GPU

Graphical Processing Unit (GPU) computing uses the computational power of modern graphic adaptors for computing, i.e. simulation, data processing etc, and is sometimes referred to as General-purpose computing on graphics processing Wikipedia entry for GPGPU.

A GPU can contain several hundred (or more) computational cores on which computation can be carried out in parallel. For some computational problems a speed up of the order of 40-80 over serial execution (on one core) can be achieved with GPU cards in the price range of ~£150. Good problems that translate well to parallel execution on a GPU require a fair amount of computation on relatively little data: data access is slow in comparison to computation. Typical examples include Molecular Dynamics simulations.

This is an area with rapid change in hardware and software and libraries.

The figure shows results of a GPU accelerated Monte-Carlo simulation (credit Marijan Beg ).

For queries about this topic, contact Hans Fangohr.

View the calendar of events relating to this topic.

Projects

A Fast Multipole Method for the Bessel potential

Marc Molinari, Simon Cox (Investigators), Neil O'Brien

The fast multipole method (FMM) proposed by Greengard and Rokhlin provides a method by which the O(N-squared) many-body problem can be reduced to O(N) complexity. In this project, a multipole method is developed to calculate the energy of a system of vortices in a high temperature superconductor, where the many-body interactions give rise to rich and complex physics. The method developed here is suitable for systems where the interactions are governed by a Bessel potential rather than the usual logarithmic potentials occurring in gravitational and electrostatic problems. We derive and apply vectorised forms of the Gegenbauer addition formulae in order to achieve the O(N) scaling associated with fast multipole methods.

Centre for Doctoral Training in Next Generation Computational Modelling

Hans Fangohr, Ian Hawke, Peter Horak (Investigators), Susanne Ufermann Fangohr, Thorsten Wittemeier, Kieran Selvon, Alvaro Perez-Diaz, David Lusher, Ashley Setter, Emanuele Zappia, Hossam Ragheb, Ryan Pepper, Stephen Gow, Jan Kamenik, Paul Chambers, Robert Entwistle, Rory Brown, Joshua Greenhalgh, James Harrison, Jonathon Waters, Ioannis Begleris, Craig Rafter

The £10million Centre for Doctoral Training was launched in November 2013 and is jointly funded by EPSRC, the University of Southampton, and its partners.

The NGCM brings together world-class simulation modelling research activities from across the University of Southampton and hosts a 4-year doctoral training programme that is the first of its kind in the UK.

Deep Optimisation

Jamie Caldwell

The project will develop the implementation and application of a new optimisation technique. 'Deep optimisation' combines deep learning techniques in neural networks with distributed optimisation methods to create a dynamically re-scalable optimisation process. This project will develop this technique to better-understand its capabilities and limitations and develop GPU implementations. The protein structure prediction problem will be used as the main test application.

Development of a novel Navier-Stokes solver (HiPSTAR)

Richard Sandberg (Investigator)

Development of a highly efficient Navier-Stokes solver for HPC.

Massively-Parallel Computational Fluid Dynamics

Simon Cox, Stephen Turnock, Alexander Phillips (Investigators), James Hawkes

Computational Fluid Dynamics (CFD) is a numerical method for modelling fluid flows and heat transfer - and is used in many industries. It can be used to model dynamics around aircraft, ships and land vehicles; and also has uses in engine design, architecture, weather forecasting, medicine, computer-generated imagery (CGI) and much more. To harness the full power of CFD, it is necessary to utilise the full power of modern supercomputers. This project aims to improve the scalabilty of existing CFD codes so that more complex problems can be tackled efficiently.

Meshless Methods for Photonic Crystal Modelling

Kamal Djidjeli, Marc Molinari, Simon Cox (Investigators), Neil O'Brien, Elizabeth Hart

We apply meshless methods to the problems of simulating photonic crystals. The meshless methods utilise compactly-supported radial basis functions (CSRBFs) and offer a promising alternative to the conventional plane-wave expansion method for calculating the band structure of photonic crystals.

Multiscale modelling of neutron star oceans

Ian Hawke (Investigator), Alice Harpole

Type I X-ray bursts are explosions which occur on the surface of some
neutron stars. It is believed that the burning begins in a localised spot in the ocean of the
star before spreading across the entire surface. By gaining a better understanding of X-ray
bursts, it is hoped that tighter limits can be determined for other neutron star properties
such as the radius and magnetic field strength.

MXL Project

Mark Taylor, Junfen Shi (Investigators)

‘MXL’ is short for “Enhanced patient safety by computational Modelling from clinically available X-rays to minimise the risk of overload and instability for optimised function and Longevity”. This is an international EU-funded project which the Bioengineering Sciences Research Group at Southampton is involved in. For more information, visit http://www.m-x-l.eu

Pushing the Envelope of Planetary Formation and Evolution Simulations

Peter Bartram

A full understanding of the formation and the early evolution of the Solar System and extrasolar planetary systems ranks among natural science's grand challenges, and at present, even the dominant processes responsible for generating the observed planetary architecture remain elusive.

pyQCD

Matthew Spraggs

A basic Python package to perform coarse lattice QCD simulations on desktop and workstation computers.

Soft x-ray science on a tabletop

Peter Horak, Jeremy Frey, Bill Brocklesby (Investigators), Patrick Anderson, Arthur Degen-Knifton

Complex numerical simulations are being performed to aid experimentalists at Southampton realize the next generation of high brightness tabletop sources of coherent soft x-rays.?

Sustainable domain-specific software generation tools for extremely parallel particle-based simulations

Chris-Kriton Skylaris (Investigator)

A range of particle based methods (PBM) are currently used to simulate materials in chemistry, engineering, physics and biophysics. The 4 types of PBM considered directly in the proposed are molecular dynamics (MD), the ONETEP quantum mechanics-based program, discrete element modelling (DEM), and smoothed particle hydrodynamics (SPH).
The overall research objective is to develop a sustainable tool that will deliver, in the future, cutting edge research applicable to applications ranging from dam engineering to atomistic drug design.

µ-VIS Computed Tomography Centre

Ian Sinclair, Richard Boardman, Dmitry Grinev, Philipp Thurner, Simon Cox, Jeremy Frey, Mark Spearing, Kenji Takeda (Investigators)

A dedicated centre for computed tomography (CT) at Southampton, providing complete support for 3D imaging science, serving Engineering, Biomedical, Environmental and Archaeological Sciences. The centre encompasses five complementary scanning systems supporting resolutions down to 200nm and imaging volumes in excess of one metre: from a matchstick to a tree trunk, from an ant's wing to a gas turbine blade.