Paul leads the work to optimise Met Office codes for High-Performance Computers.
Paul is the manager of a small group working on understanding and improving the computational performance of the Unified Model. He is currently working on understanding and improving the performance of the Met Office's main model configurations on our IBM supercomputer.
One of Paul's main areas of focus is that of scalability of the Unified Model when using more CPUs. Current supercomputer designs add performance by utilising additional CPUs and getting good performance from these systems requires numerical algorithms that scale well. Paul is therefore trying to identify and remove bottlenecks in the scaling of the Unified Model. Paul is also heavily involved in the Met Office's supercomputer procurement activities and is currently working on preparations for the mid-life upgrade of our service with IBM.
Paul has worked in Numerical Modelling since joining the Met Office in 1998. For most of that time he has worked on the technical infrastructure and releases of the Unified Model. Paul moved to his current post in 2009.
Prior to joining the Met Office, Paul obtained a BSc in Mathematics from the University of Sussex and a PhD in Numerical Analysis from the University of Bristol. He followed this up with a Post-Doctoral position at the Universityof Leeds where he worked on parallel adaptive unstructured tetrahedral meshes for unsteady fluid flow problems.