As the premium mobile market demands high-performance visual experiences and intelligence, including AI, for increased SoC design complexity, Arm announced a new total compute solution for next-generation mobile, offering improved PPA compared to the previous generation.

Arm announced its Total Compute Solution 2023 (TCS23), a mobile computing platform, on the 30th. The TCS23 is said to offer cutting-edge premium solutions for smartphones.

The TCS23 is a new, fully integrated system designed and optimized for specific workloads, delivering a comprehensive suite of advanced IP solutions to meet the growing demands of mobile user experiences.

“TCS23 features our latest Immortalis GPU based on a new fifth-generation GPU architecture for the ultimate visual experience, supporting a wide range of applications including next-generation gaming and augmented reality,” said Hwang Sun-wook, president of Arm Korea. “Compared to the previous generation, we have reduced memory bandwidth usage by up to 40% while improving performance by 15%.”

With Arm now shipping 25 billion chips and building a broad ecosystem, CEO Hwang said, “The market need for increasingly complex computing requirements and efficient solutions in terms of performance per watt for smartphones, tablets, laptops, wearables, and smart home devices is growing.” He added, “Arm Total Compute takes a solution-centric, holistic approach to SoC design, pursuing overall system design and optimization.”

While Arm's announced products are slated to power next-generation flagship smartphones, Arm is looking further into the future. Arm has recently focused on its CPU and GPU roadmaps, announcing plans to invest heavily in key IPs such as the Krake GPU and Blackhawk CPU to deliver the compute and graphics performance its partners demand over the coming years.

“The mobile market is gradually expanding to $92.2 billion in gaming and $431 billion in mobile maps,” said Jeong Seong-hun, FAE director at Arm Korea. “Demand for intelligent functions has been increasing since the advent of generative AI technology, so the TCS23 was designed with mobile in mind.”

■ Highest quality mobile graphics with the latest Immortalis GPU

Last year, Arm delivered high-quality visual experiences on mobile with its latest flagship Immortalis-G715 GPU. This technology, achieved through a partnership with MediaTek via the TCS22-based Dimensity 9200, is now featured in top-of-the-line flagship smartphones from OPPO and vivo.

Arm announced today its latest GPUs, built on its latest fifth-generation GPU architecture. Arm's most efficient GPU architecture to date, the fifth-generation architecture redefines parts of the graphics pipeline to reduce memory bandwidth, enabling highly geometry-intensive games and real-time 3D applications, while delivering smoother gameplay on mobile and complex PC and console-like experiences.

The latest Immortalis-G720 is the highest performing and most efficient Arm GPU in the Arm Immortalis series. Compared to the previous generation, it provides 15% better performance and efficiency, and 40% better system-level efficiency, supporting a more immersive visual experience and high-quality graphics.

Along with the Immortalis-G720, the latest Arm Mali-G720 and Mali-G620 join Arm's world-leading GPU portfolio. With these latest Mali GPUs, Arm is committed to bringing premium graphics capabilities to a wider consumer device market, faster.

■ CPU Performance Leadership for Intelligent AI

As part of TCS23, Arm is enhancing CPU performance with its latest Armv9 Cortex CPU compute cluster. This cluster has delivered double-digit performance gains for three consecutive years, contributing to increased efficiency.

At the heart of this high-performance cluster is the latest Arm Cortex-X4, a fourth-generation Cortex-X core that pushes the performance limits of flagship smartphones. It's the fastest CPU Arm has ever designed, delivering a 15% performance boost over the Cortex-X3.

Additionally, this new power-efficient microarchitecture consumes 40% less power than the Cortex-X3 in the same process. These performance and efficiency improvements will not only elevate on-device experiences, such as UI responsiveness and application execution times, but also enable next-generation AI and ML-based applications.

This CPU performance leadership extends to Arm's new big and little cores, the Arm Cortex-A720 and Cortex-A520. The Cortex-A720 is the industry's leading CPU IP, serving as the core of the cluster and driving continuous performance improvements. Meanwhile, the Cortex-A520 is the highest-performing, most efficient Arm CPU core ever. Use cases like AAA gaming, all-day productivity, and background tasks will all benefit from the new CPU design's 20% improved power efficiency over the previous generation.

Arm announced that it has taken its long-standing partnership with TSMC to the next level with the industry-first tape-out of the Cortex-X4 on TSMC's N3E process for next-generation CPU designs. This ensures that the Arm ecosystem can maximize the so-called PPA benefits of Arm processor technology after it is taped out: power, performance, and area.

Completing the 2023 CPU cluster is the latest DynamIQ Shared Unit, the DSU-120, designed for multi-threaded use cases and supporting a wide range of devices, from wearables to smartphones and laptops. Arm's latest CPU cluster delivers performance and efficiency tailored to the needs and demands of its partners.

■ Providing software and security worldwide

With TCS23, Arm provides the capabilities and tools to enable millions of mobile developers developing on Arm to develop secure software.

As the AI processing power of smartphones doubles every two years, new intelligent experiences like generative AI are surprising the world. Arm is leading the way in enabling developers to leverage AI and machine learning (ML) workloads by supporting Arm IP with enhanced machine learning (ML) capabilities through Arm's open source software libraries.

Arm NN and Arm compute libraries are already used in Android Google apps with over 100 million active users, enabling developers to optimize the execution of machine learning workloads on Armv9 Cortex-A CPUs and Arm GPUs.

All modern CPUs offer 64-bit computing and Armv9 security innovations, protecting against increasingly sophisticated digital threats. Arm is successfully deploying the Arm Memory Tagging Extension (MTE), which eliminates memory safety bugs that account for 70% of all software vulnerabilities, across the mobile ecosystem with Armv9 generation CPUs.