Optimus V Battery Drain Optimization Tests Part 2

mmarz

Well-known member
Mar 9, 2011
1,130
447
0
Visit site
Setup:

I ran a second set of tests on the Optimus V. You can find my first set of tests here. To summarize my procedure, I splice equipment in between the battery and the phone so that I can monitor how much battery is being used at any given time. I install various roms and test how much the battery is drained using different features. This time around I used more sensitive equipment that gave a resolution of 0.01mA (100 times higher resolution than before!). My approach to the calculations was different this time too. Instead of just considering the current, I also recorded the voltage across the battery. In this way, I can calculate the instantaneous power draw.

Each rom was a fresh install. After the install, a Wifi network was added. Wifi sleep policy was set to never sleep. A google account was added. The rom was then restarted and airplane mode was toggled to remove the 50% TWS bug. No other settings were tampered with, including CPU speeds. Besides the typical tests from last time, I also included:

  • Downloading - Docs2Go using Wifi and 3G
  • Checking Email using Wifi and 3G
  • Calling Voicemail with 3G and Wifi turned off
  • Where (the app) Running
  • Where running in the background while on the Homescreen
  • Where running in the background while on Standby

Using all of the power drainage values, I created a formula for calculating the estimated battery life for the roms if set to 3G, Wifi, or a mix of both. The formula uses best guess values for the number of hours spent talking, surfing, running programs, and left in standby. The breakdown is as follows: 2 hours running programs, 1 hour surfing, 1 hour calling, and 20 hours on standby.

I also did an independent test to see the effects of underclocking. Using SetCPU on Rodimus, I forced different clock speeds and ran the Long Bench Test. I also looked at what happens when you go into Standby underclocked.

The roms I tested:
  • Aphrodite_1.3*
  • RodimusROM.v1.2.3
  • Zefie's CM7 03012011
  • aospCMod 03252011
  • aospCMod 04032011
*I forgot to run some of the tests on Aphrodite, but that didn't effect the final calculations.

Results:



csx2d.jpg

This assumes: 2 hours running programs, 1 hour surfing, 1 hour calling, and 20 hours on standby for a given day.

This chart sums up all of the testing. The CM7 roms have a hug problem when it comes to Wifi power use in the idle state. When downloading files, all the roms use approximately the same amount of power. The average 3G power use when downloading is 1600mW which is almost twice as much as Wifi, 850mW. This is in contrast to the power use when in standby with no data transferring where 3G consumes the same amount of power as airplane mode and Wifi consumes at least twice as much with Rodimus and at most 34 times as much with the CM7 roms. Rodimus and Aphrodite have much better battery life on Wifi than the CM7 roms. This all depends on how much Wifi and 3G you actually use. The transfer speeds did not effect the power consumption. Checking email which occurred at 40 kbps consumed the same instantaneous power as downloading Docs2go at full speed. This was true for both Wifi and 3G. The "Mix" results assume you spend half of the 24 hours at home using Wifi (sleeping counts), and the other half using 3G. If you would like to experiment with the constants used to calculate the estimated battery life, download the attached xls file and manipulate only the values highlighted in green. The values highlighted in pink and chart 5 will change as a result.

Update 4/6/2011: The new asop rom seems to have improved the power consumption of the Wifi antenna drastically. There is still room for improvement compared to the stock rom, but I guess that is just a trade-off for the GB features. I have uploaded an updated xls file with the new results included.

The last test involved setting different CPU speeds and seeing how they effected standby power consumption and the power consumption during the Long Bench Test of SetCPU.


25cpg.jpg

As you can see, under and over-clocking has zero effect when the CPU is idle.


r1lj0i.jpg

This chart shows that the higher the CPU speed, the more power is consumed during the CPU intensive bench test, but this does not show the entire picture. Since the bench test is using a standard set of calculations, we can compare the time it takes to finish the test across the different CPU speeds. This value is conveniently given as the result of the bench test. Multiplying the steady power consumption to the time it took to finish the bench gives us the energy consumed to finish the bench. The clock speed with the least amount of energy use is therefor the most efficient speed (for this particular phone).


30ngaya.jpg

When we zoom in on the results at the high end of the CPU clock speeds, we see that the highest stable clock speed for this phone, 825 MHz, is not the most efficient. It is actually 806MHz that consumes the least amount of energy to complete the test. Even though 825 is faster, the increased power consumption counteracts its speed. It is also obvious that the slower clock speeds are way more inefficient for completing a CPU intensive task.

For some reason, 320MHz and 245MHz gave identical results. I think this is a kernel or hardware bug. I tested it repeatedly and got the same results.

Conclusion:

If you want to see what your battery life would be under various conditions, download the xls file and manipulate the values in green. The values in pink will change and will automatically be plotted in chart 5. Depending on your phone habits, you can pick the rom that suits you best. Remember that these are rough estimates. Your actual mileage might vary.

Turning off Wifi when the screen is off can save you battery life IF you have no backgrounds apps that send and receive. If you do have those types of apps, Wifi is better than 3G to leave on (not including the current CM7 Wifi issue). If you are using 3G, there is no advantage to turning it off when the phone goes into standby unless you again have background apps that consume bandwidth. 3G consumes the same amount of power as airplane mode if there is no data transferring! My suggestion is to eliminate those types of apps if you don't need them and leave 3G on. That will save you the most amount of battery out of all the mods you can do.

If you are going to use SetCPU, overclocking will save you battery life. Underclocking will actually cost you. I would recommend leaving the upper limit one notch lower than what is considered stable for your phone. The lower limit does not matter all that much. I now have my lower limit set at the same value as the upper limit. When idle, both high and low clock speeds consume the same power. When the CPU is needed, the higher clock speed is more efficient (to a point). I would advise against the "conservative" scaling that gradually changes your clock speed to task if you still decide to underclock. This will only slow you down and consume more battery power. You also don't need a separate profile for when your phone is on standby. Lower clock speeds can only worsen your battery life.

Lastly, underclocking is not the same as undervolting. Undervolting would probably decrease battery consumption, but it is not implemented in any rom/kernel for the Optimus V...yet.

Thank you and good night!

Update: I made a guide to create firewall profiles to block apps from using your 3G/Wifi when the screen is off. This should eliminate the need to turn off your 3G when the screen is off. Read more here: http://forum.androidcentral.com/opt...ock-apps-using-3g-wifi-tasker.html#post788344
 
Last edited:

maxpower49

Well-known member
Mar 3, 2011
47
1
0
Visit site
Thanks for the update… I love this kind of info. I recently updated to the asopCM7 rom (previously on stock). It is interesting that that battery life on 3G is comparable to the 2.2 roms, but on wifi it is considerably worse. Hopefully they get that bug worked out on an upcoming release.. I’m loving gingerbread. I guess it would be best to set wifi sleep policy back to “when screen turns off” so it uses 3G instead of wifi when the screen is off?

Very interesting about underclocking with SetCPU also… I had mine set to underclock when the screen is off, but I am removing that profile now. Going to use 480/806 interactive with all other profiles turned off and see if I notice any difference.
 

digitaljeannie

Well-known member
Feb 26, 2011
70
10
0
Visit site
LessWatts.org - Saving Power on Intel systems with Linux

Race to Idle


Current processors are quite good about saving power when idle; so much so that many show a behavior around power saving that surprises many people.

This behavior, called race-to-idle, is best explained with a simplified example:

Lets take a typical commercially available processor that consumes 34 Watts when running at full speed, and 24 Watts when running at half speed and 1 Watts when idle (using frequency and voltage scaling using P-states).

On this processor, we're decoding one second of a MP3 file or some HDTV media every second. This decoding takes 0.5 seconds at half speed, and, consequently, 0.25 seconds at full speed.

The energy consumption for one second is

Half speed: 0.5s * 24W + 0.5s * 1W = 12.5 Joules
Full speed: 0.25s * 34W + 0.75s * 1W = 9.25 Joules
Even though the above example is simplified from reality, the same paradigm tends to hold for real systems: It's generally better to run as fast as you can so that you can be idle longer.


Also, it seems that the interactive governor may be better that, say, the on-demand governor because the polling that it does to determine which speed to use constantly brings the phone out of idle for the poll whereas the interactive governor is scaled based on cpu load.
 
Last edited:

one80oneday

Well-known member
Jan 25, 2011
289
17
0
twitter.com
Nice research! I've asked a few times here and there about undervolting on the Optimus but everyone was either confused or told me to underclock which I figured could have ill effects. Hopefully an undervolted kernel will come soon :)

Sent from my LG Optimus V using Tapatalk
 

LeslieAnn

Android Developer
Feb 8, 2011
2,895
1,720
0
Visit site
Interesting results once again.
Makes me wonder what changed in Rodimus to make wifi so much better.


I find the information about underclocking the most surprising though. I have long theorized the max speed effect on battery due to how my computers have acted when using faster, hotter processors, but the underclock was a surprise. I wonder if maybe the phone is trying to do this automatically and they are fighting each other. Time to change my settings on the phone.
 

mmarz

Well-known member
Mar 9, 2011
1,130
447
0
Visit site
Aphrodite 1.3.1 has been posted and should close some, if not all, of the gap with Rodimus on this test.

This wasn't meant as a challenge :)

But I do appreciate a dev working to make their rom better. Keep in mind that there is some level of uncertainty with these results. Don't take it as an insult if your rom isn't the fastest in every category.
 

LeslieAnn

Android Developer
Feb 8, 2011
2,895
1,720
0
Visit site
Oh, I don't.

I don't expect it to, I'm just trying to make it as good as I can. I haven't looked at what Mmmark111 has done, I just implemented a few new things (wrongly). I wouldn't be surprised if Rodimus still wins this.

We have different goals with our roms.
 

Helgaiden

Well-known member
Jun 17, 2010
136
14
0
Visit site
i suppose stock + xionia would align closer to Rodimus as far as battery usage right? Or would it be closer to Aphrodite? @_@
 

trippin7464

New member
Mar 25, 2011
3
0
0
Visit site
I know that these results weren't really trying to make a competition out of the popular roms, but it definitely helps me decide that I want to try out Rodimus or Aphrodite over the others.

I am definitely concerned with battery usage, and have been using Juice Defender to turn off 3g when my screen is off with great results, (the icon says about 1.6x to 1.8x battery savings. Which is awesome, I was able to go a day and a half (phone off charger around 8am, no charge all day and all night, and got battery low warning mid-afternoon the next day), when before juice defender I wasn't even getting to 10pm before the battery warning.

Thanks so much for doing the work, not just once but twice, really interesting to see!
 

justinh

Well-known member
Mar 22, 2011
108
8
0
Visit site
I just gave my phone a "real world" test today. I am using RodimusROM v1.2.3 + xionia lots of apps that I would use everyday. I left for work today at 9:00 AM MST...I listened to slacker radio untill about 6:00 PM MST. It is now 9:43 and ever since i stopped listening to slacker radio the phone has been basically idle except texting here and there..the uptime has been 11h 56m and the battery is at 18%..hopefully this helps someone :)
 

mmarz

Well-known member
Mar 9, 2011
1,130
447
0
Visit site
I just gave my phone a "real world" test today. I am using RodimusROM v1.2.3 + xionia lots of apps that I would use everyday. I left for work today at 9:00 AM MST...I listened to slacker radio untill about 6:00 PM MST. It is now 9:43 and ever since i stopped listening to slacker radio the phone has been basically idle except texting here and there..the uptime has been 11h 56m and the battery is at 18%..hopefully this helps someone :)

Using the excel file I attached to the first post, your estimated battery life is:
Battery Life Estimate 3G (hr) 8.83
Battery Life Estimate Wifi (hr) 17.00
Battery Life Estimate Mix (hr) 11.63

This was assuming 9 hours of internet use (the radio), 6 mins of calls, and 6 mins of running apps.

I would call this fairly accurate if not conservative :)
 

justinh

Well-known member
Mar 22, 2011
108
8
0
Visit site
Using the excel file I attached to the first post, your estimated battery life is:
Battery Life Estimate 3G (hr) 8.83
Battery Life Estimate Wifi (hr) 17.00
Battery Life Estimate Mix (hr) 11.63

This was assuming 9 hours of internet use (the radio), 6 mins of calls, and 6 mins of running apps.

I would call this fairly accurate if not conservative :)

wow..i would call that freakishly close..haha
 

Eollie

Well-known member
Feb 22, 2011
1,534
258
0
Visit site
I know that these results weren't really trying to make a competition out of the popular roms, but it definitely helps me decide that I want to try out Rodimus or Aphrodite over the others.

Consider this Rodimus and Aphrodite are based on 2.2 CM7 is 2.3.
And the 2.3 is a generic build of the asop and not optimized for the optimus. (wow that was a hard sentence to make right)

Now however mmarz testing is pretty damn useful in knowing what to do to extend that battery.
 

one80oneday

Well-known member
Jan 25, 2011
289
17
0
twitter.com
Tested higher clocking only yesterday (480-864) and was surprised that battery life was the same as usual however it can't be good for the processor to never get a break.

Sent from my LG Optimus V using Tapatalk
 

Trending Posts

Forum statistics

Threads
943,139
Messages
6,917,471
Members
3,158,836
Latest member
Robbyworkman1995