Student research opportunities
Energy measurement framework for PandaBoard
Project Code: CECS_852
This project is available at the following levels:
CS single semester, Honours, Summer Scholar
Please note that this project is only for undergraduate students.
Keywords:
energy hardware measurement performance microarchitecture
Supervisor:
Professor Steve BlackburnOutline:
Energy efficiency is very important for today's computing system,
especially mobile systems. Quantitatively analyzing the power
consumption in fine granularity is the first step to optimizing
energy efficiency. This project is about building an energy
measurement/analysis framework for PandaBoard.
The PandaBoard is a low-power low-cost single-board computer board
based on OMAP4 SoC. As shown here, there are independent power rails
to provide power to OMAP4 SoC. For example, VCORE1 powers OMAP4's
Cortex-A9 subsystem. The goal of this project is to understand how
software affects the energy efficiency by measuring the energy each of
power rails consumed.
Goals of this project
The first part of this project is to build the hardware kit to measure
voltage and current on power rails. We can use either shunt registor
or hall-effect sensor to measure the current. For the voltage, we
could use power meter to measure and verify whether it is constant. We
can use similar approach in our prior work to sample the current.
The second part of this project is to investigate the relations
between the energy consumption and the software. For example, we can
measure how much energy could be saved by putting cores in deep sleep
state.
The work will be co-advised by Xi Yang and Ting Cao.
Links
Looking back and looking forward: power, performance, and upheavalThe Yin and Yang of Power and Performance for Asymmetric Hardware and Managed Software
