David Beckingsale is a post doc in the Center for Applied Scientific Computing at Lawrence Livermore National Laboratory. His research interests include performance analysis, performance modelling, code optimization, mini-applications and parallel programming models. He is currently engaged in advanced research and development in the area of performance analysis and modeling of massively parallel simulations, and developing tools and techniques based on machine learning to better understand and predict the performance of large-scale, parallel, data-dependent, and adaptive applications. These techniques ares being used to build lightweight machine learning-based models for dynamic performance tuning of data-dependent applications.
David received his Ph.D. in Computer Science from the University of Warwick, UK in June 2015. His thesis investigated the use of mini-applications for investigating scalability and performance issues on future parallel computer systems, specifically focusing on hydrodynamics with adaptive mesh refinement. As part of his Ph.D. research, David deveoped CleverLeaf, a hydrodynamics mini-application with AMR using the SAMRAI toolkit developed at LLNL. David received his B.Sc. in Computer Science in 2011, also from the University of Warwick.
Conference & Journal Papers
David Beckingsale, Wayne Gaudin, Andy Herdman, and Stephen Jarvis. Resident Block-Structured Adaptive Mesh Refinement on Thousands of Graphics Processing Units. In Proceedings of the 44th International Conference on Parallel Processing, Beijing, China, September 2015.
Dan Curran, Simon McIntosh-Smith, Christian Allen, and David Beckingsale. Developing a Future-Proof CFD Code. In Proceedings of the 27th International Conference on Parallel Computational Fluid Dynamics, Montreal, Canada, May 2015.
David Beckingsale, Oliver Perks, Wayne Gaudin, Andy Herdman, and Stephen Jarvis. Optimisation of Patch Distribution Strategies for AMR Applications. Computer Performance Engineering, 7587:210–223, 2013.
Robert Bird, Steven Wright, David Beckingsale, and Stephen Jarvis. Performance Modelling of Magnetohydrodynamics Codes. Computer Performance Engineering, 7587:197–209, 2013.
Oliver Perks, David Beckingsale, Alan Dawes, Andy Herdman, Cyril Mazauric, and Stephen Jarvis. Analysing the Influence of InfiniBand Choice on OpenMPI Memory Consumption. In Proceedings of the 5th International Conference on High Performance Computing and Simulation, Helsinki, Finland, July 2013.
Oliver Perks, David Beckingsale, Simon Hammond, Iain Miller, Andy Herdman, Ash Vadgama, Abhir Bhalerao, Ligang He, and Stephen Jarvis. Towards Automated Memory Model Generation via Event Tracing. The Computer Journal, 56(2):156–174, February 2013.
Workshop Papers & Technical Reports
Andy Herdman, Wayne Gaudin, Oliver Perks, David Beckingsale, Andy Mallinson, and Stephen Jarvis. Achieving Portability and Performance Through OpenACC. In Proceedings of the 1st Workshop on Accelerator Programming using Directives, New Orleans, LA, November 2014.
David Beckingsale, Wayne Gaudin, Rich Hornung, Brian Gunney, Todd Gamblin, Andy Herdman, and Stephen Jarvis. Parallel Block Structured Adaptive Mesh Refinement on Graphics Processing Units. Technical Report LLNL-TR-664446, Lawrence Livermore National Laboratory, October 2014.
Wayne Gaudin, Andy Mallinson, Oliver Perks, Andy Herdman, David Beckingsale, John Levesque, Michael Boulton, Simon McIntosh-Smith, and Stephen Jarvis. Optimising Hydrodynamics applications for the Cray XC30 with the application tool suite. In Proceedings of the Cray User Group, Lugano, Switzerland, May 2014.
Andy Mallinson, David Beckingsale, Wayne Gaudin, Andy Herdman, and Stephen Jarvis. Towards Portable Performance for Explicit Hydrodynamics Codes. In Proceedings of the 1st International Workshop on OpenCL, Atlanta, GA, May 2013.
Andy Mallinson, David Beckingsale, Wayne Gaudin, Andy Herdman, John Levesque, and Stephen Jarvis. CloverLeaf: Preparing Hydrodynamics Codes for Exascale. In Proceedings of the Cray User Group, Napa, CA, May 2013.
Andy Herdman, Wayne Gaudin, Simon McIntosh-Smith, Michael Boulton, David Beckingsale, Andy Mallinson, and Stephen Jarvis. Accelerating Hydrocodes with OpenACC, OpeCL and CUDA. In Proceedings of the 3rd International Workshop on Performance Modeling, Benchmarking and Simulation, Salt Lake City, UT, November 2012.
David Beckingsale, Oliver Perks, Wayne Gaudin, Andy Herdman, and Stephen Jarvis. Optimisation of Patch Distribution Strategies for AMR Applications. In Proceedings of the UK Performance Engineering Workshop, Edinburgh, Scotland, July 2012.
Oliver Perks, Robert Bird, David Beckingsale, and Stephen Jarvis. Exploiting Spatiotemporal Locality for Fast Call Stack Traversal. In Proceedings of the 2nd Workshop on High-performance Infrastructure for Scalable Tools, Venice, Italy, June 2012.
David Beckingsale. Towards Scalable Adaptive Mesh Refinement on Future Parallel Architectures. PhD thesis, University of Warwick, Coventry, UK, 2015.
Posters & Presentations
Matt Martineau, Simon McIntosh-Smith, Mike Boulton, Wayne Gaudin, and David Beckingsale. A Performance Evaluation of Kokkos & RAJA using the TeaLeaf Mini-App. November 2015.
David Beckingsale, Abhinav Bhatele, Jayaraman Thiagarajan, and Todd Gamblin. Loop-based optimization using supervised learning. In LLNL Computation Postdoc Poster Symposium, September 2015.
David Beckingsale, Wayne Gaudin, Rich Hornung, Brian Gunney, Todd Gamblin, Andy Herdman, and Stephen Jarvis. Scalable Adaptive Mesh Refinement of Thousands of GPUs. In LLNL Student Summer Poster Symposium, August 2014.
David Beckingsale and Robert Bird. Practical Performance Modelling of High-Performance Scientific Codes. In Numerical Analysis and Scientific Computing Seminars, University of Manchester, UK, February 2013.