Arm answers AI's push into PCs, smartphones with CSS for Client

The growing presence of AI PCs and AI smartphones shows that AI-driven demand for more powerful computing and greater power efficiency is not a trend confined to data centers. As AI experiences increasingly become part of humans’ everyday lives, AI support needs to extend all the way through the devices end users rely on for those experiences.

That makes client devices a fitting target for Arm’s newest compute subsystem architecture, the Arm Compute Subsystems (CSS) for Client, which aims to combine the AI performance, security and other benefits of the Armv9 CPU architecture with “validated and verified production-ready implementations of new Arm CPUs and GPUs on 3nm process nodes,” the company said in a statement, adding, “CSS for Client provides the foundational computing elements for flagship SoCs and features the latest Armv9.2 CPUs and Immortalis GPUs, as well as production-ready physical implementations for CPU and GPU on 3nm and the latest Corelink System Interconnect and System Memory Management Units (SMMUs).”

Specifically, the processor cluster serving as the foundation for CSS for Client is based on  Armv9.2, and includes the Cortex X925, Cortex A725, and Cortex A520 CPU cores, along with the new Arm Immortalis G925 GPU. As for how all of that translates into the end user experience, Arm official said during a briefing that for Android devices, CSS for Client delivers “greater than 30% increase on compute and graphics performance and 59% faster AI inference for broader AI/ML and computer vision (CV) workloads.”

The use of 3nm process node technology is significant because it is said to yield greater performance and reduced form factor and power consumption for AI.

Also, as AI software becomes more important in extending the performance benefits of AI-optimized processors throughout the user experience, Arm also announced the KleidiAI software library, which supports popular AI frameworks to enable easier developer experiences. When combined with CSS for Client, KleidiAI helps to drive application performance improvement by leveraging Arm acceleration technologies, like the NEON instruction set architecture and extensions like SVE2 and SME2.

“We are seeing significant innovation from phones to laptops, which are now being characterized as both AI smartphones and AI PCs,” said Chris Bergey, senior vice president and general manager of Arm’s Client Business. In particular he pointed to how AI capabilities in new phone models like the Samsung S24 are “redefining the ‘smart’ in smartphones.” 

He added, “With more AI integration across everything, from productivity to entertainment applications, what's possible is changing right in the palm of your hand. We're now seeing new use cases and form factors facilitating human to human interaction across language barriers, hinting I would have still yet to come. At the Samsung S24 launch [earlier this year] we saw real-time multi-language translation for better communication [as well as new demonstrations of how] smartphone cameras gather and process light for photos and videos. These features were powered by Samsung’s AI model, Galaxy AI, which runs on top of Google Gemini.”

Bergey also noted that Google Gemini is not the only AI assistant emerging, and that AI’s influence will not be felt only in PCs and smartphones, but also TVs in every consumer home, and in an increasingly wide range of wearables.