We have developed PowerMon and PowerMon2, low-cost power monitoring devices that operate inside commodity computer systems, to analyze performance and power consumption tradeoffs in computer applications. Inserted between a system's power supply and motherboard, PowerMon monitors voltage and current on six DC rails and reports measurements at a rate of up to fifty samples per second through a USB interface, allowing monitoring by the target or a separate host. PowerMon2 has a smaller form factor that fits in a standard 3.5" hard drive bay, allowing it to be used in a 1U server chassis. It features a faster measurement rate of up to 1024 Hz on a single channel or 3072 Hz divided among multiple channels. PowerMon2 also adds two measurement channels for additional peripherals, such as disks and graphical processing units. The PowerMon devices have been used to resolve and highlight variations in power consumption and energy efficiency at the subsystem level during separate operating phases of scientific performance benchmarks, such as NAS BT and SP. The device parts cost is low enough to provision an entire cluster for power monitoring and adaptation. Complete schematics, board layouts, and source code for the PowerMon devices are available under a BSD-style open source license at ilab.renci.org/powermon.
[1]
Margaret Martonosi,et al.
Runtime power monitoring in high-end processors: methodology and empirical data
,
2003,
Proceedings. 36th Annual IEEE/ACM International Symposium on Microarchitecture, 2003. MICRO-36..
[2]
Ying-Wen Bai,et al.
Estimation by Software for the Power Consumption of Streaming-Media Servers
,
2007,
IEEE Transactions on Instrumentation and Measurement.
[3]
C. Calwell,et al.
80 plus: a strategy for reducing the inherent environmental impacts of computers
,
2005,
Proceedings of the 2005 IEEE International Symposium on Electronics and the Environment, 2005..
[4]
Rong Ge,et al.
Power and energy profiling of scientific applications on distributed systems
,
2005,
19th IEEE International Parallel and Distributed Processing Symposium.
[5]
R. Fowler,et al.
Powermon 2: Fine-grained, Integrated Power Measurement
,
2010
.