In the old days, teams with more money could just write a big check to spend on bigger and better development, but now that’s not the case.
The arrival of a cost cap in Formula 1 means teams are strictly restricted in how much they can spend. For the main manufacturers, staying within the current limit of 140 million dollars (about 129 million euros) has become a real battle.
But, in addition to the budget cap, another key change has been the introduction of restrictions on aerodynamic testing .
The number of hours that teams can use the wind tunnels and the CFD work that they can do has been limited, to try to ensure that everyone has a similar opportunity.
In addition, there is a system of disadvantages based on a sliding scale, in which the more successful you are in the world championship, the more you restrict what you can do in the wind tunnel or with computer simulations.
The combined impact of all these rules should give smaller teams who haven’t been able to spend as much money in recent years a chance, but they’ve also prompted the need for big teams to look at a completely different approach.
Now it’s no longer about spending money on a problem to try to fix it; but you have to look for the highest possible profitability.
From a fan perspective, this impact is most noticeable in the way teams are limiting updates to their cars this year. Everyone is being more strategic about how and when they introduce advancements, rather than adding as many new things to the car as they can.
However, digging deeper into the areas where teams have made some changes, in an attempt to increase performance in the context of spending constraints and aero testing restrictions, it is fascinating to see how it has come to cause even a deep analysis of the computer chips by the teams.
A report published today has shown that a change to AMD’s computer processors at Mercedes in early 2020 has since helped deliver an impressive 20% performance improvement over the CFD that was used to model and test the aerodynamic flow of its F1 car.
The result was that the team was able to cut their CFD workload in half, and it was a much larger gain than the standard 1% or 2% improvements typically seen with every chip evolution.
This was the result of a switch to AMD’s second generation of EPYC processors, which are seen as offering a better balance of price and performance than the systems they had used for the previous three and a half years.
The move to the new processors came as a result of both the looming cost cap, and a change for 2021 in the way the FIA limited what teams could do for aero testing.
Simon Williams , head of aerodynamic development software at Mercedes, said the inclusion of the teams’ CFD in the cost cap meant some big decisions had to be made. That is why the second generation of EPYC processors was chosen.
“We had new regulations coming in and we were refreshing our systems,” he said. “Performance was the key to decision making.”
“We looked at AMD and the competitors. The CFD solution was the critical factor in setting the benchmarks. We had to get it right, because we’re going to be on this hardware for three years.”
“The other element was that we had the hardware on premises, and if it took up the entire data center, it wasn’t going to be an option. EPYC was the best candidate to deliver the performance and space savings we needed.”
Williams claims the push was partly motivated by the way the FIA opened up a pathway for teams to improve their CFD from 2021.
“Before, wind tunnel hours and computational fluid dynamics were regulated by a common number, so you could choose between wind tunnel or CFD,” he explained.
“The FIA has decoupled these aspects for 2021. The next thing they did was provide an increase in the amount of computing that we could have.
“The third thing was to prorate it based on where your performance is in the championship, so there’s a sliding scale that biases the ability you have to develop aerodynamics.
“We try to get the most out of that, as well as the raw calculation. There is also a regulation of how many geometries we can run in a given period, which is usually eight weeks. We try to maximize everything we can do in that period to get the most out of it. get the most out of our CFD”.
With around 1,800 new geometry simulations allowed per eight-week period in 2021, Mercedes needed to know that its processors were getting the most out of each run.
“It’s about trying to maximize the work that the CFD solution can do per clock cycle,” he added.
Williams says the 20% increase in performance has been impressive, but the absence of downtime has also been an important element: “The new system allows us to focus our efforts on aerodynamic performance.”
“Reliability has also been very good. If we lose even a couple of hours, it hurts us, so it’s important that these systems are robust and reliable. This is something that EPYC has achieved,” he said.
“The time measurements are very short, from the initial idea, through the CFD, to the tests in the tunnel and then in the car. It could be measured in weeks. The fact that we have been able to offer constant work to throughout the year has been key”.