I have setcpu running on desire with neophyte v1.2 ROM. It was running perfectly until I put it to charge overnight while sleeping.. when I woke up, the display shows that the phone is running at max speed constantly when it was set to "ondemand". I also have a profile with 245 min/max when screen off but when I wake up after 8hrs, I go to the info tab, I see that my phone is spending more time on max speed than 245 which doesn't make sense because the phone has majority of time with screen off..
What am doing wrongly here?
Sent from my HTC Desire using XDA App
webber2109 said:
I have setcpu running on desire with neophyte v1.2 ROM. It was running perfectly until I put it to charge overnight while sleeping.. when I woke up, the display shows that the phone is running at max speed constantly when it was set to "ondemand". I also have a profile with 245 min/max when screen off but when I wake up after 8hrs, I go to the info tab, I see that my phone is spending more time on max speed than 245 which doesn't make sense because the phone has majority of time with screen off..
What am doing wrongly here?
Sent from my HTC Desire using XDA App
Click to expand...
Click to collapse
need help....
what other profiles do you have?
i have 2 profiles.
one is 245 min 245 max when screen off (priority 100)
the other is 245 min 998 max when charging or full when charging (priority 50)
did i do anything wrong?
Has ur SetCPU got granted by SuperUser?
lpachuong said:
Has ur SetCPU got granted by SuperUser?
Click to expand...
Click to collapse
yup it is. in fact it was working properly until i leave it to charge. i tried uninstalling and re-installed but still it clocks at 998. the sleep profile doesnt seem to work well too
Try using a different ROM.
I experienced the same with the other ROMs when using the ondemand governor. I'm now using Nextsense 5.3 and the kernel allows "interactive" governor in SetCPU and my CPU speed stays at the lowest limits when idle.
amf said:
Try using a different ROM.
I experienced the same with the other ROMs when using the ondemand governor. I'm now using Nextsense 5.3 and the kernel allows "interactive" governor in SetCPU and my CPU speed stays at the lowest limits when idle.
Click to expand...
Click to collapse
do u have any profile for screen off? does it work?
from the info tab, i see my phone spent more time on max speed than min even though most of the time the screen is off
webber2109 said:
do u have any profile for screen off? does it work?
from the info tab, i see my phone spent more time on max speed than min even though most of the time the screen is off
Click to expand...
Click to collapse
Yup it works. Screen off profile--> Max=384 // Min=245
Try a ROM that supports interactive governor.
EDIT:
My "Time in State"
245 = 1352167
998400 = 18389
Hi,
I purchased SetCPU last night from Android market but I am unable to select the correct settings for it. I tried with 'Powersave' preset as I was going to sleep, but after 5 mins when I checked my phone it was off! and I had to turn it on. Every time I select Powersave mode it is forcing the phone to go off. Then I selected 'Ondemand' preset and went to bed at 0245 hrs. Today when I woke up at 1200 hrs ( ) the battery was at 48%, while yesterday night when I selected ondemand preset it was at 53%, it's a gap of 10 hours, in the mean time the Phone was in Flight mode and everything was set to OFF. So that's a 5% battery decrease overnight. But before that, when I was at all Default settings without SetCPU installed the battery was decreasing at a rate of 2-3% overnight under same settings. So I wonder after installing SetCPU the battery back up isn't extended but actually decreased. So I was thinking it may be a problem with my settings. Can you guys please tell me that what settings should I use to get the maximum juice out of my battery and that without compromising the clock rate of the CPU? Isn't there an Automatic settings that the CPU will run at Full speed when needed and check it's speed when in Idle? I thought 'Ondemand' preset would serve that purpose but I was wrong I guess!
Thanks in advance.
Here is a screen from SetCPU,
Regards.
I've got the same problem,any help would be appreciated.
Max 500 Min 200 and work from that adjust as needed .
jje
JJEgan said:
Max 500 Min 200 and work from that adjust as needed .
jje
Click to expand...
Click to collapse
Oh! And by setting the upper limit to 500 MHz we won't ever be able to use the CPU at 1200 MHz, great. I should've stick with my S8500 then, eh? What do you say?
ithehappy said:
Oh! And by setting the upper limit to 500 MHz we won't ever be able to use the CPU at 1200 MHz, great. I should've stick with my S8500 then, eh? What do you say?
Click to expand...
Click to collapse
aren't you eager to criticize. there are something called profiles. does your phone require to use 1.2 ghz in sleep , low battery conditions or overheating conditions? galaxy s2 uses ondemand governor at 200 to 1200 by default. meaning when it needs power it will ramp up to 1200. it does this in steps higher than battery saving governors so you spend more time in 1.2 ghz
my advice
sorted by priority
screen off = 200 - 500
in battery < 30 use 200 to 800
overheating = 200 - 800
charging = full speed
regular use when no profiles apply = full speed
What's the lowest you've set for Setcpu's 'screen off' profile? I tried 200 min/200 max and it didn't wake my phone when I tried turning it on. I have it set at 200 min/500 max right now. anyway to go lower than that? I'm trying to conserve battery as much as possible.
i got mine set to 200/500 mostly because 200/200 caused glitchy wakeups and occasional glitches when running background tasks with screen-off such as navigation/music/etc.
i also undervolted both 200/500 to 825mv/850mv so either way they both are barely using any of the battery
main reason for locking CPU to 500mhz max with screen off is in case something goes haywire and goes crazy on cpu, at 100% cpu usage @ 500mhz won't heat up the battery at all.
Just 200/500 for me, 200/200 at one point and no more than about 15 minutes later the screen wouldn't wake up.
Ok .. so, seeing many questions w/o real answers regarding the power consumption of the GN under various circumstances, I decided to create some testing conditions and generate some data.
The original idea and requirement started on Franco's kernel development thread but as this topic might be of interest I decided to create its own thread.
Test conditions:
- device: Galaxy Nexus i9250 - GSM version
- rom: bigxie's rom
- kernel: Franco kernel
*************************
1. IDLE test
2. MP3 play test
3. DISPLAY test
4. CPU Power test
5. Still image test
6. MKV play
7. Additional Display, MP3 play, YouTube and TuneIN media streaming
*************************
If you have any suggestions of tests you would like to see, please drop me a PM so I can make a list and implement them (depending on my possibilities).
*** Original post was heavily modified due to inaccurate measurements. The replacement posts linked here are the ones that matter ***
Why exactly did you decide not to run tests with wifi and 3G/2G, like grabbing and playing a youtube video?
That was the plan but seeing that I don't have enough resolution for this run, I decided to keep the test as simple as possible.
In few hours I'll do a more detailed run, with 1mA reasolution and I'll include online media too, as well as wifi versus 3g power comparison. It looks like a long friday night for me
// sent from my green robot //
ro_explorer said:
It looks like a long friday night for me
Click to expand...
Click to collapse
Haha! Well... I'm already looking forward to those higher res tests... don't forget to include 3G vs 2G as well, because I'm really not sure if there is much of a difference since you take so much longer to load... oh, maybe not a late friday but an early saturday?
thanks for doing these tests, its a great idea and cant wait for more results.
Yes. Thanks. Definitely need more.
My grandma beat me down and took my nexus. Sent from a jitterbug with beats by dre.
Thank you for support guys .
I've managed to build the new test circuit. I can now read the current with 1mA steps.
As a first set of measurements with the new setup - idle power .
Kernel and rom - see first post.
HotPlug function is a feature of Franco's kernel to turn off CPU 1 while phone is in sleep mode (in case some of you are wondering). It is supposed to be a feature to save power while phone is idle.
Voltage constant at 3.9V over all set.
1. AIRPLANE MODE - any governor, both 350-1200 and 700-1200 frequency range:
1.1. HotPlug OFF - 3 mA steady
1.2. HotPlug ON - 3/4 mA - the multimeter was unable to choose a value
So .. in airplane mode, the standby power is between 11.7 - 15.6 mW
2. 3G mode,WiFi - OFF, interactive governor
2.1. Freq: 350-1200, HotPlug - OFF - 4-8 mA
2.2. Freq: 700-1200, HotPlug - OFF - 4/5-8 mA
2.3. Freq: 350-1200, HotPlug - ON - 4-8 mA
2.4. Freq: 700-1200, HotPlug - ON - 4-7 mA
3. 3G mode,WiFi - OFF, conservative governor
3.1. Freq: 350-1200, HotPlug - OFF - 3-8 mA
3.2. Freq: 700-1200, HotPlug - OFF - 3-8 mA
3.3. Freq: 350-1200, HotPlug - ON - 3-7 mA
3.4. Freq: 700-1200, HotPlug - ON - 3-6/7mA
4. 3G mode,WiFi - ON, conservative governor
4.1 - 5GHz WiFi
4.1.1. Freq: 350-1200, HotPlug - OFF - 4-8 mA
4.1.2. Freq: 700-1200, HotPlug - OFF - 4-8 mA
4.1.3. Freq: 350-1200, HotPlug - ON - 4-7/8 mA
4.1.4. Freq: 700-1200, HotPlug - ON - 4-6/7 mA
4.2 - 2.4 GHz WiFi
4.2.1. Freq: 350-1200, HotPlug - OFF - 4-8 mA
4.2.2. Freq: 700-1200, HotPlug - OFF - 4-8 mA
4.2.3. Freq: 350-1200, HotPlug - ON - 4/5-8 mA
4.2.4. Freq: 700-1200, HotPlug - ON - 4/5-8 mA
When NOT in airplane mode, my phone has data and sync ON, and (over night at least) only Google services are being synced - gmail, g+, gtalk.
The measurement with data on/off, sync on/off are identical to the data on, sync on, when there is no activity that is. So, the the only way the phone will consume more power when data/sync is on is exactly when an event is being synced .. the rest of the time the radio being idle, identical with data/sync off.
Conclusion regarding IDLE (in my case at least): best power saving while phone idle is with WiFi off while screen off (or on only when plugged), conservative governor, 700-1200 freq. range and HotPlug - ON. Of course these values are experimental and are specific to my device but the numbers are looking reasonable.
I did not take any video of the above measurements because it was really boring .. put device at sleep and wait for it to really sleep (about 2 minutes for each test). But I'm attaching a photo of the test bench .
... to be continued with awake tests with video capture also ...
Awesome work!
I can think of one more scenario that's worth testing: when the screen is off but there is a wakelock keeping the CPU awake. This would give us a baseline of CPU awake but screen off, which would tell us exactly how much juice is going to the screen and CPU when the device is on.
Would MP3 play in background meet your expectations?
ro_explorer said:
Would MP3 play in background meet your expectations?
Click to expand...
Click to collapse
Wakelock but idle would be ideal I think, but actually an MP3 playing might be perfect if you can compare MP3 playing with the screen on in one of the previously tested states and MP3 playing with screen off. Then some simple subtraction should tell us everything we need to know.
I'd like to be able to calculate how much is used when the CPU is awake due to a wakelock, and how much power is saved by getting out of that wakelock ASAP to get back to power collapsed state. It might turn out that ideal screen off profile is 920 or even higher.
Good stuff. Glad someone is taking their time to collect this data.
Something to try: Loop a program to browse site content with white (light) heavy colors vs black (dark) heavy colors.
excellent info. if anyone's interested we did a similar test with the nexus one where we were able to use the actual resistor in the fuel gauge chip within the battery. so we did various testing scenarios, and that phone pulled very high currents. seems like our galaxy nexus is actually more efficient with some of those screen tests.
here's the link, prob boring though since it's not our shiney new GN.
http://forum.xda-developers.com/showthread.php?t=998544
This is my simple, yet efficient circuit (and quite precise I might say).
As you can see, the sense resistor is quite small not to interfere with normal operations and I think the 1% tolerance is making it perfect.
MP3 test
This is the mp3 failed test: http://www.youtube.com/watch?v=LOwnFcro0no
Why is a failed test? because the 3g, data and sync were on, so, beside the small 5-10 mA instability, there was a syncing action resulting in totally messed up readings while playing with screen off (1:42).
I hope now people will understand why these tests should be done in airplane mode. Having all the data at hand, one can easily add different values to get an approximation of his particular situation.
Now .. with airplane mode ON, I have a second mp3 play test: http://www.youtube.com/watch?v=IXHy0Ru45XU
Values from this second test:
~0.62W consumption with screen ON, autobrightness, showing homescreen
~0.72W when mp3 is started (70% volume, headphones)
~0.155W with screen off/playing mp3
~0.05W with screen off, no mp3 play and a software waiting for input (in this case, poweramp's lockscreen)
~0.011W in standby - airplane mode.
The above values are with 350-1200MHz freq range, conservative gov.
The frequency CPUs have used during screen-off play was 350MHz. If I'm increasing the min. frequency to 700, the average consumption is increasing with an average of 2-3mA.
ro_explorer said:
This is the mp3 failed test: http://www.youtube.com/watch?v=LOwnFcro0no
Why is a failed test? because the 3g, data and sync were on, so, beside the small 5-10 mA instability, there was a syncing action resulting in totally messed up readings while playing with screen off (1:42).
I hope now people will understand why these tests should be done in airplane mode. Having all the data at hand, one can easily add different values to get an approximation of his particular situation.
Now .. with airplane mode ON, I have a second mp3 play test: http://www.youtube.com/watch?v=IXHy0Ru45XU
Values from this second test:
~0.62W consumption with screen ON, autobrightness, showing homescreen
~0.72W when mp3 is started (70% volume, headphones)
~0.155W with screen off/playing mp3
~0.05W with screen off and a software waiting for input (in this case, poweramp's lockscreen)
~0.011W in standby - airplane mode.
Click to expand...
Click to collapse
Perfect. Thanks so much for taking the time to do this. I had been trying to come up with a semi-scientific way to measure some of this stuff and you just hit it out of the park with this.
Do you know what frequency the CPU was at when playing the MP3 with the screen off?
Display test
This is the hi-res (1mA) display test: http://www.youtube.com/watch?v=cqSQ5QE-4S0
Values (with device awake, of course):
- auto brightness, relative low light conditions, dark homescreen: 162mA / 3.87V = 0.14W
Maximum brightness:
- white screen: 380mA * 3.8V = 1.44W
- red screen: 287mA * 3.83V = 1.1W
- green screen: 295mA * 3.82V = 1.127W
- blue screen: 327mA * 3.81V = 1.246W
- gray screen (50%): 240mA * 3.84V = 0.92W
- black screen (but awake): 147mA * 3.87V = 0.57W
aron7awol said:
Do you know what frequency the CPU was at when playing the MP3 with the screen off?
Click to expand...
Click to collapse
No, but I can find out .. all I have to do is to turn on recording on system tuner pro application ... Let me check this and I'll update the MP3 test post with the cpu frequency data. I can tell you that I was on conservative, 350-1200 MHz settings, HotPlug ON.
ro_explorer said:
No, but I can find out .. all I have to do is to turn on recording on system tuner pro application ... Let me check this and I'll update the MP3 test post with the cpu frequency data. I can tell you that I was on conservative, 350-1200 MHz settings, HotPlug ON.
Click to expand...
Click to collapse
Thanks, my guess is that this CPU can play an mp3 at 350 MHz just fine
I've solved the mistery ... above post updated.
The cpu was in 350MHz state ... and (of course) if I'm setting the min freq. to 700, it will stay at 700. The 700MHz min value will add 2-3 mA on average during m3 play.
ro_explorer said:
I've solved the mistery ... above post updated.
The cpu was in 350MHz state ... and (of course) if I'm setting the min freq. to 700, it will stay at 700. The 700MHz min value will add 2-3 mA on average during m3 play.
Click to expand...
Click to collapse
Sorry to keep bothering you for more, but do you also know the CPU usage playing the mp3 at each frequency?
The final question I'd like to be able to answer is: Is getting to idle faster at the expense of using a higher frequency to get there beneficial?
I guess to answer that question we need to separate how much power the CPU uses at 100% CPU usage at 350 MHz, 700 MHz, 920 MHz, etc.
Hi all, first post here at xda, am fairly new to android devices and am wondering if someone could share some knowledge please? After installing my first Rom (m1ndh4x3r 2.3.5) I haven't been receiving any calls... A friend today said I may be underclocking the phone too much for it to wake on time to ring? Had it set at 200mhz with screen off. After setting it to 480mhz I received a call straight away, question is, have you had this problem too and what's the lowest setting advised that I can have it on please?
Thanks in advance for any help/advice
Your phone is built to run at 600 MHz. Not at 200. You can use apps like setcpu and no frills cpu control to control your phone. These apps let you set the minimum and maximum value of the cpu frequency. You should let it run at its full capability. Optimal settings would be something like
Min: 320
Max: 600
Governor: smartass2
You can underclock more when you are not using it or not expecting calls ( in the night).
You can also set your minimum to 120 MHz, but usually, it'll go into deep sleep mode, rather than stay on 120 MHz. Also, when listening to music, I've noticed, it stays on 320 MHz.
My minimum setting is 120 independently from which ROM (i tried 5 ROMs) I use (3month use = 0 problems).