Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

MacRumors

macrumors bot
Original poster
Apr 12, 2001
68,574
39,430


The long-rumored Apple Car could use a "C1" chip based on the A12 Bionic processor and boast in-cabin AI capabilities such as eye-tracking, according to a highly-speculative report by analyst Colin Barnden for EETimes.

Apple-car-wheel-icon-feature-purple.jpg


The report outlines the methodology and licensed technologies believed to be behind an Apple automotive-grade processor, which it tentatively calls the "C1" chip.

Since Apple will require a chip foundry with capacity for automotive processes, the report suggests that Samsung or TSMC could supply Apple. TSMC is believed to have been developing a 7nm automotive-grade process for some time, and Samsung has developed the Exynos Auto V9 SoC on its 8nm process.

With an understanding of the limitations of suppliers, the report suggests that the C1 could bear a close resemblance to the design of the A12 Bionic chip, which is already fabricated with a similar 7nm process, and be manufactured by TSMC.

Tesla's Full Self-Driving Chip has 6 billion transistors and a power consumption of 36W, falling short of Apple's A12, which has 6.9 billion transistors and a power consumption of 3.5W. Due to its parity with existing automotive SoCs, it is speculated that the C1 will be based on the A12 Bionic, before being modified for specific automotive applications.

I'm certain Apple would make some tweaks, changes and additions to keep us all guessing, but if the iCar is to enter production in 2024, then a lightly modified variant of the A12 looks like a great starting point for the C1. After all, why re-invent the wheel?

With the knowledge that Apple licenses technologies that it cannot design itself, such as Arm architectural technologies and CPU cores, the report proposes that Apple will license a number of technologies for the C1. Foremost is the Occula NPU core from Seeing Machines, which would enable Apple to implement a plethora of in-cabin AI features such as driver eye-gaze tracking.

Seeing Machines has adopted an Arm-like business model for licensing the Occula Neural Processing Unit, making it available for Apple to implement into its own custom chip designs. Occula boasts AI and computer vision algorithms, human factors expertise with naturalistic driving data, IR optical path expertise operating, and a 3-pillar embedded processing strategy. The technology is therefore believed to be Cupertino's first choice for designing AI-driven in-cabin monitoring systems.

There is some debate around how close Apple's consumer vehicle is to market, with speculated launch dates varying from 2024 to 2027. Nevertheless, there has been a striking increase in reports surrounding the Apple Car which suggest that the project is gathering momentum, with Hyundai recently confirming that it is in negotiations regarding Apple Car production.

Article Link: Apple Car Speculated to Leverage Eye-Tracking Technology With a 'C1' Chip Based on A12 Processor
 
Last edited:
No, apple doesn’t license CPU cores. And an a12 would be inadequate by 2024, especially if this is to support self-driving. Report is rubbish.

And “automotive-grade processor” is gibberish, as well. For cars you need to modify the package, not the die.
Apple certainly has licensed ARM cores in the past.
 
isn't this car years away? they're going to use a chip released 2 years ago, making it at least 3-4 years old *now*, in a car that they'll boast to be the most technologically advanced motor vehicle ever released?

hard doubt
 
Why the heck are you comparing the A12 to Tesla's FSD chip? They aren't even close to equivalent in design or purpose.

Its got much more of its chip dedicated to running neural networks than Apple does, AND its got 12 ARM cores. Of course its going to have a different power consumption profile for the number of transistors. Its designed to do completely different things.
 
I understand cars need to be very battle tested, but an A12 in 2024 seems crazy. I could see that being what they are working with now for testing, but I'd expect it to be A13-15 based for 2024.
 
I understand cars need to be very battle tested, but an A12 in 2024 seems crazy. I could see that being what they are working with now for testing, but I'd expect it to be A13-15 based for 2024.
I expect there to be multiple CPUs in there for different purposes. The chip that handles self-driving is not going to be the chip that runs the media center. And the chip that handles determining whether you are watching the road will be yet another chip. And some of those will look nothing like an A-anything, because you don’t need general purpose CPU cores for some of that stuff - you want AI cores, etc.
 
click and bait ... someone is out there dreaming ...
yes, automotive is not on the leading edge of silicon technology, but todays technology will not be leading in 5-7 years anymore, plus, if, and imho a big if, if self driving is to take off as so many believe, automotive will have to use state of the art technology and adapt to that ...
 
  • Like
Reactions: KeithBN and trip1ex
I expect there to be multiple CPUs in there for different purposes. The chip that handles self-driving is not going to be the chip that runs the media center. And the chip that handles determining whether you are watching the road will be yet another chip. And some of those will look nothing like an A-anything, because you don’t need general purpose CPU cores for some of that stuff - you want AI cores, etc.
totally agree ... and the self-serving "computers' will be exchangeable, similar to spark plugs, throw them out and replace with the latest one ... cars will still be expected to last 10 years or so, automotive will have to adopt their current thinking ...
 
The A12 ONLY if Apple fixed the Perf Ctrl HW Bug in the A12 (& in the A11) !

Otherwise, the A10 & A13 or newer are their ONLY viable candidates.

The A10, with its simplified Perf Ctrl design, was Rock Solid, Performance wise.

The A11 & A12 each had successively "Fancier" Perf Ctrl designs, which Apple didn't quite get 100% right.

IMO, Apple simply didn't test ALL cases before committing to Silicon !

The A13 does NOT have the HW Bug that the A11 & A12 both have.

The Bug ONLY affects High-Perf CPU-oriented apps.

Few in the General Public are aware of it (few in the General Public have ever had a High-Perf CPU-oriented app on ANY of their iPhones).

The Bug was first discovered on Oct 27th, 2018, the day AFTER the XR was released.

To this date, Apple has never publicly mentioned it, but the evidence is there for ALL to see that they were aware of it, & have done something to resolve it ... simply look at "the timeline" of Apple's public comments regarding the "Performance Controller," & noticed when they stopped talking about it !
 
  • Disagree
Reactions: KeithBN
totally agree ... and the self-serving "computers' will be exchangeable, similar to spark plugs, throw them out and replace with the latest one ... cars will still be expected to last 10 years or so, automotive will have to adopt their current thinking ...

Tell that to Tesla, where my MCU has failed for the second time since 2013. Grrrr.
 
A12?

I would have expected the C1 to be using Real Time CPU something like the Cortex -R series. Which would be different to the current A Series SoC.

At least I assume Real Time CPU is required in these situation. But I could be wrong.
 
Military aircraft use some ancient chips in them, because they are thoroughly tested and a very well known quality.

The last thing you want controlling your flight, or your drive, is bleeding-edge technology.
 
  • Like
Reactions: munpip214
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.