Proposed in 2021Armv9 instruction set architecture, and after being used in core architecture designs including Neoverse and Cortex, Arm has now announced that it will bring the new instruction set architecture to the edge computing platform to promote new performance improvements in the Internet of Things.
The Cortex-A320 CPU, introduced this time, will leverage the Armv9 instruction set architecture to improve the computing performance, security, and efficiency of IoT devices, thereby driving development in areas such as industrial automation and smart cameras. It also integrates the Ethos-U85 NPU design, which supports the Transformer computing subnetwork, enabling IoT devices to execute AI models with over 10 billion parameters on the terminal.
The Cortex-A320 CPU design boasts approximately 35 times the deep learning performance of the Cortex-A10 CPU, while scalar computing performance is increased by 30%. This enables IoT devices, previously constrained by cost and power, to utilize more advanced artificial intelligence applications.
The Cortex-A320 CPU also introduces advanced security features such as parameter authentication (PAC), branch target identification (BTI) and memory tag extension (MTE), allowing IoT devices to more securely process sensitive data in different fields.
In addition, Arm has also extended its Kleidi artificial intelligence framework to IoT devices, enabling them to improve performance by up to 70%. At the same time, it will also enable more than 2000 million developers worldwide to simplify the development process of edge AI applications. In addition to importing Meta's Llama 3 or Microsoft's Phi-3 artificial intelligence models into IoT devices, developers can also quickly and flexibly adjust related designs according to deployment requirements, enabling IoT devices to be applied to required scenarios more quickly.
The Cortex-A320 CPU and other designs proposed this time have been adopted and implemented by companies such as Advantech, AWS, Eurotech, Renesas Electronics, and Siemens. Arm also emphasizes that as artificial intelligence technology gradually moves towards edge deployment, edge computing platforms combined with the Armv9 instruction set architecture will become a catalyst for future IoT innovation and development. Combined with more complete software and AI technology integration, it will drive the growth of larger IoT applications.
