The latest chip from Maxim Integrated provides support for AI at the edge on battery-powered IoT devices.
Maxim Integrated’s new MAX78000 chip promises support for executing AI inferences at less than 1/100th the energy of software solutions. Such power-efficiency is vital for IoT devices which may have to last years before their batteries are replaced.
Kris Ardis, Executive Director for the Micros, Security, and Software Business Unit at Maxim Integrated, said:
“We’ve cut the power cord for AI at the edge. Battery-powered IoT devices can now do much more than just simple keyword spotting.
We’ve changed the game in the typical power, latency, and cost tradeoff, and we’re excited to see a new universe of applications that this innovative technology enables.”
While the MAX78000’s power consumption is minimal; there’s been no compromise in performance. MAX78000 executes inferences 100x faster than software solutions running on low-power microcontrollers.
In the past, devices have had to gather data from sensors and send that information back to the cloud to execute an inference. The answer would then often need to be sent back to the device. This is a slow and energy-intensive process which is unfeasible for many edge applications.
Microcontrollers like the MAX78000 integrate a dedicated neural network accelerator with a pair of microcontroller cores to enable local AI processing in real-time. This means applications such as machine vision, audio, and facial recognition can be run efficiently at the edge.
Kelson Astley, Research Analyst at Omdia, commented:
“Artificial intelligence is frequently associated with big data cloud-based solutions.
Anything that can cut the power cord and reliance on big Lithium-Ion battery packs will help developers build AI solutions that are nimbler and more responsive to environmental conditions in which they operate.”
Specialised hardware at the heart of the MAX78000 is designed to run convolutional neural networks (CNN) with minimal intervention from any microcontroller core. The cores are only engaged where required for mathematical operations that implement a CNN to maintain efficiency.
Cost is always a key consideration in what could be a large-scale IoT deployment. Maxim Integrated promises the microcontroller is just a fraction of the cost of FPGA or GPU solutions previously required to achieve decent inference performance.
Interested in hearing industry leaders discuss subjects like this? Attend the co-located 5G Expo, IoT Tech Expo, Blockchain Expo, AI & Big Data Expo, and Cyber Security & Cloud Expo World Series with upcoming events in Silicon Valley, London, and Amsterdam.