Throughout this last decade, the automotive industry has seen some immense growth. A huge milestone for this industry was the introduction of affordable, efficient, and powerful electric cars.
With that, we also got another breakthrough: autonomous driving.
Of course, AI driving is still very far from perfect, but with enough research and processing power, soon, cars will be able to drive on their own.
However, to power the artificial intelligence in a Tesla, BMW, Porsche, or any other vehicle, a powerful GPU (and CPU) is needed. In fact, GPUs are an essential part of every new vehicle today.
This article will explain the immense importance of graphics cards for vehicles and autonomous driving.
Let’s get right to it!
Table of ContentsShow
Before we get into the complicated matter of artificial intelligence, autonomous driving, and the role the GPU must take, let’s first talk about some of the more basic needs of why cars need a graphics card.
It is no secret that most modern vehicles of today have a display. And like other displays, a computer is needed (a GPU) to render the image.
However, the graphics cards found in most vehicles are not as powerful as AMD’s or NVIDIA’s current flagship GPUs.
This doesn’t necessarily mean they’re not powerful, but the tasks they need to handle are much more straightforward. Something like processing/rendering the car’s interface with which the driver interacts.
Some vehicles have a weaker GPU/CPU, leading to a slower (not responsive) infotainment system. A less powerful GPU also limits the possibilities of additional features.
Here are some features found in today’s cars that require a fast GPU:
Many new car models come with tons of optional features for the infotainment system that can assist the driver.
Nowadays, parking cameras are essential, especially for larger cars with many dead angles, like most SUVs. There are options for a camera on the front, back, or sides that help the driver avoid bumping into other cars, scraping sidewalks, etc.
Naturally, to process all these cameras/sensors, the car must have a GPU to render the image. However, rendering an image from a camera is not very demanding, so you can imagine that the GPUs are not that powerful either.
Bird’s Eye View And Environment Rendering
To further improve this feature, companies in the industry have started implementing much more advanced sensors and cameras to detect objects in the surrounding area. The latest invention is the bird’s eye view parking assistance.
Definitely a more complex feature than having a few cameras recording the car’s dead angles. Because of this, the GPU now has to process the sensors and live-render the entire environment around the vehicle so that drivers can have a much clearer understanding of their surroundings.
For these features, the automotive industry has to use more powerful GPUs. In the future, when we get life-like renders, there may be GPUs inside cars that are more powerful than today’s fastest graphics cards, like AMD’s RX 6900XT or NVIDIA’s RTX 3090.
Another relatively new and helpful feature is the heads-up display or HUD. The whole purpose of a HUD is to prevent the driver from looking down at the instruments by showing the needed information on the windshield.
This can be achieved by projecting the image onto the windshield. Alternatively, a dedicated, transparent display is placed right in front of the windscreen where the information is projected. A car HUD is usually used as a speedometer and GPS assistance.
In the image below, the HUD in this BMW acts as a speedometer and shows GPS directions. This effectively reduces the need to glance away from the road.
Again, this is not a very demanding process for the vehicle’s GPU, but it does have to process other stuff too, so it still needs to be powerful enough.
The Rise Of Autonomous Driving
The idea behind autonomous driving or Advanced Driver Assisted Systems (ADAS) is that cars in the future will not be operated by humans unless necessary. Ultimately, this will remove human error from the equation, making vehicular transport faster, more efficient, and considerably safer.
However, it will take some time until we reach that level of autonomy and build the perfect AI driver.
Fortunately, Advanced Driver Assisted Systems saw a considerable increase in popularity after Tesla’s electric cars (like the Model S, Model X, and Model 3) came out with an autopilot as we’ve never seen before.
Soon after, other brands in the automotive industry started introducing their own electric cars with similar driver assisting features.
GPUs’ Role In Autonomous Driving
We previously delved a bit into autonomous driving and that GPUs are a must to process the information on the road. But let’s go in more depth and explain how GPUs and tech giants like NVIDIA, AMD, and Intel are now a part of the automotive industry.
Highway and daily traffic are exceptionally complicated, which means that vehicles need powerful hardware to handle all those “autopilot” calculations.
While every car has a CPU, often called ECU (the brains of the entire operation), it is not powerful enough to process data for autonomous driving.
This is where graphics cards come in. Unlike processors, the GPU dedicates its vast processing power to specific types of tasks. For example, in cars, the GPU processes various visual data from cameras, sensors, etc. which is then used to automate the driving.
NVIDIA’s Self-Driving Technology
NVIDIA is a massive part of the automotive industry regarding self-driving technology. It is probably the leading company in this category.
Currently, NVIDIA is partnered with Audi, Jaguar Land Rover, Mercedes-Benz, NIO, Volkswagen, and Volvo Cars. We imagine that this list of partners will soon grow.
While Tesla doesn’t use NVIDIA hardware for its infotainment system or autopilot system, they use an NVIDIA-powered supercomputer with 5,760 A100 GPUs. A supercomputer that handles Tesla’s AI training to further develop their autopilot.
Companies like Audi and Mercedes-Benz have stuck with NVIDIA because of their DRIVE Hyperion platform for autonomous vehicles.
This computer architecture is wholly dedicated to autonomous driving. In other words, it’s specialized to process data from cameras and sensors to deliver the best self-driving experience.
NVIDIA boasts that DRIVE Hyperion is also highly customizable, which is definitely what many brands are looking for.
To get all of the benefits of the DRIVE Hyperion platform, the car needs to be equipped with a DRIVE Orin SoC.
The term system-on-a-chip (or SoC) means that all the required components for a computer are integrated into the chip itself. So, technically, DRIVE Orin is not just a GPU, but it is included in the chip and has a considerable role in the process of self-driving.
With the power of the Orin SOC, Level 5 autonomous driving is possible. A Level 5 system means that no driver is needed at all. Predecessors of ORIN like the DRIVE AGX Pegasus and Xavier could only deliver Level 2 and Level 3 self-driving
As we all know, NVIDIA constantly strives for innovation and technological advancement, like the Orin chip or the Tensor cores in their GPUs that handle DLSS.
So, naturally, Orin’s successor is already in the works. NVIDIA’s DRIVE Atlan will deliver 1,000 TOPS of computing performance.
AMD’s Role In Cars
AMD isn’t widely used in the automotive industry, but they have partnered with Tesla to power their car’s infotainment system.
Currently, Tesla has the most powerful infotainment system in the world.
Want to know the best part?
During the reveal of the Tesla Model S Plaid, Elon Musk claimed that the infotainment system actually has PS5-level of performance thanks to AMD’s Ryzen APU that can deliver 10 TFLOPS of computing power.
Elon Musk proved these claims by showcasing a Cyberpunk 2077 gameplay on the Model S’ infotainment system. That’s quite remarkable.
Inside of this APU is an RDNA 2 GPU, which enables a Tesla to run demanding games such as CP2077.
Here’s a video of the Cyberpunk 2077 gameplay in the Model S Plaid.
Why Do Tesla Users Need A Powerful GPU?
While Tesla’s powerful infotainment system is definitely impressive, it might leave many people wondering why it is even needed.
Well, here’s the deal:
Tesla is already very close to fully automated driving, so in the near future, when the car takes over the entire drive, people can enjoy all kinds of entertainment on the car’s screen. Anything from watching movies on Netflix to high-end gaming.
Even today, having such a powerful GPU is helpful because when Tesla users end up on a charging station, they can just fire up some games to make the wait times just a little bit more interesting.
Intel In The Automotive Industry
Before Tesla switched over to AMD’s powerful Ryzen APU, they relied on Intel Atom. Intel’s Atom chip was significantly slower but still capable enough to handle Tesla’s infotainment system.
Intel has also collaborated with Audi in the past to deliver a Level 3 autonomous driving system.
Intel is also advertising their Intel FPGAs, which will accelerate the production of Advanced Driver Assisted Systems.
Intel’s FPGAs are focused on flexibility, safety, security, and efficiency, but, for now, we don’t exactly know whether they are actively used in any vehicle.
However, we know that Intel’s subsidiary, Mobileye, is deeply focused on delivering chips dedicated to autonomous driving. During the 2022 CES, Mobileye announced its next-generation SoCs for ADAS, EyeQ6L, and EyeQ6H.
The L-version will be able to handle Level 2 autonomous driving systems and enter production in the middle of 2023. The more powerful chip, EyeQ6H, will go into production a bit later (in 2024) and will provide much better support for ADAS.
At the time of writing, Mobileye is partnered with BMW, Nissan, NIO, Volkswagen, Ford, and WILLER for autopilot development.
Tesla might’ve been the first to add an infotainment system that offers features similar to consoles, but we imagine that this trend will slowly spread to other car brands too.
In other words, powerful GPUs will soon become a necessity for all car models. Either for gaming, media, and different types of entertainment, or the GPUs will be used to support the Level 5 autonomous driving systems.
Expect that all future improvements and updates on car GPUs will be added to this article.