The automobile will likely be the next big battleground for tech companies, with the in-car display ripe for disruption. Compared to smartphones, in-car "infotainment" systems today are as dumb as bricks, and the long development cycles ensure that in-car tech is outdated before it even reaches the dealership.
Apple's recently announced CarPlay has gotten a lot of attention, allowing for iPhone users to use the in-car display to access select iPhone apps, as well as Siri, in models that support the system.This solves the problem of outdated in-car software, at least for iPhone users. But the prospect of driver assistance features and, eventually, semi-autonomous and fully autonomous vehicles is far more exciting. This will require a significant amount of on-board processing power, and there is no company better-positioned to provide that processing power than NVIDIA (NVDA 3.65%).
The car becomes a visual computer
NVIDIA's Tegra line of mobile processors consists of an ARM-based CPU coupled with the company's graphics technology. Some cars, like the Tesla Model S, already use older versions of Tegra to power the in-car display.
The most recently announced Tegra, the Tegra K1, is notable for two reasons. First, it provides an enormous increase in processing power compared to earlier versions. And second, it's the first version of Tegra that is built on the same graphics architecture as NVIDIA's GPUs.
This increase in processing power allows the Tegra K1 to analyze video data on the fly, enabling various driver-assistance features that previous iterations of the chip couldn't handle. At CES in January, where the K1 was first announced, NVIDIA set up a demo showing off various functions like lane detection, street sign detection, and car detection. The K1 was able to churn through the video data in real time, tagging objects in the video and displaying the current speed limit on the screen after detecting and analyzing speed-limit signs.
Pedestrian detection. Source: NVIDIA
The fully autonomous car of the future will rely on cameras, sensors, lasers, and radar to probe the world around it, and the on-board computer system will need to be able to process this raft of data in real time. GPUs are ideal for this type of application, where data can be broken into chunks and processed in parallel across many graphics cores. The Tegra K1 has 192 graphics cores, and future iterations of the chip will undoubtedly have an even greater number.
NVIDIA isn't the only company aiming to get its chips inside automobiles. Intel (INTC -0.38%) is also making a push to extend its dominance beyond PCs. Intel is positioning its low-power Atom chip as a way for automakers to reduce development time and cost. But while Atom is a compelling choice to power in-car displays, it lacks the power to do the complicated image and video processing required for driver-assistance applications.
An example of why a powerful GPU is better suited for this type of work can be seen in certain PC applications that have achieved significant performance boosts by utilizing the GPU. Adobe Premier Pro is a popular video editing application. By offloading processing from the CPU to the GPU, functions that require heavy analysis and processing can be sped up by a factor of nearly eight, according to an NVIDIA white paper on the subject.
NVIDIA is already gaining some traction with the Tegra K1. Audi, a company that uses NVIDIA's earlier Tegra chips to power the displays in some of its cars, will be using the Tegra K1 for all of its future piloted-driving initiatives. These range from automated parking to the far-off goal of fully autonomous vehicles. NVIDIA won $1.9 billion worth of automotive business in fiscal 2014, all for previous versions of the Tegra and set to be delivered over the next few years as the models come to market. This early momentum is a good sign that Tegra is beginning to become the platform of choice.
The bottom line
The amount of processing power inside of automobiles is going to explode as driver-assistance functionality and eventually fully autonomous vehicles become a reality. NVIDIA's Tegra is well-suited for the type of processing required, and being the leading graphics chip company gives NVIDIA an important advantage. While fully autonomous vehicles are still a fantasy, NVIDIA's focus on driver-assistance functions should serve the company well as its fights for automotive market share.
