ARM moves beyond mobile with DynamIQ

With a new multi-core microarchitecture that will serve as foundation for all new Cortex-A processors, ARM plans to be everywhere, from the cloud to cars.


Building on a decade of innovation, ARM is unveiling its next step in multi-core technology with DynamIQ -- a multi-core microarchitecture that will serve as foundation for all new Cortex-A processors.

While earlier ARM processing system technology targeted traditional markets like cell phones, DynamIQ's flexibility and versatility allows ARM to support intelligent systems just about everywhere, from the cloud to the network and edge devices.

"As systems get more complex...we need to redefine how multi-processing works," Nandan Nayampally, general manager of the ARM Compute Products Group, told reporters this week. That means addressing "flexibility within the CPU processing itself... but also how to enhance and integrate heterogeneous compute closer to enable very diverse, differentiated solutions."

ARM has yet to announce its new Cortex-A processors, but the Cambridge company says "multiple partners" are already working with DynamIQ technology. Those partners will be announced when the new processors are launched later this year, while products based on this technology should be available sometime in 2018.

DynamIQ builds on ARM's big.LITTLE technology, which couples a low-powered chip with a high-powered one in order to extend battery life on phones and tablets without sacrificing processing power.

Advancing the "right processor for the right task" approach, DynamIQ supports up to eight cores in a single cluster. With big.LITTLE, each cluster was targeted for a particular performance level. With DynamIQ, each core can have different performance and power characteristics, enabling faster responsiveness to machine learning and AI applications. Adding to its flexibility is a redesigned memory subsystem, enabling both faster data access and enhanced power management. DynamIQ also offers more efficiency when switching between power states and fine-grained speed control.

"Not only can you tune the clusters, you can tune individual processors and actually trade off the right performance for the right power and energy characteristics," Nayampally said.

This will help ARM move into markets beyond mobile, Nayampally explained, particularly as the demand for compute speeds up. ARM has already seen its business accelerate: Over the past four years, ARM partners shipped 50 billion chips, the company said, and it expects to ship twice that many in the next five years.

The technology should be useful for the evolving smartphone market, given the priorities around heterogeneous computing and device-based AI on mobile devices. DynamIQ also represents the sort of ARM ISA enhancements referenced by Microsoft recently in a blog post about the future of cloud infrastructure. DynamIQ also brings greater levels of responsiveness for advanced driver assistance systems (ADAS) and increased safety capabilities. Meanwhile, DynamIQ's scalability should also prove useful for networking environments.

For every industry ARM expects DynamIQ to touch, artificial intelligence is expected to play a key role. With that in mind, Cortex-A processors designed for DynamIQ technology can be optimized to deliver up to a 50x boost in AI performance over the next three to five years, ARM says, as well as up to 10x faster communications with tightly coupled on-chip hardware accelerators.