SC15 BoF : Paving the way for Performance on Intel® Knights Landing Processors and beyond: Unleashing the Power of Next-Generation Many-Core Processors
Download presentations here:
- Dynamic SIMD Scheduling
- Intel Xeon Phi HPC Users Group
- Estimating the Performance Impact of the HBM on KNL Using DualSocket Nodes
- Experiences in Optimizations of Preconditioned Iterative Solvers for FEM/FVM Applications & Matrix Assembly of FEM using Intel Xeon Phi
- PerfExpert.- Workflow for Vectorization and Parallel Scalability
- Enabling the Quantum Collisions ‘PFARM’ code on Xeon Phi Refactoring MPI communications for Symmetric Mode
- Intel® Parallel Studio XE 2016
- Vectorisation efficiency in a Gadget kernel: dealing with conditionals and data access
- Going parallel with efficiency on the future Knights family
- Hetero Streams: easing the way to task parallelism and platform features
- Uncertainty in Seismic Imaging: Exploring a New Frontier on HPC
- Particle-in-Cell Plasma Simulation on Intel Xeon Phi Coprocessors
- TAU Performance System
- Improving vectorization awareness in HPC
- Strategies for Effective Offloading in GROMACS
- Performance Improvement of Electronic Structure Calculations using MIC programming: A Sparsematrix-coupled Normal Eigenvalue Problem
- Improved Parallelization of Quantum Espresso
- SC15 Event Listing, Wednesday, Nov. 18, 5:30-7p.m
- We had approximately 180 participants
Agenda - for content, see here
Introduction and welcome (5) Richard Gerber, NERSC
Keynote: Going parallel with efficiency on the future Knights family (10) Chris J. Newburn, Intel (CJ)
Tales from the trenches - 5-minute lightning talks (35) CJ Newburn, Intel
- Uncertainty in Seismic Imaging: Exploring a New Frontier on HPC, Alvaro Coutinho, UFRJ Brazil
- Particle-in-Cell Plasma Simulation on Intel Xeon Phi Coprocessors, Sergey Bastrakov et al, Lobachevsky State Univ of Nizhni Novgorod and Institute of Applied Physics of the RAS (presented by Dmitry Durnov, Intel)
- Dynamic SIMD Scheduling, Florian Wende, Zuse Institute Berlin
- Vectorisation efficiency in a Gadget kernel: dealing with conditionals and data access, Luigi Iapichino, Leibniz Supercomputing Centre
- Hetero Streams: easing the way to task parallelism and platform features, Piotr Luszczek, UTK
- Enabling the Quantum Collisions ‘PFARM’ code on Xeon Phi - Refactoring MPI communications for Symmetric Mode, Michael Lysaght, ICHEC
- Experiences in optimizations of preconditioned iterative solvers for FEM/FVM applications and matrix assembly of FEM using Intel Xeon Phi, Kengo Nakajima, Univ Tokyo
- Estimating the performance impact of the HBM on KNL using dual-socket nodes, Zhengji Zhang, NERSC/Berkeley Lab
Discussion on tuning issues (10) Thomas Steinke, Zuse Institute Berlin
Analysis and development tools (20) Richard Gerber, NERSC
- Show don't tell: improving vectorization awareness in HPC, Mark O'Connor, Allinea (3)
- Tau Performance System, Sameer Shende, U Oregon (3)
- PerfExpert - Workflow for Vectorization and Parallel Scalability, Jim Browne and Antonio Gomez-Iglesias, U Texas (3)
- Faster Code... Faster Mike Lee, Intel (3)
- Tools from Rogue Wave for Xeon Phi systems, Martin Bakal, Rogue Wave
Open discussion and wrap up (10) David Martin, Argonne Leadership Computing Facility
- The talks that were submitted but not presented as lightning talks will also be posted shortly, and alluded to during the BoF.
- Lightning talk and tools submissions were made here using this template. We had nearly 20 entries.
- Keeping with tradition, we will feature tales from the trenches as lightning talks, updates on tools and libraries, a broader discussion as a community, and IXPUG highlights
- If you have questions, please contact CJ Newburn, Richard Gerber, Thomas Steinke or Kent Milfed.
Additional talks, online (in order of submission)
- Vectorization of pair interactions in GROMACS, Roland Shultz, John Eblen, Jeremy Smith, U Tenn Knoxville
- Strategies to Optimize Offloading to the Xeon Phi, John Eblen, Roland Schulz and Jeremy Smith, U Tenn Knoxville
- Improved Parallelization of Quantum Espresso, Taylor Barnes, NERSC
- Performance Improvement of Electronic Structure Calculations using MIC programming: A Sparse-matrix-coupled Normal Eigenvalue Problem, Hoon Ryu, National Institute of Supercomputing and Networking, Korea Institute of Science and Technology Information
- Scientific research and application optimization on RSC PetaStream massively-parallel supercomputer with Intel Xeon Phi, Alexander Moskovsky, RSC
Related
- Intel HPC Developer Conference, Nov. 14-15 in Austin, TX before SC15. Please register now! Please note the Sunday 2pm talk on data locality, which grew out of IXPUG Data Preconditioning working group sessions.