When you hear the word "microcontroller," do you immediately think of Arm? Many Elektor community members do, which is why we decided to catch up with Arm at Embedded World 2022. In this interview, Elektor editor chats with Mohamed Awad, who explains how Arm is supporting microcontroller vendors with virtual hardware prototyping and more. 
 

Arm on Microcontrollers and Virtual Prototyping

"When it comes to microcontrollers, proprietary cores have fallen largely out of favor and been replaced by technology from Arm, and we're here today to find out how Arm is supporting the semiconductor vendors bringing their microcontrollers to market faster through virtual hardware prototyping and support," Stuart Cording mentioned in the introduction. Cording chats with Awad about Arm virtual hardware, as well as topics such as machine learning, Project Cassini, and Project Centauri.
  
When it comes to virtual hardware, Arm puts it in the cloud, Awad explained. "We deliver it directly to software developers. So, we make it available at the same time we make the IP available to the semiconductor guys. We make the virtual models available to the broad ecosystem, so software developers can literally start developing from the convenience of their homes without having to worry about wires and jumpers and cables or the availability of silicon."
 
Elektor Arm EW2022

How Arm is Tackling Machine Learning

If you think about machine learning, I would venture to guess that more machine learning runs on more Arm processors than anything else, just starting with the Cortex-M family, Awad noted during the discussion.

"To talk about machine learning specifically, we actually have two recent Corstone designs which we've released," he added. "One is the Corstone-300 that you pointed out. It's got a Cortex-M55 and then our Ethos-U55 NPU on it. And then we also have a Corstone-310, which has our newest, fastest, most secure Cortex ever, the M85, along with the U55. What we do here is in both of those cases, whether it's the Cortex-M55 or the Cortex-M85, we actually have machine learning acceleration built right into the CPU core. But beyond that, we've actually added the optional U55 next to it for those next-level applications."

Watch the full conversation.