Arm CSS for Client Deliver AI-powered Consumer Experiences

Updated: May 29 2024 21:37
Arm CSS for Client is a purpose-built compute platform designed to deliver a step-change in performance, efficiency, and scalability across a wide range of consumer devices. Building on the success of Arm's Total Compute solutions, CSS for Client includes:

• The latest Armv9.2 Cortex CPU cluster
• Arm Immortalis and Mali GPUs
• CoreLink Interconnect system IP
• Production-ready physical implementations for CPUs and GPUs on the 3nm process

This powerful combination of technologies provides the fastest path to production silicon for Arm's partners, enabling them to unlock the full potential of the leading-edge 3nm process while maintaining the flexibility to create highly customizable silicon designs.



At the heart of CSS for Client is the latest Armv9.2 CPU cluster, which integrates Arm's highest-performance Cortex-X925 CPU, most efficient Cortex-A725 CPU, and refreshed Cortex-A520 CPU. This trio delivers unprecedented performance and efficiency for AI and other real-world compute workloads.

• Cortex-X925 handles 'bursty' workloads like launching applications and web browsing
• Cortex-A725 provides sustained performance for common AI workloads and AAA gaming
• Cortex-A520's high efficiency is best for light media, idle, and background tasks

The system integration and expansion of CSS for Client are achieved through the latest CoreLink Interconnect. The integrated system-level cache (SLC) enables best system power efficiency by reducing DRAM bandwidth and accesses, while the System Memory Management Unit (SMMU) provides enhanced security through stage-2 translation to support virtualized security frameworks like the Android Virtualization Framework (AVF).

Flagship smartphones powered by Arm's v9 CPU technologies, such as the MediaTek Dimensity 9300-powered vivo X100 and X100 Pro, Samsung Galaxy S24, and Google Pixel 8, are leading the way in delivering unprecedented opportunities for AI innovation. As AI workloads continue to become more compute-intensive and complex, Arm is laying the foundation for next-generation AI with its latest Armv9.2 CPU cluster. The Arm Cortex-X925:

• Delivers 36% single-threaded (peak) performance improvements and 46% better AI performance compared to the previous generation Arm Cortex-X4 CPU
• Optimized 3nm implementation complemented by a premium subsystem and packaging enables more than 30% higher performance scores on next-generation consumer devices
• Improvements to the microarchitecture, including up to 3MB private L2 cache, provide enhanced configurability for CPU cluster implementations

Designed to power a wide range of consumer devices from flagship smartphones to smartwatches, the new lineup includes the Arm Immortalis-G925, Mali-G725, and Mali-G625 GPUs, each targeting different market segments and performance levels. The flagship Immortalis-G925 GPU is Arm's highest-performing and most efficient GPU to date, offering:

• 37% better performance (fps) compared to its predecessor, the Immortalis-G720
• 30% less power consumption when providing gaming performance on par with Immortalis-G720
• 46% average performance improvement in leading mobile games, such as Genshin Impact and Roblox

These performance gains are driven by Arm's commitment to addressing the evolving needs of developers and ecosystem partners, as they strive for greater gaming realism through:

• Increased scene geometry complexity
• Complex fragment shading techniques
• Improved ray tracing capabilities

In addition to gaming improvements, the new Arm GPUs also deliver significant AI performance uplifts:

• Immortalis-G925 provides 34% faster inference across AI and ML networks compared to Immortalis-G720
• 41% performance improvement in image processing tasks
• Nearly 30% improvement in super sampling tasks
• 50% performance improvement in natural language processing and speech to text

Arm is collaborating with ecosystem partners like Unity to bring int8 support into the Sentis ML framework, resulting in a 44% performance uplift and smaller memory footprint for improved ML-based mobile gaming experiences.

Arm is also working with Epic Games to enable Unreal Engine 5 desktop renderer on Android, ensuring desktop-quality rendering and graphics on mobile devices. Arm is partnering with Google and MediaTek on the Android Dynamic Performance Framework (ADPF) to optimize user experience and performance based on real-time thermal-state information of mobile devices.



CSS for Client is Arm's fastest platform for Android to date, with significant improvements across key benchmarks and general compute use cases compared to the TCS23 platform:

• 36% improvement in peak performance (Geekbench 6 single-core score)
• 33% faster application launch times on average
• 60% faster web browsing (Speedometer 2.1 browser benchmark)
• 30% peak graphics performance improvements on average

CSS for Client is the platform for AI-powered consumer device experiences. Earlier this year, Arm showed how large language models (LLMs) can run locally on Arm CPUs on mobile devices. With CSS for Client, LLMs will run even better on Arm CPUs with faster response times. The platform delivers a 42% faster time-to-first token when running the Llama 3 LLM and a 46% faster time-to-first token when running the Phi-3 LLM.


Moreover, CSS for Client achieves a significant performance leap for AI inference across a broad range of general AI networks due to advances in the new Arm CPUs and GPUs:

• 59% faster inference on Cortex-X925
• 36% faster AI inference on Immortalis-G925
• 2.7x performance uplift in AI inference across 17 popular networks for int8 and fp16 data types

These improvements enable seamless user experiences across various AI use cases, such as computational photography and AI camera. CSS for Client achieves a 24% increase in bokeh performance compared to TCS23, allowing users to enjoy faster and smoother bokeh effects on their photos and videos without compromising battery life.



Arm is committed to enabling AI for everybody, and CSS for Client scales across a broad range of consumer devices and form factors:

• Next-generation AI PCs: Cortex-X925 delivers 50% more TOPS compared to Cortex-X4
• Mass-market consumer technology segments: Cortex-A725 acts as the main workhorse and developer target for AI processing
• Area-optimized deployments: Cortex-A725 allows efficient deployments of generative AI workloads across various consumer technology segments

CSS for Client is the purpose-built platform for the next generation of AI experiences across a broad spectrum of consumer devices. It allows Arm's ecosystem to do more, whether it's unleashing more performance, more AI, more application experiences, or more advanced silicon.

As consumer demand for technology continues to grow, with users expecting more advanced experiences, Arm's new Armv9 CPUs are set to elevate these experiences through advanced compute capabilities. From faster web browsing and applications to enhanced AAA gaming and generative AI workloads, the new Armv9 CPUs are defining the future of consumer technology.