The intelligent cockpit is the most similar part of a car to consumer electronics, such as smartphones, and it is also a very important entry point for Internet IT giants to enter the field of intelligent cars. Through this entry point, we can not only state the facts and compare traditional car companies with how they cannot keep up with the trend of the information age, but also provide a convenient and diversified means of human-machine interaction and inter-personal interaction. Therefore, having more in-cabin sensors and intelligent cockpits that are more closely tailored to the application scenarios of cars will be the inevitable choice of the next wave of intelligent car giants.
Of course, the intelligent cockpit is also a product that is constantly evolving, and with the increasing degree of autonomous driving, it has more application scenarios. If we simply equate the intelligent cockpit with “one machine with multiple screens” or only in terms of functional voice interaction, real-time navigation, whole-vehicle function setting, and driver status detection, it would be too one-sided. Here, let’s share Tesla’s approach with you.
Tesla’s Intelligent Iteration
In the two major systems of intelligent cockpit and autonomous driving, Tesla’s hardware iteration speed is also very fast. As shown in the following figure, from the timeline perspective, MCU1 was launched in 2012, using Nvidia Drive CX, which is a processor based on Tera X1; MCU2 was launched in 2017, switching from Nvidia to Intel’s Atom E3950; and with Model S Plaid, it uses AMD Ryzen+Navi 23 GPU. In the field of autonomous driving, from the first-generation Autopilot HW1 with Mobileye Eye Q3 to the second-generation Autopilot HW2 mainly using Nvidia’s Drive PX2, which uses a Parker and Pascal GPU, two Parkers and Pascal GPU were added one year later. When Tesla entered the chip field in the third generation, it used its own FSD chip.
In terms of urgency, Tesla spent more energy on intelligent driving, while the main chip iteration of the cockpit took about 4 years. In an earlier article, Tesla introduced the “Model 3 Information Entertainment Module MCU Update.”
In a Skywell Auto document, a more complete depiction of the MCU2 block diagram was provided for reference.
On the known Tesla MCU3, a 17-inch 2200×1300 resolution screen is used, along with AMD’s NAVI 23 GPU chip, which logically corresponds to the instrument panel of the Model S Plaid, with several video outputs (2.7Gb/s HBR and 5.4Gb/s HBR2), as well as an HDMI output, corresponding to the displays of the instrument panel, central control screen, and rear-seat screen. The system also utilizes 16Gb of GDDR6 memory, with the central processing unit being an AMD Ryzen processor. Essentially, Tesla is pursuing more extensive capabilities beyond traditional vehicle functions, as evidenced by their impressive display abilities.
Features supported by the smart cockpit
The chart below divides the functions of the car computer and basic Autopilot functions from the User Manual of the Model S Plaid:
Based on current software function descriptions, it appears that the MCU3 can provide more gaming functions. Nevertheless, one can imagine that on this platform, more far-reaching vehicle-related ecosystem development can be explored, as these hardware iterations mainly pave the way for the future.
In conclusion, the importance of the infotainment system and smart cockpit computer cannot be overstated, as they are paramount for individual car models. Furthermore, I am also conducting some systematic statistics on the chip sector, and Tesla seems to be the most unique enterprise in this regard.
References:
- Figure 3 is taken from “Future Trends of Multi-screen Applications in Smart Cars” published by Cheshi.
This article is a translation by ChatGPT of a Chinese report from 42HOW. If you have any questions about it, please email bd@42how.com.