_tutorial
Scalable and portable HPC in Python using Parla and PyKokkos
Milos Gligoric and Will Ruys, University of Texas at Austin
Texas github.com/ut-parla/nuwest
Pragmatic performance-portable solids and fluids with Ratel, libCEED, and PETSc
Jed Brown, University of Colorado Boulder
Colorado github.com/jedbrown/nuwest24
1200 - 1300 Lunch break Garden
1300 – 1350 Conceptual Overviews Garden
Acceleration and Abstraction of Python based Monte Carlo Compute Kernels for Heterogeneous machines via Numba
Joanna Piper Morgan, Oregon State University
overview (pptx)
MIRGE -- A lazy evaluation framework in Python
Andreas Kloeckner, University of Illinois Urbana-Champaign
overview (pdf)
OpenCilk: A Modular and Extensible Software Infrastructure for Fast Task-Parallel Code
Tao Schardl, Massachusetts Institute of Technology
overview (pdf)
MPI Advance - Optimizations and Extensions to MPI
Purushotham V. Bangalore, University of Alabama
overview (pdf)
1350 - 1400 Break
1400 - 1600 Code-alongs
Acceleration and Abstraction of Python based Monte Carlo Compute Kernels for Heterogeneous machines via Numba
Joanna Piper Morgan, Oregon State University
Arizona slides (html)
notebook
github.com/jpmorgan98/nuwest-mcdc-jpmorgan
MIRGE -- A lazy evaluation framework in Python
Andreas Kloeckner, University of Illinois Urbana-Champaign
Nevada github.com/illinois-ceesd/nuwest-mirge
Writing Fast Task-Parallel Code Using OpenCilk
Tao Schardl, Massachusetts Institute of Technology
Texas mxm_demo.pdf
speedcode.org
github.com/OpenCilk/opencilk-project/releases
opencilk.org/doc/users-guide/install
MPI Advance - Optimizations and Extensions to MPI
Purushotham V. Bangalore, University of Alabama
Colorado github.com/mpi-advance/MPIPCL
github.com/mpi-advance/locality_aware
1600 - 1615 Concluding remarks
Luke Olson, University of Illinois Urbana-Champaingn
Garden
1700-1900 (optional) social at Bow and Arrow
Food truck on site; food and drinks are on your own.

Related technologies and tools: