2013 Agenda

9/19/2013 Session ChairSpeakerOrganization
8:30 Workshop Introduction and Opening Remarks Rich Loft   NCAR
8:40 Session 1 - Common Features and Challenges of Using Multi-core Systems Chris Kerr   NOAA
  Performance Results from Multi-core Platforms   Nicholas Wright NERSC
  The GFDL FMS: Progress for Climate Science on Multi-core Platforms   Jeff Durachta NOAA
  A comparison of climate applications on accelerated and conventional architectures   Srinath Vadlamani NCAR
  Performance Analysis and Tuning Techniques on GPU and MIC   Youngsung Kim NCAR
10:00 Discussion period 1      
10:30 Morning Break      
10:50 Session 2 - Optimization for Hybrid GPU systems Ilene Carpenter   NREL
  Porting the Community Atmospheric Model - Spectral Element (CAM-SE) to utilize hybrid GPU architectures   Matt Norman ORNL
  Optimization of MPI Message Passing in a Multi-core NWP Dynamical Core running on NVIDIA GPUs   Jacques Middlecoff NOAA
11:30 Discussion period 2      
12:00 Lunch - on your own in NCAR Cafeteria      
1:20 Session 3a - Porting and Optimizing for Intel Xeon and Xeon Phi Will Sawyer   CSCS
  Intel Xeon Phi Coprocessor Application Experiences   Mike Greenfield Intel
  Performance of Cubed-sphere Atmospheric Dynamical Core on Intel Xeon and Xeon Phi architectures   Hoang Vu Nguyen KAUST
2:00 Discussion Period 3a      
2:20 Afternoon break      
2:40 Session 3b - Porting and Optimizing for Intel Xeon and Xeon Phi Mark Govett   NOAA
  Approaches to porting, validating and optimizing NOAA weather models to the Intel Xeon Phi architecture   Jim Rosinski NOAA
  Atmospheric Simulation Based on FPGAs and MICs   Haohuan Fu Tsinghua University
  Discovering Bottlenecks without Pain: Profiling and Optimizing the Intel Xeon Phi with Allinea MAP   Beau Paisley Allinea
3:40 Discussion Period 3b      
4:10 First day summary Chris Kerr   NOAA
4:20 Adjourn      
9/20/2013 (Friday)      
9:00 Session 4: Common Features and Challenges of Using Multi-core Systems Mark Govett   NOAA
  Weather and Climate Modeling on GPU and Xeon Phi Accelerated Systems   Mike Ashworth STFC Daresbury Lab
  Continued Efforts in Adapting the GEOS-5 GCM to CUDA Fortran, OpenACC and Intel Xeon Phi   Matt Thompson NASA
9:40-10:00 Discussion Period 4      
10:00-10:20 Morning Break      
10:20 Session 5: Porting and Optimizing for Intel Xeon and Xeon Phi Bill Putman   NREL
  Code restructuring to improve performance in WRF model physics on Intel Xeon Phi   John Michalakes NREL
  Heterogeneous Multi-Core Debugging   Ed Hinkel Rogue Wave Software
  Explicit vectorization/parallelization through the Intel Compiler   Milind Girkar Intel
  Early Experiences and Progress Incorporating Block Algorithms into GEOS   Kareem Sorathia NASA
11:40 Discussion period 5      
12:20 Lunch - on your own in NCAR Cafeteria      
1:30 Session 6a: Optimizing for GPUs using CUDA and OpenACC Rich Loft    
  Approaches to porting the LMDZ General Circulation Model to CUDA Fortran   Sourav Mukherjee CAS/IIT
  OpenACC Implementation of the GFDL Single Column Model   Eric Dolven Cray
  OpenACC Programming for Weather Codes   Carl Ponder NVIDIA
  NVIDIA Update   Stan Posey NVIDIA
2:30 Discussion 6a      
3:00 Afternoon break      
3:20 Session 6b: Optimizing for GPUs using CUDA and OpenACC Chris Kerr   NOAA
  Using OpenACC compilers to run FIM and NIM on GPUs   Mark Govett NOAA
  Multi-node OpenACC Implementation of the ICON Non-hydrostatic Dynamics Core   Will Sawyer CSCS
4:20 Discussion period 6b      
4:50 Workshop Summary Chris Kerr/Rich Loft    
5:00 Adjourn