Arm Unveils Its Largest Portfolio of Automotive Processors


Arm recently announced a new suite of Automotive Enhanced (AE) processors to accelerate automotive computing. The new lineup focuses particularly on AI-driven use cases and safety-critical applications. It leverages the advanced Armv9 architecture and integrates server-class performance and enhanced AI capabilities. Arm designed these devices to cater to the burgeoning demands of modern software-defined vehicles (SDVs).

 

Arm’s new AE processors include a server-class, automotive-focused Neoverse CPU, two Armv9 A-class CPUs, an R-class CPU, and an image signal processor (ISP).
 

All About Circuits sat down with Dipti Vachani, Arm’s SVP and GM of automotive, to learn more about the new set of AE processors and what they bring to the table. 

 

The New Automotive Enhanced (AE) Processor Lineup

The Neoverse V3AE processor is at the core of the AE lineup—a significant move to port Neoverse technology into the automotive sector for the first time. The V3AE is designed to deliver server-class performance, providing a more than 50% uplift in per-core performance over its predecessor, the Cortex-A78AE. The Neoverse V3AE integrates Armv9’s novel security features, such as memory tagging extensions, pointer authentication, branch target identification, and hardware-based virtualization support. 

“From our discussions with partners, it’s very clear that with this growing demand for AI, there was a need for more high performance in automotive,” Vachani said. “These autonomous use cases demand more and more raw CPU performance for the AI workloads. To deliver on this need, we looked to our cores that were being developed by the data center market.”

 

Arm’s new lineup of AE processors

Arm’s new lineup of AE processors
 

Complementing the Neoverse V3AE are the Arm Cortex-A720AE and Cortex-A520AE processors. The Cortex-A720AE, purpose-built for automotive applications, offers sustained performance and system-on-chip (SoC) design flexibility. The device is tailored for a broad spectrum of SDV applications, from infotainment systems to more complex autonomous driving functions. The Cortex-A520AE, on the other hand, is engineered for power efficiency, incorporating functional safety features to scale across a wide array of automotive applications, ensuring both performance and safety.

 Arm positions the new Cortex-R82AE processor as its highest-performing real-time processor for functional safety. For the first time, it introduces 64-bit computing to real-time processing, addressing the demanding requirements of real-time automotive systems.

“Now, with our v9 architecture, software developers can develop and test their software in the cloud and deploy it to automotive edge devices using the same instruction set architecture,” Vachani said. “This is huge for the software-defined vehicle because it can now reliably and confidently deploy software to the vehicle in real time.”

The new Arm Mali-C720AE image signal processor (ISP) rounds off the AE portfolio. This device allows developers to tune multiple parallel pipelines across computer vision (CV) and human vision use cases. By accommodating both use cases, the ISP reduces memory use, saves SoC power and space, and cuts the time to process images. Users can also add or remove processing blocks depending on their image processing requirements and tune image quality for CV-based applications. 

Each processor in the AE lineup is further complemented by a suite of configurable system IPs, enabling the Arm silicon ecosystem to produce scalable, high-performance automotive SoCs. 

 

Compute in the Era of Software-Defined Vehicles

The conventional sequential development process—wherein hardware design and manufacturing precede software development—poses significant challenges in this new era. This model extends product development cycles, delays the introduction of new features, and diminishes an automaker’s competitive edge. Because many vehicles are now reliant on software for both core operational functions and differentiating features, OEMs must be able to rapidly iterate and deploy updates.

“The serial methodology of building the hardware, waiting for the silicon to get to market, and then developing software on silicon is no longer going to serve the automotive market. It’s just not possible anymore,” Vachani said. “These hardware and software enablement platforms must all come to market together at the same time from day one to truly have an impact on the time to market for automakers.”

Furthermore, the computational demands of AI and machine learning algorithms are orders of magnitude higher than traditional automotive computing tasks. These algorithms require not only high-performance computing cores but also specialized processors that can efficiently handle parallel tasks, neural network processing, and real-time data analytics—all within the stringent power and thermal constraints of automotive environments.

 

Example of an automotive compute subsystem

Example of an automotive compute subsystem
 

To address these challenges, there is a pressing need for automotive processors that can provide the necessary compute performance and efficiency. This includes processors with advanced architectures that support faster, more efficient processing of AI workloads, enhanced security features to protect against evolving cyber threats, and the flexibility to support a range of automotive applications from ADAS to fully autonomous systems. Additionally, virtual prototyping and software development platforms help developers begin software design and testing well ahead of hardware, significantly accelerating the overall development process and enabling faster time to market for new innovations. 

“Our virtual prototyping solutions make this announcement unique compared to anything we’ve ever done with Arm and its ecosystems before,” Vachani said.

 

“Think about the digital twin technology that’s in the factory automation space. We’re essentially bringing this concept to the automotive market. This is what allows the automotive developers to work together early ahead of physical design being available.”

 
Arm Does What Arm Does Best: Automotive Edition

For automakers to thrive in the era of software-defined vehicles, they must take a holistic approach to automotive development, integrating advanced processing capabilities with new software development methodologies. With Arm’s new lineup of AE processors, the company hopes to provide automotive engineers with the computing resources and development support necessary to thrive in this era.

“This stack of solutions is a stepping stone toward supporting our ecosystem with a full CSS to bring them faster, more agile, and better TCO,” Vachani said. “There will be pre-configured production quality RTL subsystems that are specifically designed for the safety and software needs of the vehicle application. This empowers our ecosystem and our partners to focus their limited resources on key differentiation, while Arm does what we do best: deliver a scalable and efficient compute foundation.”

 

 

All images used courtesy of Arm.



Source link

Leave a comment