Not logged in
Log in now
Create an account
Subscribe to LWN
LWN.net Weekly Edition for December 5, 2013
Deadline scheduling: coming soon?
LWN.net Weekly Edition for November 27, 2013
ACPI for ARM?
LWN.net Weekly Edition for November 21, 2013
ICS works acceptably even on hardware like the HD2, released 2009, which originally came with Windows *Mobile*.
"Last year's phone"
Posted May 16, 2012 12:15 UTC (Wed) by Richard_J_Neill (subscriber, #23093)
Yes, android is prettier, but we make a terrible performance tradeoff. Even a basic "flashlight" app is hundreds of kB, and something like a train-timetable app is ~ 4 MB.
Then again, why code for efficiency? CPU and Flash are now really cheap, and battery life is dominated by the backlight anyway.
Android App Bloat??
Posted May 16, 2012 23:38 UTC (Wed) by ldo (subscriber, #40946)
Even a basic "flashlight" app is hundreds of kB...
I don’t understand how you come up with that figure. Here is an app of mine that does something reasonably interesting, and includes built-in help, in a package that is just 59kB in size. And the built APK for this sample app is just half of that.
Dalvik is a very compact byte code—half the size of Sun JVM bytecode. And all the additional XML resources for your app are stored in a compressed binary form. So I don’t see where the bloat should come from...
Posted May 17, 2012 16:57 UTC (Thu) by zlynx (subscriber, #2285)
Posted May 17, 2012 22:30 UTC (Thu) by ldo (subscriber, #40946)
Posted May 17, 2012 23:49 UTC (Thu) by zlynx (subscriber, #2285)
I thought that the ODEX (Optimized DEX) format can contain a cached copy of the JIT-compiled machine code. I also believe that the ODEX format is commonly used for all applications which are included in a ROM, because each ROM is hardware specific anyway.
Posted May 18, 2012 1:09 UTC (Fri) by ldo (subscriber, #40946)
Posted May 17, 2012 12:46 UTC (Thu) by jschrod (subscriber, #1646)
Posted May 31, 2012 15:49 UTC (Thu) by nye (guest, #51576)
What? How can that be? Are you using your phone with the display on constantly, while still using nearly zero CPU?
Have you actually tested this, or is it just an intuitive assumption? If the former, what kind of screen does your device have? If the latter, your intuition is incorrect.
Unless I'm using the power-guzzing Google navigation, 'cell standby' is always at the top of my battery usage list, followed by 'phone idle', followed by 'Wi-Fi' if I've had it enabled, usually followed by 'Android system', with perhaps a few other entries, and 'display' dead last - and that's if I've had it on for the hour or so that it takes to even show up in the list (I think there's a threshold of 1 or 2 percent).
If I do nothing after unplugging the phone but tap the screen every now and then to keep it on, 'display' will go up to 20%, with 'Android' system in the 60s, so if that were more efficient it would have a substantial impact on battery life (although in the real world case 'cell standby', 'phone idle', and 'Wi-Fi' tend to come to around 90% of the total battery use).
Posted Jun 1, 2012 9:17 UTC (Fri) by rschroev (subscriber, #4164)
Cell standby 18%
Dolphin Browser HD 9%
Phone idle 7%
Android OS 5%
Android System 5%
The screen is most certainly not on most of the time. Also I didn't realize Dolphin uses that much power; from the stats you'd think I do nothing but surfing the web all the time, which is not true at all.
The phone is a Samsung Galaxy S (GT-I9000) running Cyanogenmod 7.1 (but I seem to remember that the start where more or less the same when the phone still had the stock software).
Copyright © 2013, Eklektix, Inc.
Comments and public postings are copyrighted by their creators.
Linux is a registered trademark of Linus Torvalds