Thoughts? - Establishing Long-Term Effects of Android BOINC on Battery Life

[Atai Barkai]

Hello everybody,

As some of you may know, the long-term capacity of Lithium-ion batteries is greatly affected by the temperature of the battery.

For example, see below the table taken from:

http://batteryuniversity.com/learn/article/how_to_prolong_lithium_based_batteries

Temperature	40% charge	100% charge	Table 3: Estimated recoverable capacity when storing Li-ion for one year at various temperatures Elevated temperature hastens capacity loss. The capacity cannot be restored. Not all Li-ion systems behave the same.
0°C 25°C 40°C 60°C	98% 96% 85% 75%	94% 80% 65% 60% (after 3 months)

While the desire to keep uncomfortable statistics "out of sight and out of mind" is one that I too share, I think it's important to quantitatively establish the effect that the BOINC app has on the device's long-term battery life.

Since by default the app never lets the battery temperature go above 40.0C, and most users presumably only run it for 6-8 hours a day, I expect that the effect is not too detrimental.

Measuring the effect may alleviate some worries that prevent some people from installing the app.

As mentioned in the following link, the only way to establish this is by getting real-world feedback from the community.

https://boinc.berkeley.edu/trac/wiki/BOINConPhones

*** 

A first step would be measuring the battery capacity of existing BOINC users, and correlating it with the device type, as well as with age.

I expect that after getting the information from BOINC community members, comparing it with the general Android community wouldn't prove too difficult.

Of course, for starters, we must establish a test that would be a good measure of the effect, but would also be feasible to perform.

There are 2 ways we can go about this, as far as I can see... 

1) Ask members to manually measure this using a separate battery-benchmarking app that the community can agree upon

2) Add the battery benchmark test to the standard BOINC app / a beta version available only for testers (could be as simple as tracking battery depletion rate over a week or so).

***

Thoughts anyone?

Is anyone else curious about this?

[Eric J Korpela]

I can't say that I'm surprised.  Battery failures have seemed more common to me on my devices once BOINC was installed, but that could be my impression rather than reality.  Because of my impressions, I set a temperature maximum of 35C for running BOINC on my devices.  I will also never buy another device that doesn't have replaceable batteries for the same reason.

Another potential problem with your plan is that there is no control group using the same devices without BOINC.  Batteries degrade without BOINC, so with no comparison it's hard to say what the effect of BOINC is.

--
You received this message because you are subscribed to the Google Groups "BOINC Android Testing" group.
To unsubscribe from this group and stop receiving emails from it, send an email to boinc-android-te...@googlegroups.com.
To post to this group, send email to boinc-andr...@googlegroups.com.
Visit this group at http://groups.google.com/group/boinc-android-testing.
To view this discussion on the web visit https://groups.google.com/d/msgid/boinc-android-testing/b20fcd6a-7cf8-4255-83e7-bc8933290f62%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[David Anderson]

I forwarded this question to HTC, who have a BOINC-based app called Power to Give.
They presumably have a laboratory for studying battery life;
maybe they have some data.
-- David

[Atai Barkai]

Thanks for your reply, David.

It would be great if you could keep us updated regarding HTC's response.

In case the HTC-route does not bear results, not all hope is lost:

(This is also a reply to Eric's comment)

There are a number of smartphone battery benchmarks out there, and some of them publicly post their data online.

If we can get the BOINC community to also benchmark their devices, we can get a good idea of the effect of the BOINC app on a device's battery.

As I mentioned in my original post, this can either be achieved by including a battery benchmark (+ instructions) with the BOINC app, or by simply asking members to manually run a battery benchmark and share the results.

For example consider GeekBench, which have been including a battery benchmark with their Android app for some time.

You can see the results here:

 https://browser.primatelabs.com/battery3/top

It's going to be difficult to get exact results out of this data, as this database is not aware of the time in which the device was (1) manufactured and (2) purchased.

However this database does contain the device kind, which can give us some ballpark estimation of the aforementioned 2 parameters.

(If somebody is aware of a better database, please share).

Regarding the option of benchmarking the devices directly through the BOINC app, is this something that would have the developers' support?

Are the developers even reading this Google Group?

GeekBench mentions on their website that they sometimes make source code available for a price. Perhaps they'd be willing to freely share it due to the nature of BOINC.

Even if not, it shouldn't be too difficult to come up with a separate battery benchmark with results that we can learn to "translate" into the GeekBench language, allowing us to use their database.

If this proves too difficult, we can consider other ways to get members to manually download and run the GeekBench app, and post links to their results.

I'd be happy to personally see this through, both on the coding side of things and on the coordination side of things, if whoever runs the BOINC app wishes it.

-Atai

To unsubscribe from this group and stop receiving emails from it, send an email to boinc-android-testing+unsub...@googlegroups.com.