2018 MultiCore 8 Workshop

Agenda
Agenda Track

September 18, 2018 

8:00 - In-person registration begins

8:30 - Workshop Introduction, Rich Loft (NCAR)

8:45 - Keynote: Reflecting on the goal and baseline for exascale computing in weather and climate modelling, Thomas Schulthess (ETH Zurich and CSCS)

9:45 - Break (15 min)

Session 1: Exascale trends and Models / 1

  • Chair: Mark Govett

10:00 - Purpose-Built High-Performance Computing for Earth System Models, Daniel Duffy

(NASA Center for Climate Simulation)

10:25 - NEPTUNE Development for Next Generation HPC, John Michalakes (UCAR/NRL)

10:50 - A Tale of Three Applications: NCAR Experiences in Performance Portability, Richard D. Loft

(National Center for Atmospheric Research)

11:15 - FPGA Acceleration of the LFRic Weather and Climate Model in the EuroExa Project Using Vivado HLS, Mike Ashworth (University of Manchester)

11:40  Lunch

Session 2: Vendor updates and new technologies / 1

  • Chair: Jed Brown

12:45 - Intel update: Running ESM workloads on Intel architecture, Andrey Ovsyannikov (Intel)

1:10 - GPU Developments in Atmospheric Sciences, Stan Posey (NVIDIA Corp)

1:35 - Arm's HPC Solution, Srinath Vadlamani (Arm, Inc.)

Poster Session.  

  • Chair: John Michalakes

2:00 - Poster Presenter Lightning talks

  • Using Graphical Processing Units for Massively Parallel Computations of Fluid Parcel Trajectories, Kelton Halbert (U. Wisconsin)
  • Walking the tightrope: a shared infrastructure for model physics for application on modern high performance computing systems, Dom Heinzeller (NOAA/CIRES)
  • Memory Management with ScaLAPACK for Eigensolver Algorithms on the K Supercomputer, Daniel Howard (Notre Dame/RIKEN)
  • Recent Developments Targeting Manycore CPU and GPGPU Architectures in PTESc, Richard Tran Mills (Argonne NL)
  • How fast is fast: batched linear algebra on GPUs, Kasia Swirydowicz (NREL)

2:30 Break  

Session 3: Portability, Performance and Software

  • Chair: John Michalakes

2:45 - Improving Scientific Software Quality, Christopher Harrop (Cooperative Institute for Research in Environmental Sciences)

3:10 - On performance and portability for the libCEED finite element library, Jeremy L Thompson (CU Boulder)

3:35 - Performance portability from a single source code, Jon Rood (C2SM)

4:00 - A performance portable implementation of HOMME via the Kokkos programming model., Luca Bertagna (Sandia National Laboratories)

4:25 - Choosing Your Candidate in the Programming Model Primaries, Phil Jones (Los Alamos National Laboratory)

5:00 - Adjourn

 

September 19, 2018

8:30 - Keynote: A Portable Applications Driven Approach to Scalability on Present and Future Exascale Systems, Martin Berzins, University of Utah

9:30 - Panel Discussion: Sustainability and Performance Portability

  • Chair: Jed Brown

10:15 - Break (15 min)

Session 4:  Vendor updates and new technologies / 2

  • Chair: John Dennis

10:30 - You can do that on a Cray? Earth Science work on Cray systems, Ilene Carpenter (Cray Inc.)

10:50 - Effect of Network Hardware Support for MPI Collectives and Tag Matching Offloads Gerardo Cisneros-Stoianowski (Mellanox)

11:10 - OpenMP: Current and Future Directions, Helen He (Lawrence Berkeley National Laboratory)

11:35 - Enabling Performance Portability of the Albany Land Ice solver using Trilinos and Kokko, Jerry Watkins (Sandia National Laboratories)

12:00 - Lunch

Session 5: Exascale trends and Models / 2

  • Chair: Hans Johansen

1:00 - LFRic: Meeting the challenges of scalability and performance portability in Weather and Climate models Chris Maynard (Met Office/University of Reading)

1:25 - Data Assimilation plans at NOAA/NCEP and potential implications for high performance computing, Daryl Kleist (NOAA/NWS)

1:50 - Redefining reproducibility for E3SM on multicore systems, Kate Evans (ORNL)

2:15 - Refactoring the Energy Exascale Earth System Model (E3SM) Super Parameterization for GPUs, Matt Norman (Oak Ridge National Laboratory)

2:40 Break

Session 6: Port, Measure, Repeat - performance studies

  • Chair: Ilene Carpenter

3:00 - Evaluating the performance of NEC Aurora and Cavium Thunder X2 processor on NCAR kernels, John Dennis (National Center for Atmospheric Research)

3:25 - MPAS on GPUs Using OpenACC: Portability, Scalability and Performance, Raghu Raj Prasanna Kumar v(National Center for Atmospheric Research)

3:50 - Multicore Performance for a Nonhydrostatic Atmospheric Dynamical Core with Adaptive Mesh Refinement, Hans Johansen (Lawrence Berkeley Laboratory)

4:15 - Acceleration of WRF on the GPU, Daniel S Abdi (TempoQuest)

4:40 - Performance Optimization Techniques for Accelerating WRF Physics Codes on Micro-architectures, Timbwaoga Aime Judicael Ouermi (University of Utah)

5:05 - Discussion and suggestions for next workshop

5:15 - Adjourn