The 38-member EEMBC, best known for its CoreMark processor and ULP (ultra-low power) microcontroller benchmarks is now heading up towards the system with working groups addressing entire edge-node power consumption, including RF communications, and another on security.
Both benchmarking exercises are expected to build on the energy measurement platform and profile approach used for ULPBench.
However, because of the complexity, the output from these working groups is likely to be more than benchmarking to aid purchase decisions, two working-group co-chairs told EE Times Europe. It could also include information on system decomposition and partitioning, the exposition of numerous use-cases, design guidelines, examples of best practise and Wikipedia-style continuously updated web pages.
Quantifying edge-node complexity
The IoT edge-node power consumption working group is the most advanced and has developed a physical test-rig to benchmark the combination of microcontroller plus RF communications.
The benchmark is intended to be communications protocol agnostic. Brent Wilson, a director of applications engineering, at Silicon Laboratories Inc. (Austin, Texas), who serves as co-chair of the working group, said "We're looking at things like battery-powered edge-node performance whether that includes ZigBee, Bluetooth, Thread." He added that LoRa "is in the field of view" as is SigFox's proprietary long-range communications for IoT. He added that partly it is a matter of resources and expertise.
Part of the approach is to deconstruct a use case activity – such as the operation of a ZigBee light switch – into a number of sub-tasks that contribute to power consumption. So a sub-task might be the energy cost of sending two bytes of data over the communications channel likely to be Bluetooth Low Energy (BLE). These sub-tasks can then be individually measured and added together to produce a total benchmark. This at the very least is expected to shed light on where energy inefficiencies may lie. The working group has started by developing a list of prominent use cases: these are the wireless