OCIAM and NA Group

Numerically Intensive Scientific Computing


MSc in Mathematical Modelling & Scientific Computing
Special topic (Week 8) TT
Dr Stef Salvini

Synopsis

This course provides a hands-on introduction to Parallel Numerical computing through both formal lectures and extensive practical sessions. Both distributed memory and shared memory approaches, using MPI and OpenMP, respectively, will be introduced and used. Hardware and algorithmic issues will also be covered.

Aims

Numerical High Performance Computing (HPC) is an expanding field, of increasing economic importance, for scientific, engineering as well as commercial/financial applications. Currently, HPC can only be achieved through parallelism. Hence, this course will illustrate the main techniques in parallel computing and some of the issues arising, in the context of HPC for numerical applications:

These will be presented in the context of specific applications, for example, Monte Carlo simulations and Finite Difference (FD) solutions of 2D diffusion equation.

Practical sessions are an essential part of the course and can be carried out either in C or Fortran, according to each individual attendee's preference. Code templates, problem sheets, etc will be provided. These sessions will culminate, towards the end of the week, in parallelizing an application, such as in the examples mentioned above.

DPhil students and others are very welcome to attend the lectures and are encouraged to also participate in the practical sessions.

Reading list


Timetables | Course Material
OCIAM OCIAM OCIAM NAGroup NAGroup NAGroup