Power Performance Benchmark
In Consultation with NASA and GCN
If a machine is faster but burns more power, does it cost more to perform
a specific activity? In other words, is it more effective to use a greater
amount of power for a short duration (faster system), or a smaller amount of
power for a longer duration (slower system)?
Alterion, in consultation with NASA and GCN, has developed a series of
Power Performance Benchmarks to address this question. This white paper
details the strategy, environment, and testing conditions of the Power
Performance Benchmarks.
Overview
Alterion, in consultation with NASA and GCN, has developed a series of
Power Performance Benchmarks. These benchmarks measure the amount of power a
computer consumes while executing small but significant and common tasks.
Test Environment
The basic measurement test bed consists of a WattsUp Pro power meter from
Electronic Educational Devices (www.doubleed.com),
the test computer system, and a monitoring system. The power cord of the test
system plugs directly into the WattsUp meter, which measures power consumption
while providing electricity to the test system. The monitoring system connects
to the WattsUp meter via serial cable, allowing it to record power consumption
measurements at a rate of once per second.
Testing Methodology
Individual test applications are executed on the test and monitoring systems.
The test system repeatedly executes small and well-defined tasks from the
Alterion ALP suite, while the monitoring system records power consumption. Each
test runs for a fixed duration. Afterwards, the total power consumed is divided
by the number of tasks completed. The results represent the power consumption
required to perform the test activities, measured in watt-seconds.
We test this way to answer a critical question: If a machine is faster but
burns more power, does it cost more to perform a specific activity? In other
words, is it more effective to use a greater amount of power for a short duration
(faster system), or a smaller amount of power for a longer duration (slower
system)?
The current Power Measurement Test Set consists of five subtests: No activity,
Excel chart drawing, Zip file compression, Photoshop filtering, and Application
loading. The first test measures the power consumed when the system is idle.
By most estimates, a system will be idle almost all of the time. The remaining
tests measure small and fast activities such as chart redrawing (5000 points,
which typically takes a second or two to redraw) and compressing 5 MB of data.
Each of these tests uses system components in a fashion that differs from the
other tests. For example, the Excel chart test makes greater use of the graphics
controller in combination with data access and CPU utilization, while the
compression test balances CPU utilization and memory access with disk I/O.
Power Performance Benchmark
This test suite is part of the continuing Alterion ALP benchmarking program.
Alterion (www.alterion.com) is widely
known for its work in systems integration support and testing efforts for U.S.
Government agencies such as NASA and the Departments of State, Justice, and
Defense. Alterion is based in Conshohocken, PA.
About Alterion
Alterion was founded in March, 2000 with the goal of becoming a premier
provider of IT Program IV&V Management and Support services.
Headquartered in suburban Philadelphia, Alterion has since been successfully
involved in numerous programs for corporate and government organizations.
Alterion provides a complete set of Independent Verification and Validation
services through its proprietary SharpThought™ methodology. As a
cost-effective method for assuring successful IT system acquisition and
deployment, SharpThought minimizes risk during all phases of an engagement by
identifying problems, and their corresponding corrective actions, when they
can be mitigated at the lowest cost to a program.
Through SharpThought Services™, Alterion delivers IV&V expertise
in areas such as Quality Management, Risk Mitigation, Requirements Analysis,
Process Improvement, Performance Testing, and IT Consulting. Alterion's
capabilities are further augmented by its rich network of industry partners,
enabling Alterion to provide comprehensive and distinct services.
Alterion falls under the classification of a Small Business, as defined by
the United States Government. For more information visit Alterion's website at
www.alterion.com.
For More Information Contact:
Matthew Shapiro at 610.832.9450
m_shapiro@alterion.com
|