Arm enables SoC makers to create custom instructions for embedded CPUs

The new Customs Instructions feature should open up the door for a greater diversity of devices, Arm says.

Arm on Tuesday announced it's opening the doors for SoC (system on chip) designers to add unique, application-specific features to Cortex-M CPUs, without risking software fragmentation. The new Custom Instructions feature for the Armv8-M architecture should make it easier to co-design hardware and software for specific embedded and IoT use cases.

"A world of a trillion secure intelligent devices will be built on a diversity of complex use cases requiring increased synergy between hardware and software design," Dipti Vachani, SVP and GM of Arm's Automotive and IoT Line of Business, said in a statement. 

Arm will offer Custom Instructions as a standard feature in Cortex-M CPUs, starting with Arm Cortex-M33 CPUs in the first half of 2020, at no additional cost to new or existing licensees. 

To accommodate Custom Instructions, Arm is modifying the CPU to reserve encoding space where designers can add custom datapath extensions. Along with the existing co-processor interface, the new feature will enable Cortex-M CPUs to be extended with accelerators optimized for edge compute use cases, including machine learning ML and AI.

Meanwhile, in effort to strengthen its IoT partner ecosystem, Arm also announced on Tuesday that it's changing the partner governance model for Mbed OS, its free, open-source IoT operating system. Over 10 years, more than 425,000 third-party software developers have used Mbed OS. 

Arm is shifting the governance model to allow developers to have more direct influence over Mbed OS features and functionality. To support this new model, Arm is launching monthly Product Working Group meetings where the company will work with silicon partners to prioritize and vote on which new capabilities to add to the operating system. Several Arm silicon partners are already participating in the working group, including Analog Devices, Cypress, Maxim Integrated, Nuvoton, NXP, Renesas, Realtek, Samsung, Silicon Labs and u-blox .

The working group's partners, Arm said, have already contributed to the development of new low-power battery optimization to extend the battery life of devices in the field.