As you read this, chances are that a product based on ARM Ltd.’s microchip line is within arm’s reach, perhaps in the smartphone in your pocket, or the TV on your wall or the tablet on your lap. The ARM processor is one of the most universal technologies in the world. An estimated 86 billion total chips shipped last year, according to ARM.
ARM’s latest innovation could make another technology equally pervasive: artificial intelligence.
ARM recently introduced the Cortex-A75 and A55, the first processors from the company to use its DynamIQ technology that is designed to accelerate performance, allowing the microprocessor to better perform AI tasks such as machine learning. ARM said the new processor can boost the performance of these tasks by as much as a factor of 50. Machine learning is a branch of AI that involves equipping computers with the capability to gain knowledge without human intervention.
Chip in demand
Imagine software that could detect new types of malware or security cameras that could detect dangerous or criminal activity by themselves. That’s just a few examples of what machine learning can do.
With all the practical uses for artificial intelligence, demand for chips that can conduct AI functions is growing. Global shipments of chipsets supporting a form of machine learning called deep learning are expected to rise by a factor of nearly 50 from 2016 to 2025, according to the market-research firm Tractica.
In the future
Since ARM processors are already at the heart of nearly all the world’s smartphones, including Apple and Android models, the AI capabilities of the A-75 could have a major impact on the mobile market.
“The big story is you are getting a more powerful generation of chips that will be used in cell phones and other applications that will be more powerful and deliver incrementally better performance when running AI, just like anything else,” said David Kanter, analyst and editor with the Microprocessor Report, a publication owned by the market research firm The Linley Group.
In addition to mobile devices, Kanter predicted the A-75 would be used in other applications where integration is more important than top performance, such as routers. He anticipated that the A55 would be used in low-end embedded systems, such as sleep tracking devices.
Many companies are already moving to enhance their smartphones and tablets with AI capabilities. For example, Apple may be adding facial-recognition user-authentication capabilities to its upcoming next-generation iPhone, according to a market expert. The Bixby digital assistant technology in Samsung’s new Galaxy S8 smartphones is capable of recognizing images and providing information to users about real-world objects that are nearby.