Research

My research is focused primarily on the utilisation of so-called “many-core” architectures (e.g. AMD and NVIDIA graphics cards, Intel MIC) for the acceleration of high-performance scientific computing workloads.  I am also interested in the development of predictive performance models, both to guide future application optimisations and to examine the performance of codes at massive scale.

A lot of my work to date has focused on a class of application known as “pipelined wavefront codes”, but I have also worked on computational fluid dynamics and molecular dynamics codes.

Publications

2011

• Perks, O.F.J. and Hammond, S.D. and Pennycook, S.J. and Jarvis, S.A. (2011) WMTools – Assessing Parallel Application Memory Utilisation at Scale. In: Proceedings of the 8th European Performance Engineering Workshop (EPEW11), 12-13th October, 2011, Borrowdale, United Kingdom. (BibTeX)
• Wright, S.A. and Hammond, S.D. and Pennycook, S.J. and Jarvis, S.A. (2011) Light-weight Parallel I/O Analysis at Scale. In: Proceedings of the 8th European Performance Workshop (EPEW11), 12-13th October, 2011, Borrowdale, United Kingdom. (BibTeX)
• Pennycook, S.J. and Hammond, S.D. and Mudalige, G.R. and Wright, S.A. and Jarvis, S.A. (2011) On the Acceleration of Wavefront Applications using Distributed Many-Core Architectures. The Computer Journal. ISSN 0010-4620 (In Press) (BibTeX)
• Perks, O.F.J. and Hammond, S.D. and Pennycook, S.J. and Jarvis, S.A. (2011) WMTrace – A Lightweight Memory Allocation Tracker and Analysis Framework. In: Proceedings of the UK Performance Engineering Workshop (UKPEW’11), 7-8th July, 2011, Bradford, United Kingdom. [PDF] (BibTeX)
• Wright, S.A. and Pennycook, S.J. and Hammond, S.D. and Jarvis, S.A. (2011) RIOT – A Parallel Input/Output Tracer. In: Proceedings of the UK Performance Engineering Workshop (UKPEW’11), 7-8 July, 2011, Bradford, United Kingdom. [PDF] (BibTeX)
• Pennycook, S.J. and Hammond, S.D. and Jarvis, S.A. and Mudalige, G.R. (2011) Performance Analysis of a Hybrid MPI/CUDA Implementation of the NAS-LU Benchmark. ACM SIGMETRICS Performance Evaluation Review, 38 (4). ISSN 0163-5999 [PDF](BibTeX)
• Perks, O.F.J. and Hammond, S.D. and Pennycook, S.J. and Jarvis, S.A. (2011) Should We Worry About Memory Loss? ACM SIGMETRICS Performance Evaluation Review, 38 (4). ISSN 0163-5999 [PDF] (BibTeX)

2010

• Perks, O.F.J. and Hammond, S.D. and Pennycook, S.J. and Jarvis, S.A. (2010) Should We Worry About Memory Loss? In: 1st International Workshop on Performance Modeling, Benchmarking and Simulation of High Performance Computing Systems (PMBS 10), held in conjunction with IEEE/ACM Supercomputing 2010 (SC’10), New Orleans, LA, USA. (BibTeX)
• Pennycook, S.J. and Hammond, S.D. and Mudalige, G.R. and Jarvis, S.A. (2010) Performance Analysis of a Hybrid MPI/CUDA Implementation of the NAS-LU Benchmark. In: 1st International Workshop on Performance Modeling, Benchmarking and Simulation of High Performance Computing Systems (PMBS 10), held in conjunction with IEEE/ACM Supercomputing 2010 (SC’10), New Orleans, LA, USA. [PDF] (BibTeX)
• Pennycook, S.J. and Hammond, S.D. and Mudalige, G.R. and Jarvis, S.A. (2010) Experiences with Porting and Modelling Wavefront Algorithms on Many-Core Architectures. In: Daresbury GPU Workshop 2010, Daresbury, UK. [PDF] (BibTeX)
• Pennycook, S.J. and Mudalige, G.R. and Hammond, S.D. and Jarvis, S.A. (2010) Parallelising Wavefront Applications on General-Purpose GPU Devices. In: 26th UK Performance Engineering Workshop (UKPEW10), University of Warwick, Coventry, UK. [PDF](BibTeX)