Inside Mobileye’s EyeQ Ultra chip for future self-driving cars
At the Consumer Electronics Show, Mobileye unveiled the next generation of its flagship system-on-chip (SoC), called EyeQ Ultra, specifically designed for cars that can almost entirely drive themselves.
Unveiled at CES 2022, the EyeQ Ultra aims to provide enough processing power for what experts call “Level 4” autonomous driving, which means cars can operate in autonomous mode in most situations, but not all.
Intel’s self-driving unit has become an early leader in brain-locking self-driving cars. For years it offered camera-based systems with all of its hardware and software inside to companies trying to add safety features like emergency brake assist and warnings. Lane Departure and Blind Spot Assist, also known as Advanced Driver Assistance Systems (ADAS). — to cars. There are millions of these systems on the road today.
The company previously said it has shipped over 100 million EyeQ processors to date.
Mobileye has promoted its ADAS systems as a stepping stone for self-driving cars. Last year, the company announced that it had started testing prototype self-driving cars in several major cities, including New York. The self-driving giant is also developing internal lidar sensors to complement the radar sensors. Combined with its camera-based systems, the sensors will help vehicles “sense” their surroundings more.
With EyeQ Ultra, it’s raising the bar in its battle with Nvidia and Qualcomm to propel self-driving cars into the next decade. Mobileye already sells a set of hardware and software that can add Level 4 autonomous driving mode to cars, but it uses several chips that are too power-hungry for consumer-grade cars. Integrating such a system into a single SoC will improve its cost and power efficiency, which will also allow it to operate cooler.
under the hood
Harnessing 5nm process technology takes the performance of the EyeQ Ultra chip to another level.
The EyeQ Ultra can deliver up to 176 trillion operations per second, or TOPS, for performing artificial intelligence workloads in cars and up to 4.2 TFLOPS for handling more general processing tasks. the company said.
Mobileye said the EyeQ Ultra delivers performance equivalent to 10 of its latest EyeQ 5 chips combined in a highly power-efficient package of less than 100W. The EyeQ 5 is expected to be used in production vehicles in 2022.
While not on par with Nvidia’s Orin and Qualcomm’s Snapdragon Ride chips in terms of performance, Mobileye CEO Amnon Shashua said the Ultra chip has more than enough muscle to control a consumer autonomous vehicle. It also gains in energy efficiency, he said. Power reduction is becoming increasingly important for electric vehicles, as self-driving systems will need to run on the same battery as the powertrain.
“It’s not just about TOPS,” Shashua said during a press conference at CES this month. “Efficiency comes from understanding the very, very close interaction between hardware and software.”
Mobileye also controls costs by packing more performance into a single monolithic chip. The Ultra will cost less than $1,000 when it becomes widely available to automakers, Mobileye said.
Mobileye said prototypes of the Ultra will be ready in 2023, with full automotive-grade production expected in 2025.
Hit the accelerators
The EyeQ Ultra packs 12 dual-threaded CPU cores based on the open source RISC-V architecture, a move away from the MIPS CPU cores of its predecessors. Also under the hood is a graphics processing unit (GPU) from Arm that can run up to 256 billion FLOPS and a vision processing unit (VPU) to help cars “see” the world around them. . The chip integrates an image signal processor (ISP) and a set of H.264/5 video encoding cores.
The chip is complemented by four classes of proprietary accelerators, each suited to a specific set of tasks. The chip integrates 64 such cores, including 16 convolutional neural network (CNN) accelerators, divided into separate clusters. There are also 48 general-purpose accelerator cores, with eight programmable cores based on its coarse-grained reconfigurable architecture (CGRA); 16 cores based on its hybrid Very Long Instruction Word (VLIW) and Single Instruction, Multiple Data (SIMD) design; and 24 “cylindrical-threaded” CPU cores.
The chip adds internal redundancies for security as well as a security subsystem. The EyeQ Ultra can be paired with a separate microcontroller (MCU) for ASIL-D systems, Mobileye said. It also supports LPDDR5X memory.
Mobileye said Ultra can process data from two sensor subsystems – one camera-only system and the other combining radar and lidar – and the vehicle’s central computer system, high-definition map and traffic policy software. conduct. All cameras, radars and lidar sensors provide the vehicle with 360 degree coverage of the world around it.
Ultra is the “crowning jewel” of its offerings, Shashua said, likening it to an “autonomous vehicle on a chip”.
Whereas it previously bundled its chips with its in-house software, sensors and apps in a “black box”, Mobileye said it has decoupled software and hardware from the EyeQ Ultra chip to give customers more flexibility. Mobileye said Ultra is not required to run its in-house software, which allows auto giants to buy the hardware or software a la carte and run their in-house apps on top of it — or to purchase a complete package from Mobileye.
eyes on the road
At CES, the automotive chip giant also announced a pair of 7nm processors for ADAS: EyeQ 6L and EyeQ 6H.
The EyeQ 6L is ideal for car front camera systems. The chip is faster and more energy efficient than its predecessor, the EyeQ 4, meaning it can be mounted behind a car’s windshield to monitor road hazards. The chip, which fits in a 50% smaller package than its predecessor, offers more performance to run advanced camera-based safety features such as automatic emergency braking (AEB) and lane control.
Mobileye started sampling the EyeQ 6L last year, and it’s expected to go into production by mid-2023.
The EyeQ 6H is intended for more advanced safety features in cars that rely on multiple surround cameras, such as lane change assist and automated parking. Mobileye said the EyeQ 6H, which will be the most advanced driver assistance chip in the family when it becomes widely available, will have three times the performance of its EyeQ 5 while burning only 25% of more power, which will give him a boost for performing AI chores in cars.
The EyeQ 6H will begin sampling later this year and enter full production by the end of 2024.
A long way to go
At CES, the company said it was continuing to strike deals to supply self-driving car brains to automakers. Mobileye has announced plans to partner with automaker Zeekr to deploy what the companies say will be the world’s first Level 4 autonomous vehicle in China by 2024.
The companies said the model would use six EyeQ 5 chips and use digital maps generated by Mobileye’s Road Experience Management (REM) system, which aggregates data from all cars on the road with Mobileye hardware.
Mobileye said it also agreed to provide its road mapping data to Ford and Volvo. Digital road maps are accessed via the cloud in real time to provide up-to-date information that helps supplement a car’s on-board sensors.
But last year Intel said it planned to build part of its self-driving unit, in a bet the automotive chip business would become an increasingly important market for the chipmaker. Intel, which is trying to raise funds to build chip factories and bridge the gap with TSMC and other rivals that have ended Intel’s leadership in chip manufacturing, said it plans to retain a majority stake in the company after Mobileye’s IPO in 2022.
Intel said splitting the unit would give the standalone company a higher profile and the ability to attract more business. Intel said the two companies will continue to collaborate on technologies for the automotive sector.
Intel acquired Mobileye in 2017 for around $15.3 billion.