Samsung' 5nm selected by Ambarella for New Automotive AI Central Domain Controller
New CV3-AD685 System-on-Chip advances Autonomous and Advanced Driver Assistance System (ADAS) applications, providing single-chip solution for multi-sensor fusion and path planning
Integrated Ambarella CV3-AD685 system-on-chip built on Samsung Foundry’s 5nm technology
Samsung Electronics, a world leader in advanced semiconductor technology, and Ambarella, Inc., an edge AI semiconductor company, announced that Samsung’s Foundry business is providing its 5-nanometer (nm) process technology to Ambarella for its newly announced CV3-AD685 automotive AI central domain controller. This collaboration will help transform the next generation of autonomous driving vehicle safety systems by bringing new levels of AI processing performance, power and reliability.
The CV3-AD685 is the first production version of Ambarella’s CV3-AD family of automotive AI central domain controllers, with Tier-1 automotive suppliers announcing they will offer solutions using the CV3-AD system-on-chip (SoC) product family. Samsung’s 5nm process technology is optimized for automotive-grade semiconductors, with extremely tight process controls and advanced IP for exceptional reliability and outstanding traceability.
Ambarella will rely on Samsung’s 5nm process maturity and the technology’s solid track record. This 5nm process is backed by the company’s extensive experience in automotive foundry process, IP, and service package development to enable manufacturers to create cutting-edge innovations in assisted and automated mobility.
“Ambarella and Samsung Foundry have a rich history of collaboration, and we are excited to bring their world-class 5nm technology to our new CV3-AD685 SoCs,” said Fermi Wang, President and CEO at Ambarella. “Samsung’s proven automotive process technology allows us to bring new levels of AI acceleration, systems integration and power efficiency to ADAS and L2+ through L4 autonomous vehicles.”
The CV3-AD685 integrates Ambarella’s next-generation Cvflow AI engine, which includes neural network processing that is 20 times faster than the previous generation of Ambarella’s CV2 SoCs. It also provides general-vector and neural-vector processing capabilities to deliver the overall performance required for full autonomous driving (AD) stack processing, including computer vision, 4D imaging radar, deep sensor fusion and path planning.
“Samsung brings 5nm EUV FinFET technology to automotive applications for unprecedented ADAS and vision processor performance,” said Sang-Pil Sim, executive vice president and head of Foundry Corporate Planning at Samsung Electronics. “With Tier-1 automotive suppliers already adopting the technology, we believe other automotive companies will also consider using the Ambarella CV3-AD SoC product family manufactured in Samsung’s 5nm process.”
The CV3-AD685 will be the first in the CV3-AD product family to use Samsung’s 5nm process, and this SoC integrates advanced image processing, a dense stereo and optical flow engine, ARM Cortex A78AE and R52 CPUs, an automotive GPU for visualizations, and a hardware security module (HSM). It features an “algorithm first” architecture that provides support for the entire autonomous-driving software stack.
The high-performance, power efficient and scalable CV3-AD family, which is built specifically for ADAS, complements a wide range of solutions for assisted driving, while advancing vehicle automation. The integrated CV3-AD685 SoC enables information from various sensors to be fused for robust L2+ to L4 autonomous driving. Samsung Foundry’s industry-leading process technology and advanced 3D-packaging solutions are powering many of the latest mobile, HPC and automotive solutions.
Samsung’s 5nm process is also backed by the Samsung Advanced Foundry Ecosystem (SAFE) program. The SAFE program ensures close collaboration between Samsung Foundry, ecosystem partners, and customers to deliver robust SoC designs based on certified key design components including Process Design Kits (PDK), reference flows with Design Methodologies (DM), a variety of Intellectual Properties (IP), and on-demand design support.
Author: Srinivasa Reddy N