Energy-efficient Artificial Intelligence system-on-chip runs
AI is used for a vast array of useful applications, such as predicting a machine’s lifetime through its vibrations, monitoring the cardiac activity of health patients and incorporating facial recognition capabilities into video surveillance, for example. But the downside to the technology is that it generally requires a lot of power and in most cases must be permanently connected to the cloud, which raises issues related to data protection, IT security and energy consumption.
Minimising power consumption
Now engineers at Swiss-based CSEM think they have been able to get around these issues. The new system-on-chip they have developed runs on tiny batteries, or small solar cells and executes Artificial Intelligence at the edge, i.e. locally on the chip, rather than in the cloud. In addition, the system is fully modular and can be tailored to any application where real-time signal and image processing is required - especially when sensitive data is involved.
The system works through an entirely new signal processing architecture that minimises the amount of power needed to function optimally. It consists of an Application-Specific Integrated Circuit (ASIC) with a RISC-V processor - also developed at CSEM - and two tightly-coupled machine learning accelerators.
A new generation of devices with processors
This innovation now opens the door to an entirely new generation of devices with processors that can be run independently for more than six months. It also sharply reduces the installation and maintenance costs for such devices and enables them to be used in places where it would be hard to change the battery.
The new device will be presented at the 2021 VLSI Circuits Symposium in Kyoto, Japan, this month. The prestigious event was first organised in 1981 with the hope of offering opportunities for the world’s top technologists to engage in an open exchange of ideas. Since then, it has been held annually and grown to be an important and valuable event for people who work in the Very Large Scale Integration (VLSI) chip design and other frontier technical areas.
