High-performance computing puts Ferrari in pole position in Formula One

Technological advances are making all the difference in Formula One

On a sunny mid-week day in October, the racetrack at the Ferrari headquarters at Maranello in Italy is empty save for the occasional hum of a familiar red sports car being put through its paces.

Today, the real activity is away from the track. The Formula One team is making its way to China for the next race, while in Italy, the firm is announcing a technical collaboration with Microsoft in high-performance computing (HPC).

This is strategic to Ferrari because of how it uses technology in designing and building cars. “Our business in IT in Formula One is following the lifecycle of the car," explains Piergiorgio Grossi, head of IT at Ferrari Gestione Sportiva.

Ferrari Formula One cars have a relatively short lifespan of 18 months, from initial concept and testing through to the final lap of the race season. “The car is not built at the beginning up front. The car evolves during the season," says Grossi.

Whereas before, cars had to be physically assembled to be tested, now models can be simulated first on supercomputers.

When each race car is built, it is fitted throughout with sensors and the large volumes of data from test drives are fed back into supercomputers for analysis.

According to Grossi, HPC comprises some 20pc of Ferrari’s total IT budget. That figure is a closely guarded secret.

Such is the importance of testing and modeling at Ferrari – and the staggering volume of data that needs to be calculated – that its HPC servers run 24 hours a day, seven days a week.

Ferrari’s IT team writes software to mimic the effects of wind drag around the chassis, the aerodynamic balance of the car and to test changes in its centre of gravity. These criteria change depending on the prevailing conditions at each Formula One racetrack and the team must devise software models for every environment.

“We help the team with specific software to lead and to win. We try to empower the engineers at the track to make the right choices," says Grossi.

Ferrari also uses high-performance systems from other suppliers, notably some Linux-based technology, but Grossi points out that much of the software Ferrari uses in other parts of the business comes from Microsoft.

“We run the company on Windows so our users and our systems engineers are very used to Windows, and a Windows service like HPC Server can be easily plugged into this infrastructure. It’s easier for us to manage – it’s more compatible with the rest of our ecosystem."

Antonio Cisternino, assistant professor at the University of Pisa, who has been consulting with Ferrari on the HPC project, strongly endorses Microsoft’s HPC platform, saying it has many advantages over open source tools such as Linux.

Ferrari is one of the first companies to test Windows HPC Server 2008, but Microsoft doesn’t believe its HPC efforts will be restricted to prestigious customers with specific needs such as the Formula One team.

“Windows HPC Server has applications in a variety of different industries, as well as Government," says Vince Mendillo, director of HPC with Microsoft.

Microsoft’s move into HPC marks another step beyond the company’s traditional stronghold of desktop software applications and operating systems. Over the past decade or more, it has expanded its range of server and applications products for companies with enterprise-level computing needs.

Chris Phillips, Microsoft’s general manager of Windows Server Solutions, acknowledges as much. “When I joined in 1994, Microsoft dreamed of trying to crack the data centre. The idea we would be running a Fortune 500 company’s financial systems was just a dream. Fourteen years ago, if you asked me if Windows could be powering a Cray [supercomputer], I’d have been fairly skeptical."

Now, Microsoft believes it has a credible offering. For the first time in June, a supercomputer running Windows HPC Server was ranked in the top 25 of the world’s 500 largest supercomputers with the National Centre for Supercomputing Applications.

Phillips says Microsoft’s goal is “about taking features at the highest end of computing and bringing them into the mainstream". Its track record with other technologies suggests it would be unwise to bet against such an outcome.