Search results for

All search results
Best daily deals

Affiliate links on Android Authority may earn us a commission. Learn more.

ARM announces its next-gen CoreLink system for Heterogeneous Computing

ARM has announced its new CoreLink CCI-550 interconnect and DMC-500 memory controller, promising bandwidth improvements for next-gen mobile devices

Published onOctober 27, 2015


Today, ARM has announced its new CoreLink system IP, promising performance improvements for the next generation of mobile devices. The newly announced CoreLink CCI-550 interconnect enables ARM big.LITTLE processing with a fully coherent GPU, while the new DMC-500 memory controller provides higher bandwidth and improved latency response for processors and displays.

CoreLink CCI area scaling
In terms of performance, ARM’s new CCI-550 boasts a 60 percent peak bandwidth increase and a 20 percent reduction in latency. ARM also states that this interconnect can save 100’s of mW in power due to its integrated snoop filter. The CCI-550 can scale between 1 and 6 ACE and memory interfaces to save on area, making it suitable for a wide range of applications, from mobile to networking.

This announcement is also a notable step forward for heterogeneous compute implementations, as communication between various processing components can be limited by a lack of bandwidth. Faster connected system will enable processors to work on the same data without unnecessary cache maintenance or memory copying. For mobile, applications could range from deep learning to augmented reality.

“To provide advanced features such as 4K video recording/playback, 120fps cameras, and quad-HD displays, they must integrate heterogeneous CPUs, GPUs, and accelerators into a cache-coherent system while keeping within tight power budgets.” – Mike Demler, senior analyst, The Linley Group

The new DMC-500 dynamic memory controller brings a 27 percent increase in memory bandwidth utilization, a 25 percent reduction in average CPU latency, and support for up to LPDDR4-4267 RAM. Increased memory bandwidth is particularly important as consumers demand higher resolution content on their mobile devices.

ARM Premium Mobile Application with Fully Coherent GPU

There’s also a little mention of the next generation of ARM Mali GPU technology, which will be called Mimir. One slide states that Mimir will be a fully coherent GPU with 1 to 4 ACE and Shared Virtual Memory, but we don’t have any additional details about this architecture just yet.

Production silicon sporting these latest ARM technologies is expected by late 2016, so it might not appear in consumer products until early or mid-2017.

You might like