Android L improves battery life by 36%

mercurial

mercurial

MyBB Legend
Joined
Jun 12, 2007
Messages
40,900
Reaction score
2,693
Location
/\/¯¯¯¯¯\/\
Android L improves battery life by 36% in Ars Technica’s benchmark

Android L is Google’s biggest update to its mobile operating system. We are expecting wonders when it comes to design, performance and new features, but it seems we didn’t give battery life improvements too much attention. This is probably due to us simply assuming it wouldn’t really be a huge jump, but we may have underestimated Google’s work.

Google’s “Project” tradition started with Jelly Bean and Project Butter, which made the UI incredibly smooth. KitKat offered Project Svelte, which allowed lower-end devices to run the software more efficiently. Next up is Project Volta, which is much more than a small improvement in battery efficiency.

The test

Ars Technica’s latest tests prove the new software boosts battery life by about 36%! This is no normal battery test, either. Ars Technica went all out and was as precise as possible. Both software versions were tested with the same Nexus 5 smartphone, managing to avoid battery variations. Screen brightness was set to 200 cd/m2, which was verified by a colorimeter.

The test consists of leaving the display on and loading a website every 15 seconds (over WiFi) until the device finally gives up and dies. This was ran on Android 4.4.4 first. Android L was then flashed to the same device and the test was applied again.

What is different about Android L?

Project Volta relies on different techniques to extend battery life. For starters, the new battery-saving mode is said to add an extra 90 minutes of battery life, but this test didn’t quite rely on that.

Google discovered waking the device for a second takes away about 2 minutes of standby time. And this doesn’t necessarily involve turning the screen on, this happens every time an app wants to perform a background action and wakes the processor. The new JobScheduler API allows these tasks to be grouped together, instead of having them act individually.

It’s also important to note Android is now using ART instead of Dalvik, which is known for offering better power efficiency.

battery-life.jpg

Wrap up

This is only a glance at what Android L could offer. More apps need to be optimized to run well with Android L, meaning battery performance could get even better in the future! We are certainly more excited about Project Volta now that we have seen it in action.
Click to expand...

Source
 
On a Nexus 5, maybe. It's impressive but I wonder how big the real-life difference will be on other devices, if any.
 
TJ99 said:
On a Nexus 5, maybe. It's impressive but I wonder how big the real-life difference will be on other devices, if any.
Click to expand...

Agreed. They should run the same test on the Nexus 7, and compare the results. It would be a better indicator.
 
Maverick Jester said:
Agreed. They should run the same test on the Nexus 7, and compare the results. It would be a better indicator.
Click to expand...

Or something that's not a Nexus, remember all the crap the manufacturers add on also has an effect.
 
As someone who is running L on a Nexus 5 I beg to differ.
 
TJ99 said:
Or something that's not a Nexus, remember all the crap the manufacturers add on also has an effect.
Click to expand...

Definitely, but as of right now, only those two devices have the L preview available. At the very least they should collate all possible results, considering that by their own admission, the test was as scientific as possible.

I've copied the original article run in ArsTechnica for reference:

Examining Project Volta: We put Android L through our battery test

Do Google's power optimizations work in the real world?

For each Android release, Google likes to have a "Project" that picks an area of weakness and focuses on tuning the whole OS to fix it. In Jelly Bean we had Project Butter, a concerted effort to make Android's UI animations run at 60fps. In KitKat there was Project Svelte, which aimed to get the OS running on only 512MB of RAM. Now in Android L we have "Project Volta," an attempt to boost the battery life of Android devices. Like the projects that came before it, Volta is a wide-ranging effort that touches many different parts of the OS.

Google did some testing and found that waking a device up for one second burns two minutes of standby time. This isn't just turning the screen on—it also means an app waking the processor and modem to check for updates. To cut down on this battery burn, Android L has a new "JobScheduler" API that allows the OS to batch up unimportant app requests like database cleanup and log uploading. Developers can also request that housekeeping functions be delayed until the device is plugged in. Further, the OS will not wake the device up for a network task when there is no network connectivity.

Google also made an enhanced battery stat tracker called "Battery Historian." The tool added a ton of new power-tracking capabilities and allowed the company to visualize all that data in a pretty chart. Better analytics gave Google more insight into what was using all that power, which allowed it to tweak things all over the OS for better battery life.

There's the switch from Dalvik to ART, a faster virtual machine that compiles apps once instead of every time they are run. ART is also generally faster and more efficient than Dalvik, which means less time spent processing and less power usage.

That all sounds great, but does Project Volta actually work? We decided to find out by running our review battery test. Just like in our reviews, we were as scientific as possible. We did the test on a single device to remove variances in battery, which meant flashing to 4.4.4, signing in, updating apps, charging up, running the test, and then flashing the same device to the L preview. Our battery test keeps the screen on and automatically loads webpages over Wi-Fi every 15 seconds until the battery dies. For each run, the screen brightness was set to 200 cd/m2, as verified by a colorimeter, and for consistency, we averaged two runs each. The results?

View attachment 129319

Thirty-six percent more battery life. Installing the L preview on our Nexus 5 gave us two extra hours of runtime. These tests were done on a beat up, daily-driver phone, so the final numbers aren't necessarily indicative of what the L preview could achieve on a new device. It's the difference that matters.

Android L also has a "battery saver" feature that lowers the device performance and cuts the background data and screen brightness when the device hits 15-percent battery, but we disabled this feature for our tests. We like to have our devices at full power, and we don't want a hyper-aggressive phone-crippling feature messing with our results. Presumably, you could gain even more runtime by turning the battery saver on. This test is more about the general under-the-hood improvements that are running all the time.

This is just an early look at Android L's battery life. The current release is a "developer preview," but Google has already been able to make a meaningful difference in the runtime of L devices. Some of the new power-saving features, like the job scheduler, require app support, so there's probably room to improve not only in the OS but in Google's bundled apps, too. By the time the final version of L comes out, things could be even better. Google definitely looks to be on the right track.
Click to expand...
 
Wait till Samsung loads Splanner Smemeo Shealth Svoice SamsungApps Samsunghub S***** then come back with the results
 
This tells us more about the inefficiency of the last Android release than the efficiency of Android L.

It's still impressive though.
 
tiaanels said:
As someone who is running L on a Nexus 5 I beg to differ.
Click to expand...

I'm also sceptical with the 30+% improvement. From the Google I/O talk they reckon it might be 15-20% if I remember correctly. A lot of it has do do with the background applications and scheduled wake-up to perform certain tasks. Also not all apps will immediately take advantage of this.
 
It is an improvement. The results will differ from device to device and manufacturer to manufacturer. The first battery saving benchmark is done and it is pretty significant. This is a win for all Android users who will be able to load "L" on their devices once released.

Zyzzyva said:
This tells us more about the inefficiency of the last Android release than the efficiency of Android L.

It's still impressive though.
Click to expand...

I won't say that it reflects any inefficiencies of KK. KK was already an improvement over JB (as JB was an improvement over GB, etc etc). This is the result of constant tinkering where it matters - Google. "L" uses ART. ART is better at managing the OS than Dalvik and therefor there should be battery usage improvements. It is just progress.
 
IIRC on Kk Rom Carbon AOSP it gave a choice of ART or Dalvic cache, found it on XDA. Must give it a run to see if it saves the battery.
 
Stock KK have got the option, but ART was very new and the devs at Xposed didn't code for it. That was the major reason I could never be bothered with it.
 
You must log in or register to reply here.
Back
Top
Sign up to the MyBroadband newsletter
X