Related
Hi, I'm a noob to undervolting here (just rooted and flashed Vanilla Froyo to my Defy a week ago, and I'm hoping to squeeze out as much battery life as I can); just some quick questions about undervolting that I haven't found answers to in other threads:
1) What exactly is "vsel" and how does it relate to the actual battery voltage? For example, even though I've changed the maximum voltage in setvsel and Milestone Overclock (down to about 45 vsel at 800 Mhz so far, haven't tried lower), the battery voltage reported by Spare Parts is still reading between 3800 mV and 4000 mV. Does this mean that setvsel/milestone aren't working or does a ~30% decrease in vsel *not* equal a 30% decrease in battery voltage?
2) If anyone else has undervolted their Defy, can they testify as to what kind of battery life improvements they've gotten? I've heard that custom ROMs are already fairly undervolted
If laptops are anything to go by.. laptops can be undervolted to make them run less hot, thereby extending their life. Though battery life is increased too, this increase is virtually negligible.
Since the limits of undervolting (and over/underclocking), are done on a per-cpu/gpu basis for computers, because they are not all exactly the same and may not work with the same non-default values, I think it's a bad idea to assume that phones will run without problems when using values only tested by people with a phone that is not yours.
If you want to undervolt/over- or underclock, test your device, find out the lowest/highest values and then inrease or decrease that value, to have a safety margin, do not copy someone elses' settings.
A common issue with undervolting on 2.1 (at least with me on 2.1) is that if the minimum clock speed is below 300 mhz, it takes around 3-5 sec to actually respond to incomming call on your defy when defy is on stand-by mode.
Undervolting + Ondemand made my battery work at 23ºC vs +29ºC
I did not wrote down numbers, but battery life increased AT LEAST in 8 hours.
My values
vsel 1; 18 -300
vsel 2; 33 - 600
vsel 3; 43 - 800
(-15 points each value VS factory values)
Removing bloatware, root apps, autostart at boot apps, widgets (they autostart too) etc. or freezing them with "antitek app manager"
Decreased phone memory ussage from 140Mib to 80Mib; Also decreased cpu ussage from 10%-15% average idle to just 1%-5%
Apps / Widgets / Many System preinstalled apps, are autoloaded when phone is turned on, and they run 24/7. waste of resources and app auto killers just aren't good enough, coz apps reload themselves, you have to remove them or freeze them.
fortunely modders are realeasing their custom roms with most of this optimizations
45 vsel
using milestone oc I got 1ghz @ 45vsel from the day I got it I have owned the device for 1 1/2 months now not one single problem but it depends from device to device I must have a bin'd cpu
Ok EVO nation!
Here's my scenario. I'm running a fresh flash of MikFroyo 4.5 and Netarchy's 4.2.2. cfs nohavs. With almost nothing else running on the phone.
So far I've seen a the stand-by battery drain hover around 100-180mA, with an occasional dip into the 50's. The CPU usage is always just below 10%.
I've seen references all over the forums about users claiming low 40's and idle CPU usage around 3% and I'm wondering WHY I'm not able to achieve the same thing.
Now I've done all of the standard battery tweaks, tried various kernels, and can remember a time last fall when one kernel I flashed seemed to give me incredible battery life.
So...Let's hear what your average stand-by battery drain and CPU usage is.
(And any thoughts you might have about my battery situation)
Get 4.3.2 with more havs. It'll drop the voltages nice and low. When I had mikfroyo with netarchy 4.3.2 with setcpu screen off at 128mhz I had about 30-40 idle drain.
Sent from my (insert daily ROM name here) Evo 4g
I wouldn't worry too much about your idle current draw unless you're having a battery drain problem. The issue with the current draw methodology is it's taking the current consumption that instant in time. Unfortunately, you can't monitor instantaneous current draw without affecting the reading itself.
If you idle for an hour or two (or 5), and your battery only goes down 0-1%, i'd say you're in good shape. Also, it helps if you specify what software you're using to monitor these values with? Mine is based on Battery Monitor Widget (bmw).
Edit: There's more to cpu tweaks than just kernels. Along with kernels, there are governors, and then there are governor parameters. Each will directly affect your battery consumption and cpu load.
I'm definitely having a battery drain problem. At idle, it's falling 3-4% per hour, according to the Battery Monitor Widget.
I've got SetCPU set to 128min/998 max with Conservative scaling, and I've one profile enabled for Screen off settings of 128min/128max.
I did notice that when I try to set the min cpu to 128, the clock speeds jump all over the place. But if I set the min cpu to 245 it quiets right down.
Perhaps I a have CPU that's become finicky?
fsmith3x said:
I'm definitely having a battery drain problem. At idle, it's falling 3-4% per hour, according to the Battery Monitor Widget.
I've got SetCPU set to 128min/998 max with Conservative scaling, and I've one profile enabled for Screen off settings of 128min/128max.
I did notice that when I try to set the min cpu to 128, the clock speeds jump all over the place. But if I set the min cpu to 245 it quiets right down.
Perhaps I a have CPU that's become finicky?
Click to expand...
Click to collapse
I had an issue similar to this where my battery would drain quicker than usual when the CPU min was set at 128 and having a screen off profile of 128/128. I think the problem stems from just setting the CPU way too low and having it use more power to bring the phone 'up to speed'. From my personal experience, I think you'll still be just as good setting your mins to 245, but by all means you're more than welcome to continue experimenting with 128
First off, you need to determine if you have a wake lock issue. Use "spare parts" to see. Look in the battery history, partial wake lock section. It will show you what programs are causing wake lock, and what amount of time they are in a state of wake lock.
Example, my phone was last off the charger 42 hours ago (awake time 7h36m), battery still at 32%, and should last the remainder of the day so long as I don't do too much data. My two biggest wake lock listings are Android system at 1h26m, and my email at 20m.
All it takes is 1 misbehaved program, or even a particular setting within a program to cause wake lock. If that is the cause, then regardless of what your sleep settings are, the phone will still draw the battery down quickly.
You can also look in "/sys/devices/system/cpu/cpu0/cpufreq/stats" at the time_in_state file using a text viewer/editor. It will show how much each cpu frequency is being utilized. I don't recall what unit it's in. But the point here is that if your 128/245 Mhz values are lower than your upper mhz values, something is not right.
The governor parameters will control how much cpu swing is occurring under various load. That is, if you're reading an email, there's little reason for the cpu to be pegged at 998mhz. These parameters control the thresholds (contingent upon which governor is used) at which the cpu scales up or down.
Edit: Each value in time_in_state represent 10ms, or .01seconds. So multiple each value below by .01 to get actual time in seconds. If you notice, the total time in us equal to my awake time of 7h36m, not total uptime.
Mine looks like this
Code:
Mhz cycles seconds
128000 1264017 12640
245000 201083 2011
384000 72728 727
422400 4 0
460800 133450 1335
499200 0 0
537600 72174 722
576000 53479 535
614400 78771 788
652800 38230 382
691200 0 0
729600 87258 873
768000 29579 296
806400 98273 983
844800 18476 185
883200 0 0
921600 599503 5995
960000 0 0
998400 2821 28
Total 2,749,846 27,498
27498seconds = 458min = 7.63h or 7h37m.
gpz1100 said:
Example, my phone was last off the charger 42 hours ago (awake time 7h36m), battery still at 32%, and should last the remainder of the day so long as I don't do too much data. My two biggest wake lock listings are Android system at 1h26m, and my email at 20m.
Click to expand...
Click to collapse
gpz1100, Please describe your setup and usage. I'm curious to know how you're getting that kind of battery life.
Anyway....
Good morning gang!
I gave this a couple of days for the fresh install to settle in, and hopefully let it learn how to use the battery more efficiently. But it's still a power hungry beast.
Yesterday I calculated it's use ate 3.9% an hour, based on battery % / time unplugged.
Mind you, I made 2 short phone calls yesterday, and checked my Exchange mail a few times throughout the day. No Angry birds while I'm trying to sort out batter life.
It's still using 100-150mA in standby, and it seems like the governor (SetCPU) is working like it should.
Right now the battery is at 97% and has been off the charger for 1.5 hours. That's 2% an hour. Which isn't bad, but I doubt the average will be that low by the end of the day.
For those of you who like numbers, here's some stats for you. Do you see anything that raises any red flags?
Looking at Spare Parts, in the wake lock section, I see these values.
Android System: 9m9s
UID 10020: 54s
Facebook: 39s
Voicemail: 30s
Mail: 14s
Maps: 11s
Tasker: 9s
WidgeLocker: 6s
Other Usage...
Running: 36.1%
Screen on: 16.7%
Phone on: .9%
Wifi On: 100%
Wifi Running: 100%
The values from my 'stats' file:
128000 0
245000 84095
384000 20371
422400 9
460800 0
499200 127
537600 4743
576000 100
614400 9
652800 75
691200 4238
729600 0
768000 23
806400 1358
844800 2439
883200 0
921600 0
960000 976
998400 18361
1036800 0
1075200 0
1113600 0
1152000 0
1190400 0
1228800 0
1267200 0
fsmith3x said:
gpz1100, Please describe your setup and usage. I'm curious to know how you're getting that kind of battery life.
Click to expand...
Click to collapse
My setup is as described in the sig. The only additions would include custom conservative governor settings of 80/55/10 % on the up/down/freq steps thresholds; min/max cpu of 128/921 Mhz respectively. I don't use setcpu, tasker, or anything similar. Background updates is on, but nothing is checked (no gmail, facebook, etc.). I used to have email push enabled (using k9 mail), but decided to turn off push altogether, and just poll the server manually when I need to check email when away - little reason to getting emails as they come in if I have access to a computer at the same time.
I have found that even with wifi, the quality of your voice signal drastically affects battery life. I'd say I have average signal, (3-4 bars most of the time) for voice. However, sprint provided an airave because data speed is terrible here. A byproduct of the airave is it provides excellent voice signal. So, just with that in place, consumption while the phone is sleeping/idling is reduced by a good 10-20% over connecting to the tower directly. I'd say most of my usage is voice, some texting/emailing, and on occasion, heavy browsing. No games or videos, no facebook/twitter. It's hard to quantify use, but I think that's the bulk of it.
Yesterday I calculated it's use ate 3.9% an hour, based on battery % / time unplugged.
Mind you, I made 2 short phone calls yesterday, and checked my Exchange mail a few times throughout the day. No Angry birds while I'm trying to sort out batter life.
It's still using 100-150mA in standby, and it seems like the governor (SetCPU) is working like it should.
Right now the battery is at 97% and has been off the charger for 1.5 hours. That's 2% an hour. Which isn't bad, but I doubt the average will be that low by the end of the day.
Click to expand...
Click to collapse
What you need to keep in mind, is that unless you're using an sbc kernel, your battery may or may not stay above 90% for too long after being removed from the charger. Unfortunately, battery consumption is not constant. Demand on the battery is changing depending on what the phone is doing. Also, 1.5 hrs really isn't a large enough time period to draw an accurate average from. Based on your cpu usage below, it indicates your device was actually on for 22 min.
Your wakelock numbers don't look bad, but the data is inconclusive. The question is, when these readings were taken, what was the up time, and what would you say was the time with the screen off (rough estimate is the diff between uptime and awake time). Leaving wifi on is a good idea, as it draws much less energy than using 3g.
One number that does seem surprising (assuming total run time of 1.5h's since removed from charger) is your running % of 36.1%. If the phone was not used during this time, that seems excessive. In fact, looking at the next value, screen on - 16.7%, would suggest that 20% of the awake time was when the screen was off.
I would start by disabling or uninstalling(if you use titanium backup, the FREEZE) tasker and widgetlocker. Running % - screen on % = % of time spent not sleeping even though the screen is off - things like receiving email, performing weather updates, etc...
Just curious, what's your lowest vsel?
Mine are currently set as: 18/300 - 33/600 - 39/800.
Sent from my MB525
On 2.1 my 300mhz waz 14 but now im on 2.2 i av to use 16 but heres mine 300/16 600/28 800/39
Sent from my MB525 using XDA App
Running eclair 2.1 blur version 2.21 & my settings are:
300/18
700/32
1100/56
Could probably get the 300 & 700 lower (1100 is at the lowest stable I could set) but I'm happy with what they are.
Up_threshold set to 75%.
Very snappy & good battery life - only uses 4% battery when sleeping over night for 8 hours.
My threshold iz on 99% and i get 3days owt of mine
Sent from my MB525 using XDA App
I have used this for the first time tonight and I'm running the following without any issue
300 / 18
600 / 32
1100 / 56
Mine is 1000/50. I did some testings and it is stable.
sp8y said:
Running eclair 2.1 blur version 2.21 & my settings are:
300/18
700/32
1100/56
Could probably get the 300 & 700 lower (1100 is at the lowest stable I could set) but I'm happy with what they are.
Up_threshold set to 75%.
Very snappy & good battery life - only uses 4% battery when sleeping over night for 8 hours.
Click to expand...
Click to collapse
Now running the leaked UK 2.2 froyo 3.4.2-117 & I've kept my vsel the same but now have the up_threshold set to 70%.
if i set the lowest vsel, will my battery live longer?
albertocalle said:
if i set the lowest vsel, will my battery live longer?
Click to expand...
Click to collapse
Less voltage consumption = more use time without needing to charge.
Less charges = Increased lifespan of battery (ussually batteries are rated at 1000 charge cicles)
Less voltage also means less temperature, which may increase lifespan in the long term too.
In climates of 32ºC+ for stores who sell it, its very recommended to store them in refrigerators not in the coolest compartment but next to the drinks.
One battery stored at 10º vs 30º for a few months could mean a lot.
btw my battery lasts at least 12/24 hours more per charge, thanks to setvsel
My settings are
19 300 30 600 40 800
Aply at boot + Limit to vsel1 while screen is off.
In my case:
300/24
600/34
1000/50
At the above voltages, the stability test can run at least 30 minutes with interruptions, like pulling the status menu, showing the instructions etc, from time to time.
hi,
I am running stock jellybean with franco kernel. just wondering what is the optimal CPU setup that doesnt sacrifice too much performance while increasing battery life. Im not too concerned about games as I rarely play them. I ussualy use the phone for music, calls, text, email, and webbrowsing.
I am using the Franco updater app. I have 1228 max, 230 min
and screen off max is 384.
I also noticed in cpu spy that my highest two frequencies 1036, 1228 are combined less than 10%.
with all that information, what do you suggest?
hshaikh said:
hi,
I am running stock jellybean with franco kernel. just wondering what is the optimal CPU setup that doesnt sacrifice too much performance while increasing battery life. Im not too concerned about games as I rarely play them. I ussualy use the phone for music, calls, text, email, and webbrowsing.
I am using the Franco updater app. I have 1228 max, 230 min
and screen off max is 384.
I also noticed in cpu spy that my highest two frequencies 1036, 1228 are combined less than 10%.
with all that information, what do you suggest?
Click to expand...
Click to collapse
I suggest to decrease your max CPU until you feel it affects performance too much. You can also try switching to a governor less aggressive than interactive (try ondemand).
Don't expect magic though. I've played with decreasing max CPU clock, max CPU screen off, governor settings etc with three different kernels. Gathered stats for at least one week each time. Never noticed a difference large enough to actually matter to me. For maximizing battery life, you can gain more by hunting apps that cause a lot of (partial) wakelocks and alarms, and use low screen brightness.
hshaikh said:
and screen off max is 384.
with all that information, what do you suggest?
Click to expand...
Click to collapse
don't limit screen off cpu speed to 384. it will take longer to perform operations under wakelock, thus killing of more battery than it actually saves.
it would be nice if listening to music, since it's not an intensive task and it requires constant cpu usage. still, you'll most likely have stuff syncing in the background so i don't think you're doing any good by limiting it to 384 on screen off.
power isnt going to be conserved with lower clock speed. lower clock speed means it takes longer to finish the task. to conserve power, your phone has to do less. for example, lower brightness, no sound, less/no syncing.
Darunion said:
power isnt going to be conserved with lower clock speed. lower clock speed means it takes longer to finish the task. to conserve power, your phone has to do less. for example, lower brightness, no sound, less/no syncing.
Click to expand...
Click to collapse
Well, there is an optimum somewhere. Higher CPU speed means tasks are executed more quickly, but higher clock speeds also draw more current from the battery. I agree that tweaking this has little effect on battery life though.
Petrovski80 said:
Well, there is an optimum somewhere. Higher CPU speed means tasks are executed more quickly, but higher clock speeds also draw more current from the battery. I agree that tweaking this has little effect on battery life though.
Click to expand...
Click to collapse
you are correct. there is a magic middle ground because power consumption doesnt scale in a linear way. but finding that spot would take massive testing and even getting to the center, would still probably only gain about 10-15mins average use on a battery charge :/
bk201doesntexist said:
don't limit screen off cpu speed to 384. it will take longer to perform operations under wakelock, thus killing of more battery than it actually saves.
it would be nice if listening to music, since it's not an intensive task and it requires constant cpu usage. still, you'll most likely have stuff syncing in the background so i don't think you're doing any good by limiting it to 384 on screen off.
Click to expand...
Click to collapse
what speed show i set to max when screen is off.
did you read anything that Darunion and Petrovski80 wrote? I keep mine at it's max, i don't care, i don't live in the woods with no electricity.
thanks for your inputs. i have experimented and I found out that changing the cpu speeds has minimal effect of battery life. the clock speed is not killing the battery the screen is. no matter what cpu settings i use i get 3-4 hours screen on time.
if i dont use the phone alot that day (like 1 hour screen on time) the battery will still be 40% after a day.
hshaikh said:
thanks for your inputs. i have experimented and I found out that changing the cpu speeds has minimal effect of battery life. the clock speed is not killing the battery the screen is. no matter what cpu settings i use i get 3-4 hours screen on time.
if i dont use the phone alot that day (like 1 hour screen on time) the battery will still be 40% after a day.
Click to expand...
Click to collapse
Exactly. I get similar performance.
Sent from my Galaxy Nexus using Tapatalk 2
on the nexus 7 a lot of apps mostly games use a lot more power than they need so limiting the clockspeed should fix that but using francos per app modes or just changing the max frequency but the battery still drains at the same speed
for example:running a game without limiting clock i get 3 minutes per percent but when limiting to 384 which should give nearly double that it still drains at 3 minutes/percent
does anybody else have this problem