ARM today announced DynamIQ, a new way to manage multi-core systems that gives them the ability to scale exactly to the tasks or needs at hand. DynamIQ expands on ARM's big.LITTLE technology, which allows for paired sets of high-power and low-power processor cores to tackle specific tasks based on the computing requirements. The big change is that DynamIQ doesn't require paired sets; instead, it can handle one high-power processor with seven low-power processors and other combinations to give device makers more granular control over how the cores handle computing operations. ARM says this is particularly helpful for machine learning and artificial intelligence. The company claims DynamIQ will deliver a 50x boost in on-device AI performance over the next few years when compared to what's available today. The fact that DynamIQ can handle AI tasks directly on mobile devices -- rather than in the cloud -- means quicker response times and better security for end users. Other benefits of DynamIQ include a redesigned memory subsystem that's faster and more power efficient; refined control over thermal budgets; and faster software switching between processor cores. ARM sees DynamIQ making a big impact on smartphones, but it also intends to bring DynamIQ to robots, cars, and other autonomous devices. ARM plans to debut new processor cores specifically for DynamIQ later this year, but the company didn't say when it expects DynamIQ-enabled devices to reach the market.


More...