Deploying AI Models at the Edge

Slate of ARM chip designs could make phones bigger players on the edge

Slate of ARM chip designs could make phones bigger players on the edge

Chip designer Arm Holdings Plc. has introduced its latest mobile chip design as well as a program through which buyers can tweak the performance of CPUs based on the Cortex-A design. Chips based on all of the new designs are expected to make phones more formidable edge devices.

In announcing the new mobile products, UK-based Arm said the convergence of marketable artificial intelligence and 5G telecommunications deployments mean phones need to be more powerful and efficient.

Arm’s new product designs are the Cortex-A78 CPU, Mali-G78 GPU, Mali-G68 and Ethos-N78 NPU.

Also debuting via the Cortex performance-customization program is the Cortex-XI CPU, which will have a 30% increase in peak performance over the Cortex-A77. The XI is ARM’s most powerful CPU and the first of its CPU designs in the performance customization program.

The company claims that the Cortex A-78 design delivers a 20% sustained performance increase over its A-77 mobile predecessor. An A78 chip will enable “multi-day immersive 5G performance… with greater on-device machine learning performance.”

The Mali-G78 GPU, based on the recently introduced Valhall architecture, will perform 25% better than the preceding G77. It can support 24 cores, and its architecture will extend phone battery life, according to ARM. Designs for the Mali-G68, which supports up to six cores, also was announced.

The Ethos-N78 neural processor design is billed as delivering 25% greater efficiency over the N77 while providing deeper on-device machine learning capabilities. Configurations range from one TOP per second to 10 TOP per second.

In February, ARM introduced designs for its new Cortex-M and Ethos-U chips.

Arm’s new designs are central to its bid to remain a key player in the market for processors used in power-constrained edge devices like cell phones and tablets, while long time licensees like Nvidia are pushing Arm’s architecture into data center deployments with multi-core GPUs that are more power-efficient than x86-based chips.

Article Topics

 |   |   |   |   | 

Comments

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Featured Edge Computing Company

Edge Ecosystem Videos

Automating the Edge

“Automating

Deploying AI Models at the Edge

“Deploying

Latest News