Rockbox Technical Forums
Rockbox General => Rockbox General Discussion => Topic started by: Colombo on February 17, 2007, 02:07:18 PM
-
Original Apple OS is 14 hours playing.
Rockbox is ~6-7. I think it bcuz mp3 player get more CPU time and so more battery power. Can developers make low battery use?
p.s. sorry, bad english.
-
'bcuz' is not a real word. Our forum guidelines require using real words. Please follow these rules in the future.
It's much more complicated than simply how much CPU time is being used. While it's being worked on, this is very well known, and is not likely to be an easy solution.
-
and what is debugging -> CPU frequency modification? What is boost ratio? How it influences on battery use? I set 30mhz and music became 3 sec play and 1 no play :D
-
rockbox is very good, many features etc, but simply playback time is 2x less than apple os. I must return back to apple os.
So when rockbox will be low CPU usage, notice it in main page :] thx
-
It is most likely not CPU usage causing the battery life alone, but likely a combination of this (solved through optimization), and perhaps some components not being turned off properly. It is very hard to code for chips with no data sheets or public information available ...
(standard disclaimer: as I understand it)
-
We could test that theory if someone were to measure the power consumption with a DMM on the ipod at a few different clock speeds and no boost. Then we could figure out if the high power use is from high CPU use or from devices which are constantly bleeding off power independent of what rockbox is doing.
Probably easier to just work on the CPU nuse issue first though, since that needs fixing regardless :)
-
I think mp3 playing using many cpu without rest like windows.sleep(...); in the thread.
-
I think developers first must make low battery usage like in original OS, before any other daily changes, because nobody want using firmware with 2x less playback time.
-
I think developers first must make low battery usage like in original OS, before any other daily changes, because nobody want using firmware with 2x less playback time.
On most other devices Rockbox runs on, it consumes less power than the original firmware, and obviously we want the same thing on the ipods.
However, battery life is not every user's #1 priority, as shown by the number of people who are running it on their ipod. If you only listen to music for 5 or 6 hours a day, and can easily charge your ipod overnight, then what's the problem?
Work _is_ being done to improve battery life on ipods, it's just a slow progress, and will be a combination of small incremental improvements and sudden breakthroughs. No-one can predict when we'll reach the same playback life as Apple's firmware does though.
Also, don't forget that not all the developers who work on Rockbox even own ipods - Rockbox is far more than just a replacement ipod firmware.
-
I think developers first must make low battery usage like in original OS, before any other daily changes, because nobody want using firmware with 2x less playback time.
You are wrong.
-
I did some tests using a 4.5v power supply.
Idle after booting
apple
~15.6 mA
~60.5 mA with Back Light
rockbox
~25.8 mA
~71.7 mA with BL
~34.7 mA cpu boosted no BL
during playback. My multi meter is a bit slow so these are estimates.
~29mA
~41mA boosted.
Doesn't look like reducing the amount of time the cpu is boosted will help that much.
EDIT
Tests done on a nano.
-
the_winch: Which iPod? I assume it has frequency scaling enabled?
Assuming Apple uses the same idle frequency as Rockbox, that would certainly imply theres some hardware running in Rockbox that should be disabled.
-
the_winch: Which iPod? I assume it has frequency scaling enabled?
Nano
-
These numbers are very interesting.
The only question I have is, did you account for battery charging in any way?
I think the most valid way of doing this would be to unplug the battery, and wire in the external power supply directly, not using the dock connector at all.
-
I think he did use a 4.5 V power supply, though maybe that was just to measure power use?
Still, this reinforces my suspicions that getting that Sansa emulator working so we can read off all the registers and I2C traffic the retail PP firmwares use is going to be absolutely critical to getting the PP ports working as well as retail.
-
I think the most valid way of doing this would be to unplug the battery, and wire in the external power supply directly, not using the dock connector at all.
That's what I did. I didn't use the battery because I had no second multimeter to measure the battery voltage.