Arm, a processor design company, has announced Armv9, the first new Arm architecture in a decade.
The new development aims to respond to market needs. Higher performance, greater security and dedicated DSP, machine learning and artificial intelligence features for ubiquitous computing. The company also highlighted how the new design builds on the success of the current Armv8; which already marks a state-of-the-art level of performance-per-watt, in every sector and area of computing.
The number of Arm-based chips deployed continues to grow, with more than 100 billion devices shipped in the past five years; 100 percent of the data shared around the world will soon be processed on Arm; whether it is on the endpoint, on data networks or in the cloud.
This pervasiveness also comes with Arm’s greater commitment to provide more security and better performance, as well as more new features.
The new features in Armv9 will accelerate the transition from general-purpose to more specialized computing in each application; at a time when artificial intelligence, the Internet of Things (IoT) and 5G are gaining momentum globally.
Armv9 faces today’s biggest technological challenge
The biggest technological challenge today is to keep data safe. The Armv9 roadmap introduces the Arm Confidential Compute Architecture (CCA). Confidential computing protects portions of code and data from being accessed or modified while they are in use; even from software that has privileges, performing the calculation in a secure hardware-based environment.
The Arm CCA will introduce the concept of Realm dynamically created and usable by all applications; in an area that is separated from both the safe and the unsafe world. For example, in business applications, realms can protect commercially sensitive data and codes from the rest of the system; either while they are in use, or not. Furthermore, the ubiquity and range of artificial intelligence workloads require more diverse and specialized solutions.
According to Arm, it is estimated that there will be more than eight billion AI-assisted devices in use by the mid-2020s. 90% or more of on-device applications will contain elements of AI along with AI-based interfaces. for example the voice).
To address this growing need, Arm partnered with Fujitsu to create Scalable Vector Extension (SVE) technology. It is the heart of Fugaku, the fastest supercomputer in the world. Building on that work, Arm developed SVE2 for Armv9. This allows for better machine learning and increases digital signal processing (DSP) capabilities in a wider range of applications.
SVE2 improves the processing capacity of 5G systems; virtual and augmented reality; machine learning workloads that run locally on CPUs, such as image processing and smart home applications.
In the coming years, Arm will further extend the AI capabilities of its technology. Through substantial improvements in the multiplication of matrices within the CPU; in addition to the innovations underway in the Mali GPUs and in the Ethos NPUs.
Armv9 will also boost CPU performance. However, Arm said, as the industry moves from general-purpose computing to specialized ubiquitous computing; the two-digit annual increases in CPU performance are not enough.
In addition to improving specialized processing, Arm’s Total Compute design methodology will accelerate overall computing performance; through system-wide hardware and software optimizations and increases in use case performance. The technologies at the Armv9 system level will embrace the whole range as well as improve individual areas. By applying Total Compute design principles across the entire IP portfolio of automotive, client, infrastructure and IoT solutions. In addition, Arm is developing several technologies to increase the frequency; bandwidth and cache size; reduce memory latency to maximize Armv9-based CPU performance.