Abstract
Remote Performance Monitoring (RPM): Full-System Power, Energy, and Performance Profiling for Resource-Constrained Devices
by: Selim Gurun, Priya Nagpurkar, and Chandra Krintz
Abstract:
Understanding the full-system power and energy behavior
of real, resource-constrained, battery-powered devices is vital
to accurately characterize, model, and develop
effective techniques for extending battery life.
Unfortunately, extant approaches to measuring and characterizing
power and energy
consumption focus on high-end processors,
do not consider the complete device, employ inaccurate (program-only)
simulation, rely on inaccurate, course-grained battery level
data from the device, or employ expensive power measurement
tools that are difficult to share across research groups and students.
In this paper, we present RPM, a remote performance monitoring
system, that enables fine grained characterization of embedded
computers.
RPM consists of a tightly connected set of components which
(1) control lab equipment for power measurements and analysis, (2)
configure target system characteristics at run-time (such as CPU and
memory
bus speed), (3) collect target system data using on-board hardware
performance monitors (HPMs)
and (4) provide a remote access interface.
Users of RPM can submit and configure experiments that execute
programs on the RPM target device (currently a Stargate
sensor platform that is very similar to an HP iPAQ)
to collect very accurate power, energy, and CPU performance data
with high resolution.
We use RPM to investigate whether CPU-based performance data
in the form of HPM metrics or program phase behavior correlates
well with full-system energy or power behavior. Prior work
shows that both accurately estimate processor power consumption for
high-end CPUs.
In resource-constrained devices, such as the one we study, however,
the processor consumes a much smaller portion of the total power in the
system than for high-end processors. Our experimentation
with RPM for the Stargate and set of embedded system benchmarks, show
that CPU-based metrics do not correlate well with full-system energy
and power consumption. Moreover, we find that full-system energy and
power varies significantly with the type of memory device and file
system.
Keywords:
energy and power estimation, resource-constrained devices, remote monitoring
Date:
February 2006
Document: 2006-02
