Right now, the perf lab machines run on Windows 10 version "Threshold 1" (10240), which was released on July 29, 2015.

We believe that a certain bug causing power noise in our benchmarks could be resolved if we updated these machines to a newer version. Right now, the newest version available in Google is "Redstone 1" (14393) released on August 2, 2016.

In order to update these machines, we need to evaluate whether doing so will have a significant impact on any benchmarks that we run. This is a similar problem to the one that we had when updating our Android bots from non-swarming to swarming (https://bugs.chromium.org/p/chromium/issues/detail?id=705134), so I plan to use a similar process.

I'm going to compare the performance of the Windows High-DPI FYI bot before and after the update to Redstone 1 across a few benchmarks:

- system_health.memory_desktop (swarming@  http://crbug.com/705135 )
- speedometer (swarming@  http://crbug.com/705138 )
- memory.desktop (swarming@  http://crbug.com/705136 )
- smoothness.top_25_smooth (swarming@  http://crbug.com/705139 )
- loading.desktop (swarming@  http://crbug.com/705140 )
- thread_times.key_silk_cases (swarming@  http://crbug.com/714750 )