Cookies

We use cookies to ensure that we give you the best experience on our website. You can change your cookie settings at any time. Otherwise, we'll assume you're OK to continue.

School of Engineering and Computing Sciences (ECS)

Profile

Dr Pedro Gonnet

Personal web page

Telephone: +44 (0) 191 33 42519
Room number: E113

(email at pedro.gonnet@durham.ac.uk)

Bibliometric Information

Biography

Pedro Gonnet is a Lecturer in Computer Science in the School of Engineering and Computing Sciences at Durham University. He recieved a Diploma from the Department of Computer Science at the ETH Zürich in 2003 and later a PhD in Computer Science, under the supervision of Prof. Walter Gander, in 2009.

Following his PhD, he worked as a Post-Doctoral researcher in the group of Prof. Jörg Stelling at the ETH Zürich before moving to Oxford University, to work with Prof. L. N. Trefethen as a member of the Chebfun development team with funding from the Swiss National Science Foundation.

Research

My research, in a nutshell, focuses on new algorithms for scientific computing on massively parallel shared-memory architectures.

In the past few years, we have witnessed an important paradigm shift in computer architecture: Although Moore's Law is still in effect, instead of getting faster, computers are getting more parallel. Most methods and algorithms in the computational sciences, however, were designed for single-processor machines or clusters thereof and do not readily lend themselves for shared-memory multi/many-core parallelization, SIMD-vectorization or GPU computing. In order to exploit this new paradigm, algorithms and methods designed specifically for these new architectures are urgently needed.

My research is centered around generalized approaches to shared-memory parallelization which avoid concurrency implicitly. More specifically, this is applied to Molecular Dynamics and Smoothed-Particle Hydrodynamics (SPH) simulations on multi-core and GPU architectures, and Numerical Linear Algebra on GPUs.
For more information, please visit the Desktop-HPC Group page here.

Software

  • mdcore: A platform-independent, Open-Source library for Molecular Dynamics simulations on shared-memory and GPU architectures.
  • odeSD: Second-Derivative ODE Integrator with parameter sensitivity analysis.
  • quadcc: Robust, doubly-adaptive quadrature routine using Clenshaw-Curtis rules with an improved error estimate.
  • Chebfun: I am a member of the Chebfun Developer Team.

Research Groups

School of Engineering and Computing Sciences

  • Innovative Computing Research Group

Department of Computer Science

  • Innovative Computing

Research Interests

  • Algorithms for Systems Biology
  • GPU Computing
  • High-Performance Computing
  • Hybrid Shared/Distributed-Memory Parallelism
  • Numerical Algorithms for Interpolation and Approximation
  • Particle-Based Simulations

Selected Publications

Books: sections

Conference papers

Journal papers: academic

Show all publications