Ambarella launches 16-core CV3 AI chip for autonomous vehicles

Hear from the CIO, CTO and other C-level and senior executives on data and AI strategies at the Future of Work Summit on January 12, 2022. Learn more

Ambarella has launched its latest CV3 AI domain controller family to power autonomous vehicles.

The California-based maker of AI chips, Santa Clara, made the announcement at the CES 2022 tech trade event in Las Vegas.

The Chip Family is another addition to the scalable, power-efficient CVflow family of system-on-chips for the automotive industry.

Ambarella said the chip has the highest AI processing performance of up to 500 eTOPS, showing a 42-fold increase over Ambarella’s previous automotive family.

Featuring up to 16 Arm Cortex-A78AE CPU cores, the CV3 provides a 30-fold increase in CPU performance compared to the previous generation in support of the Autonomous Vehicle (AV) software application.

Above: Block diagram of Ambarella CV3.

Image credit: Ambarella

This family enables centralized, single-chip processing for multi-sensor perception – including high-resolution vision, radar and ultrasonic – as well as deep fusion for multiple sensor modalities and AV path planning.

The result is a robust Advanced Driver Assistance System (ADAS) and L2 + to Level 4 Autonomous Driving (AD) systems, with higher levels of environmental perspective in challenging lighting, weather and driving conditions for both driver viewing and machine perception.

“With our new CV3 AI domain controller family, we are now able to run full ADAS and AD stacks with a single chip, while providing unprecedented performance and power efficiency,” Fermi Wang, CEO of Umbrella, said in a statement. “Improvements in our on-chip ISP, along with simultaneous radar processing that can take advantage of our Oculii adaptive AI algorithm, have helped the Ambarella automotive industry achieve greater levels of perception accuracy in all environmental conditions to realize its promise of autonomous driving. “

CV3 is the beginning of Umbrella’s next generation CVflow architecture, which continues the company’s algorithm-first design philosophy.

The company said this has led to the development of an on-chip neural vector processor (NVP), with up to 500 eTOPS of AI compute, support for the latest advances in industry-leading power efficiency and neural network (NN) predictions.

Advanced radar perception software, such as the Oculus Adaptive AI software, also enhances NVP to run algorithms. This is complemented by a new floating-point general vector processor (GVP), designed to offload classical computer vision and radar processing from NVP engines, and floating-point-intensive algorithms from arm CPUs, the company said.

Ambarella said the CV3 family’s unique hardware scalability allows automakers to integrate their software stacks into their entire retail portfolio, while minimizing the cost and complexity of software development. This scalability directly addresses the growing complexity of automotive software by offering the option of a fragmented ADAS SoC offering from competitors.

Above: Ambarella CV3 chip photo.

Image credit: Ambarella

In addition, the CV3 accelerates the development timeline of family automakers while facilitating the deployment of new features by providing headroom for single, robust over-the-air (OTA) update implementation, Umbrella said.

This new family also supports both central and zonal architectures. In addition to covering the entire AD stack, the CV3 can simultaneously process in-cabin sensing applications, including driver and occupant monitoring.

Ambarella’s AI Perception SoCs are uniquely positioned to support a full range of viewing, recording, sensing and path planning applications from basic ADAS to L4 AVs. This means that automakers no longer need to develop separate software stacks for their entry-level, mid-range and premium vehicles, as they can use Umbrella’s integrated CVflow platform across all models, saving engineering costs and market trends. Can respond quickly.

Using a comprehensive set of mature software tools and a robust SDK, Ambarella’s flexible CVflow AI platform provides customers with an unprecedented opportunity to innovate and differentiate their products in the market. To further expand this capability, Ambarella has developed a comprehensive ecosystem of software partners for the CVflow platform, working closely with them to port and optimize their application, the company said.

Ambarella also said that CV3 also integrates Ambarella’s next generation ISP, which extends the company’s lead in image signal processing quality. Supporting 12 physical or 20 virtual cameras simultaneously, a single CV3 can process the entire sensor suite, including 10 cameras, five radar modules, and numerous ultrasonic sensors for typical L2 + deployments. In addition, high-performance stereo and dense optical flow engines provide greater depth and speed perception.

The CV3 Car Graphics Processing Unit (GPU) enables applications such as 3D surround-view rendering, hardware security modules, fast interfaces and processing headrooms for over-the-air updates.

The first SoCs in the CV3 family are expected to be available for sampling in the first six months of 2022.


VentureBeat’s mission is to become a digital town square for technical decision makers to gain knowledge about transformative technology and practices. Our site delivers essential information on data technology and strategies to guide you as you lead your organization. We invite you to access, to become a member of our community:

  • Up-to-date information on topics of interest to you
  • Our newsletters
  • Gated idea-leader content and discounted access to our precious events, such as Transform 2021: Learn more
  • Networking features and more

Become a member

Similar Posts

Leave a Reply

Your email address will not be published. Required fields are marked *