The main rule changes for 2017 are an increase in tyre diameter and changes to body shape to increase downforce. F1 management expects these changes to knock between three and five seconds off each lap.
As ever to a large extent the races will be won or lost away from the track – in development and modelling in the data centre before the season starts and in between races. But those data centres are so important that F1 rules now cover them just like they do the actual racing cars.
One of the key skills of a great racing driver used to be their ability to sense what was going on with the car on the track and clearly communicate that information to the engineers and team management.
Of course there is still a role for the driver but the amount of data available from car sensors and data logging systems means the driver is no longer the main sensor in the car.
Today each engineering group – such as the engine group – will use their own bespoke analysis tools at track side. Back at base there is another data centre also analysing data in real time.
The results of this analysis will lead to changes in the car for almost every race of the 20 race series – continuous development cycles just like software.
But equally important is the role of high performance computing in testing and developing cars before the season starts – something the latest rules also recognise.
F1 rules cover not just every aspect of the cars but even the chipsets used in the teams’ data centres.
The avowed aim is to level the playing field for small and large teams alike. In recent years this has meant restrictions on the amount of wind tunnel modelling and data analysis. Computing, or more precisely Computational Fluid Dynamics, used to be the way poorer teams competed against the richer teams who could afford the expense of wind tunnelling.
But now both wind tunnelling and CFD are restricted by F1 rules and all teams must follow a strict timetable which includes a two week shutdown in the summer.
Teams can now use 25 hours of wind tunnel time or 25 teraflops of compute time – use more of one and you have to use less of the other.
So the battle now is how to do more with less.
Team Sauber chose HPE to help it get the most out of its technology infrastructure. HPE moved the team away from one platform onto a hybrid infrastructure to maximise efficiency and cost savings.
The Sauber team partnership with HPE allowed them to more than double throughput within the teraflop limits. They used Moonshot architecture to make this possible. It also allows instant replacement of servers in case of problems.
The flexibility of the Moonshot system also allows Sauber to increase computing power for short bursts when required – such as for intensive testing of new components.
Just like running an enterprise data centre the key is not raw power but the best utilisation of limited resources. F1 teams don’t get yearly upgrades according to Moore’s Law either – the restrictions have actually gone down since they were first introduced.
Instead they must use the most intelligent management tools to get the very best utilisation of the limited power they have – just like the cars on the track.