Battery usage on tablets is too high

[dmurfet]

I currently recommend to people giving a metauni lecture on a metaboard that they have their tablet plugged into the power, because battery consumption is so high. This is annoying, because a natural vision for metauni is that you can listen to or deliver a lecture anywhere just with your tablet, stylus and headphones. E.g. find a nice quiet spot in the park.

But not if I'm terrified of my battery running out half way through the talk.

Besides turning down the graphics quality, are there obvious things that we should be doing to reduce battery consumption on mobile devices?

[dmurfet]

A very preliminary experiment with iPad Pro 12.9inch 2nd gen at roughly 2/3 brightness and at lowest graphics quality settings.

• 11:30 running TRS, using boards drained the battery 21% --> 17% (a drop of 4%).
• 11:30 running a baseplate and not moving 17% --> 14% (a drop of 3%).

Given the lackadaisical nature of this experiment the difference between 4% and 3% seems within the margin of error for me, so at least for the moment it seems like battery usage with metaboard is consistent with battery usage due to Roblox in general and is not particularly bad.

[dmurfet]

The battery issues on iPads appear to be completely resolved by simply setting the Graphics Quality to minimum.

My tests in yesterday's event with iPad Pro 12.9inch 2nd gen starting at 100% battery:

• Euclid session 100% --> 92% about 20min writing and 10min loitering (game running, looking at a board without any activity). So about 0.4% per minute of battery usage.
• SLT session 92% -> 64% about 60min of writing. About 0.47% per minute of battery usage.

Let's round up and say 0.5% per minute while using the iPad in metauni. That's an expected battery life of ~3.3hours. This is acceptable for now, relative to the catastrophic situation before with higher graphics quality where you couldn't be sure you would get through a single talk. We can aim for better though (the iPad is rated for 10hours of wireless web surfing).