Home / ARM / Advanced processing for ADAS

Advanced processing for ADAS

Like so many things in life, cars are getting smarter. And all those smarts are requiring more and more computing power. Even not-so-smart cars run on embedded systems and complex electronics. As David Blaza pointed out in a recent post on Embedded, the 2011 Chevy Volt had 10 million lines of code in it.

For really smart cars, ARM has predicted that, by 2024, vehicles deploying Advanced Driver Assistance Systems (ADAS) will need a minimum of 100x more computing performance than 2016 models.

“Today, premium cars have more than 100 processors on board utilizing tens of millions of lines of code. To meet future ADAS demands, ARM expects processor performance compared to 2016 vehicles to increase 20x by 2018, 40-50x by 2020 and 100x by 2024. Meeting this ambition will require deeper functional safety support and higher performance, energy-efficient SoCs.” (Source: ARM)

To do their part, “ARM is licensing functional safety support across its Cortex-A, Cortex-R and Cortex-M processor families.”

As cars get smarter – more and more reliant on electronics and less and less reliant on the driver – functional safety becomes more and more important. Yes, a lot of what’s going into smart cars is on the entertainment-side, but that’s relatively minor compared to ADAS. It’s one thing when your music system fails an you can’t access your playlist. It’s quite another when the actual driving is systems-dependent. Malfunctioning systems are no longer an annoyance; they’re a clear danger.

Our partner, TI is one of the companies jumping on board, and will be licensing the latest ARM Cortex-A72 processor. (We don’t – at least for now – play in this space. But in case you’re wondering, Critical Link does have some Cortex-based SOMs. The MitySOM-335x, which is Cortex-A8, and the MitySOM-5CSX, which is Cortex-A9.)

Anyway, that 100x increase in compute performance required in less than 10 years caught my eye.