The company, which currently lies second in this year's F1 Constructors' Championship, has added "several hundred" HP ProLiant Intel-based servers to its Oxfordshire headquarters.

"Last year showed us that our chassis was a model of reliability but that there's still room for improvement, particularly on the aerodynamic front," said Patrick Head, technical director at Williams.

A spokesman for Williams F1 added that the main driver behind the decision to expand the team's supercomputing resource was the need to reduce the time taken to perform a complete analysis for a given size of model.

"This shortens our overall time to produce a design iteration, which means we can bring performance to our car more quickly."

Williams F1 is now studying the scalability of the Linux cluster and other solutions in order to understand how it can increase model sizes while continuing to get solutions to run overnight, "so that we can get more precision in our analysis without sacrificing any time", the spokesman said.

Tim Bush, engineering manager for HP EMEA, who was responsible for the installation on HP's side, added: "There was genuine surprise at the performance and its impact on the design cycle from the Williams F1 personnel."

Linux is popular for exploiting off-the-shelf applications that require heavy compute capability, while extra performance had come through using a very high-bandwidth, low-latency processor interconnection, explained Bush.

The result, according to Williams F1, has been a threefold enhancement of its simulation capabilities through more detailed computational fluid dynamics simulations.

This halved design, development and testing time has also provided more capacity to experiment with new car design concepts.

The multi-rack Intel-Xeon processor-based system, which is controlled through a head node HP Integrity (Itanium 64-bit) server, was delivered preconfigured in May.