Speeding-up the execution of credit risk simulations using desktop grid computing: A case study

Abstract

This paper describes a case study that was undertaken at a leading European Investment bank in which desktop grid computing was used to speed-up the execution of Monte Carlo credit risk simulations. The credit risk simulations were modelled using commercial-off-the-shelf simulation packages (CSPs). The CSPs did not incorporate built-in support for desktop grids, and therefore the authors implemented a middleware for desktop grid computing, called WinGrid, and interfaced it with the CSP. The performance results show that WinGrid can speed-up the execution of CSP-based Monte Carlo simulations. However, since WinGrid was installed on non-dedicated PCs, the speed-up achieved varied according to users’ PC usage. Finally, the paper presents some lessons learnt from this case study. It is expected that this paper will encourage simulation practitioners and CSP vendors to experiment with desktop grid computing technologies with the objective of speeding-up simulation experimentation.