Power Baseline

  This page is monitored by the OLPC team.

Goal

The purpose of this set of measurements was to attempt to establish a metric for a power measurement that could be used by QA to determine if a given build used more or less power while in the idle state.

Conclusions (Still Ongoing)

• For the original definition of idle it is not feasible to have a single run test that compare against a baseline. The variance is just too large and heavily dependent on the WLAN activity during the test. Achieving a useful number would require the average of many, many runs.
• Measurements with the WLAN disabled had variance that was much less but still showed a machine to machine variation of x