GemStone is a suite of five software tools that, together, identify and evaluate the sources of error in gem5 models against a reference hardware (HW) platform by utilising statistical and machine learning approaches. Independently, the tools serve many purposes: Automating the running of experiments and collection of Performance Monitoring Counters (and temperature and power on supported platforms) on Arm-based devices (ARMv7 and ARMv8); Automating the running of gem5 simulations and aggregating the results; Applying Powmon power models to the hardware results; Applying Powmon power models to gem5 simulation data; Combining the HW and gem5 model characterisations and applying statistical and…

Being able to accurately estimate CPU power consumption is a key requirement for both controlling online CPU energy-saving techniques and design-space exploration. Models built and validated using measured data from an actual device are extremely valuable as their accuracy is known and trusted. PowMon produces power models that are both accurate and stable. The open source PowMon software tools consists of a number of distinct parts: Software tools which allow researchers to implement our automated model building methodology, and produce power models of a platform that are both accurate and stable. Downloadable power models for existing mobile CPUs (quad-core Cortex-A7 and quad-core…