• After 15+ years, we've made a big change: Android Forums is now Early Bird Club. Learn more here.

Root Root + OverclockWidget = Insane Battery Life 40+ Hours

keatre said:
It does help. But you are right, it must be idle. The only reason I'm continuing to run it is because I'll come home from a pub and forget to plug the phone in. In the morning, and two aspirin later, I can see the phone is still charged in its sleep mode. Without the widget, the battery would have been much lower (from my my own tests).

As I explained earlier, this does little or may hurt when the screen is on, but with the screen off, I believe it's able to streamline the CPU someway to improve battery life. If you use your phone non-stop until the battery dies, uninstall this widget. If you use it for the "bursts" and then let it sleep for hours, then I would advise playing with the widget to see if it helps at all.
Click to expand...

This makes some sense. Although correlation does not mean caus.. wait what?

Ok if its working as I intended the post. Excellent!

Here's my current best record with my EVO as posted in a screenshot.

35hours.png
 
nick325i said:
This makes some sense. Although correlation does not mean caus.. wait what?

Ok if its working as I intended the post. Excellent!
Click to expand...

I think what he's saying is, just because battery life may have improved some, we cannot scientifically prove that it is the widget because so many other factors can play a part.

As I said in an earlier post, I'm fairly sure the CPU is not being overwritten to what the widget tells it to, BUT it is 'streamlining' it when idle, causing battery life to improve.

Out of all my tests, (including airplane idle drain), I found this widget to improve battery life based off sleep times, but to drain battery life when the screen is on (all be at a very small amount, but still faster than without installed).

So again, if you use your phone constantly and let the thing die, then this widget probably won't help.

If you use your phone, sleep it, and wait for a new SMS, phone call, or a random instance to check something on Wiki/imdb, this widget may be beneficial when the phone is slept for long periods of time. If the phone sleeps for 2-5 mins, you won't see much improvement.

Questions, comments? Let me know.
 
nick325i said:
This makes some sense. Although correlation does not mean caus.. wait what?

Ok if its working as I intended the post. Excellent!

Here's my current best record with my EVO as posted in a screenshot.
Click to expand...


okay two questions.

1. which version of battery indicator are you using? i cannot for the LIFE of me get mine to display a timer. it just says unplugged at 730am. annoying

2. Taskiller. can you set programs that will NOT be killed when autokill is on? for instance, OCwidget, battery indicator, etc?

thanks!
 
First let me say, I've seen some serious increase in battery life out of this method. My juiceplotter graph practically flatlines when the phone is idle. It's even playing pandora great without stutter, when it's downclocked when the screen is idle.

Love it, and using it all the time. I'll need to continue gathering data on what exactly the increase is.

Now on to the sticky part. I was able to reproduce the battery indicator prompt saying "unplugged for X hours". Unfortunately, I only had it off the charger for 5.5 hours at the time. It's showing 34 hours. I'm not sure what went wonky, but here's what I did to reproduce:

restarted the phone yesterday (about 34 hours ago)
didn't start battery indicator
plugged the phone in about 6-7 hours later
kept the phone plugged in all night
took the phone off the charger, used it for the next 5.5 hours, though not heavily
started battery indicator
battery indicator shows 34 hours off the charger, in the form that normally it doesn't give

here's a screenshot. I'm not saying all/any of the screenshots are bogus, because I've seen some serious increases in battery life myself, but wanted to note this anyway, in case someone ran across it and called foul.

snap20100614132619.png
 
nick325i said:
This makes some sense. Although correlation does not mean caus.. wait what?

Ok if its working as I intended the post. Excellent!

Here's my current best record with my EVO as posted in a screenshot.
Click to expand...

I see that you're taking a jab at everyone that's skeptic of your method. I hate to say this but it seems like you have selective hearing. How do we explain the people who have tried your method but have seen no change in battery life?

Maybe we should stop worrying about whether or not you were right, and instead on WHY OverclockWidget saves battery IF it does anything?

I've posted a method to test whether OverclockWidget is able to touch the CPU frequency in any way here http://androidforums.com/all-things...nsane-battery-life-40-hours-7.html#post929496

The people who've responded seem to agree that No, OverclockWidget is unable to change CPU frequency on the HTC Evo 4G right now. Then in what other way does it save you battery? Don't you think that's at least an interesting thing to discuss? You shouldn't just blow us off like that.

And here's another hypothesis (just a theory!): You guys all seem to be running Task Killers. That has a huge possible effect on battery life, especially if you set it to auto-kill. I tested a task-killer before that auto-killed all apps when the screen goes to sleep. That may be a possible explanation for why idle battery life may seem to increase.

I'm just trying to come up with explanations for this increase in battery life, considering how OverclockWidget does not seem to be able to change the CPU frequency at all.
 
pingpongboss said:
And here's another hypothesis (just a theory!): You guys all seem to be running Task Killers. That has a huge possible effect on battery life, especially if you set it to auto-kill. I tested a task-killer before that auto-killed all apps when the screen goes to sleep. That may be a possible explanation for why idle battery life may seem to increase.

I'm just trying to come up with explanations for this increase in battery life, considering how OverclockWidget does not seem to be able to change the CPU frequency at all.
Click to expand...

ATK auto-kill was present in both rooted+widget and non-root, no widget tests from my account. The widget seems to be doing something.

I agree, I would like to know what is going on behind the scenes, since from my account, the widget appears to be doing something, but I can almost definitely say it is not soft-changing the CPU clock.

ALSO! I seem to have discovered the possible battery drain while on might be from SuperUser permissions. I checked applications total opening on the phone, and SuperUser had somewhere up in the 200's, only thing that happen to beat it was HTC Sense.
 
keatre said:
ATK auto-kill was present in both rooted+widget and non-root, no widget tests from my account. The widget seems to be doing something.

I agree, I would like to know what is going on behind the scenes, since from my account, the widget appears to be doing something, but I can almost definitely say it is not soft-changing the CPU clock.

ALSO! I seem to have discovered the possible battery drain while on might be from SuperUser permissions. I checked applications total opening on the phone, and SuperUser had somewhere up in the 200's, only thing that happen to beat it was HTC Sense.
Click to expand...

Haha you're right that theory doesn't hold much water. Are you using Spare Parts to check your usage? My home replacement app has the most usage as expected, and Superuser Permissions has only 1 count. Might just be something on your Evo.
 
I did the overclock widget frequency change, I have installed Auto Memory Manager and multitasking manager , is this going to conflict, I uninstalled the task killer app. Wichita one should I keep , let me know.
 
pingpongboss said:
Haha you're right that theory doesn't hold much water. Are you using Spare Parts to check your usage? My home replacement app has the most usage as expected, and Superuser Permissions has only 1 count. Might just be something on your Evo.
Click to expand...

I doubt it, since every time I turn the screen off, I think superuser permissions kicks in. I have spare parts installed, but I check it via the *#*#4364(orwhatever)*#*# code...

But you could be correct, might just be me.
 
keatre said:
I doubt it, since every time I turn the screen off, I think superuser permissions kicks in. I have spare parts installed, but I check it via the *#*#4364(orwhatever)*#*# code...

But you could be correct, might just be me.
Click to expand...

Could it be the every time you turn the screen off, OverclockWidget attempts to lower the CPU frequency and requests root permissions?
 
pingpongboss said:
Could it be the every time you turn the screen off, OverclockWidget attempts to lower the CPU frequency and requests root permissions?
Click to expand...

Yeah, thats where I was heading... Which makes me think thats why the battery decreases more significantly with the screen on, or rather, thats what I thought. That battery decrease spike might be when I hit screen off, superuser request.

Or maybe I'm looking for something out of nothing, dunno, haha.
 
pingpongboss said:
I see that you're taking a jab at everyone that's skeptic of your method. I hate to say this but it seems like you have selective hearing. How do we explain the people who have tried your method but have seen no change in battery life?

Maybe we should stop worrying about whether or not you were right, and instead on WHY OverclockWidget saves battery IF it does anything?

I've posted a method to test whether OverclockWidget is able to touch the CPU frequency in any way here http://androidforums.com/all-things...nsane-battery-life-40-hours-7.html#post929496

The people who've responded seem to agree that No, OverclockWidget is unable to change CPU frequency on the HTC Evo 4G right now. Then in what other way does it save you battery? Don't you think that's at least an interesting thing to discuss? You shouldn't just blow us off like that.

And here's another hypothesis (just a theory!): You guys all seem to be running Task Killers. That has a huge possible effect on battery life, especially if you set it to auto-kill. I tested a task-killer before that auto-killed all apps when the screen goes to sleep. That may be a possible explanation for why idle battery life may seem to increase.

I'm just trying to come up with explanations for this increase in battery life, considering how OverclockWidget does not seem to be able to change the CPU frequency at all.
Click to expand...

If a bullet proof vest stops a bullet and saves your life, do you cut it apart to explain why?

I'm following every bit of the discussion in this thread and find it very interesting. I have answered all questions and commented on any post directed at me (so far as I can tell). I have also e-mailed the developer of the widget asking him the skeptic questions about reported clock speeds and why the widget shows differently than cat /proc/cpuinfo. I also e-mailed flipz and toast for any input. I would post what they wrote but i haven't gotten a response from any of them.

There is no ultimate mystery here. OC Widget is doing something for some of us. Some good, some no result and some bad. Try it, if it works, thats great. If not, move on.

If it can't control the clock great. Whatever its doing it's owning my uptime plus many others in this thread and i'm fine with that without knowing exactly what its doing.
 
Perflock is an HTC driver on stock kernels that prevents SetCPU and Overclockwidget from doing anything. Set the speed, and it sets itself back immediately. I've been working on this stuff for months, and I am familiar with the exact lines of kernel code that does this. I'm certain what you're seeing here is a placebo.

I have developed an overclocked/undervolted kernel on which you will see a real difference in battery life (because we're physically pumping less voltage to the processor - the Nexus One sees huge gains from this). You can also actually overclock if you want. I've also enabled 128MHz as a speed and disabled perflock.

By the way, you can get SetCPU on XDA-Developers from my sig. It has quite a few more features.
[1.267GHz!!!] [Devs] Overclocking/undervolting patches + test zImage - xda-developers
YouTube - EVO 4G overclocked to 1.267GHz!
 
nick325i said:
If a bullet proof vest stops a bullet and saves your life, do you cut it apart to explain why?

I'm following every bit of the discussion in this thread and find it very interesting. I have answered all questions and commented on any post directed at me (so far as I can tell). I have also e-mailed the developer of the widget asking him the skeptic questions about reported clock speeds and why the widget shows differently than cat /proc/cpuinfo. I also e-mailed flipz and toast for any input. I would post what they wrote but i haven't gotten a response from any of them.

There is no ultimate mystery here. OC Widget is doing something for some of us. Some good, some no result and some bad. Try it, if it works, thats great. If not, move on.

If it can't control the clock great. Whatever its doing it's owning my uptime plus many others in this thread and i'm fine with that without knowing exactly what its doing.
Click to expand...

Hey I'm glad I got an honest response from you. Sorry if I'm pushing it a little hard here, but if you take a look at your responses on the last few pages to other skeptics, you put them down and really blew off any arguments that they had, telling them all that they were "missing the point" (1, 2, 3, 4). Well, that's not nice. Also, you had the support and thanks for a great deal of people (over 20, then over 100!!!), which I think got to your head (1, 2). The way you're talking now is a lot different than the way you talked to these people who were skeptical of the method. As someone that wants to find the truth, I can't help but speak up.

Let me apologize for digging into this thread so much, but I don't want people reading my accusations against you and thinking that I'm a total douche who has nothing better to do. You have a million people thanking you and you even have that nifty green GUIDE sign. This is my way of letting people know why I'm even debating in the first place.

Now, onto the issue at hand. You claim (and this method depends on the fact that) OverclockWidget forces the processor to scale down when the screen is off (1, 2, 3). If we had to prove otherwise, we'd have to somehow do a benchmark when the screen is off. Impossible, right?

Well I took a second just now to think about it, and it's actually quite easy to do. SetCPU has a LONG BENCH benchmarking feature that takes long enough for us to see if turning off the screen does indeed trigger the CPU to scale down. You may argue that the benchmark might still be too short, but at least we can all agree that OverclockWidget shouldn't have a NEGATIVE effect right? Well that's exactly what it turns out to be.
http://www.youtube.com/watch?v=xR_uTRW6n9g
  1. Uninstall OverclockWidget, Install SetCPU
  2. Set MAX and MIN in SetCPU down to 245. Test that SetCPU has in fact done nothing by hitting REFRESH a bunch of times (the # on the upper left goes up and beyond 245)
  3. Go into System Info and do some LONG BENCH with the screen on. Should get around 1000.
  4. Now do some more LONG BENCH, but immediately turn off the screen when the benchmark starts. You may get around 1000, or something even HIGHER. This means that the CPU is DOWNSCALING itself and the test is taking LONGER to complete. This is what you'd expect Android to do.
  5. Install OverclockWidget and set it up according to the OP. Make sure you save.
  6. Go into SetCPU and do some LONG BENCH with the screen on. You should get around 1000 like usual.
  7. Now do some LONG BENCH, but turn the screen OFF immediate when the benchmark starts. Wait for it to finish, and look at your score. You should get around 1000 still. However, in my tests, I have NEVER gotten a HIGHER number, meaning the CPU NEVER downscaled when OverclockWidget was active and the screen was off. This actually runs counter to what you claim, since if OverclockWidget some how did magically FORCE the CPU to downscale, then the number should be MUCH higher.
  8. I hope this has proved that OverclockWidget indeed does NOT force the CPU to downscale when the screen is off. Now people may argue that it does SOMETHING else to save battery, but until that happens, I think it is wrong to think that skeptics are just on the wrong track or "missing the point". If you've found these steps confusing, a Youtube video is on the way.

Hey I realize you were just trying to help at first, but it turned into a thing where "hey, if you're a skeptic, you're missing the point". Plus you've managed to convince soooo many people that this works, I think it's about time we figured out why.

Thanks for listening and I truly do appreciate your hard work. But I just totally believe that your method does not work. If OverclockWidget is in fact doing something, heck let's find it and use it, instead of pushing the side affect
 
pingpongboss said:
Hey I'm glad I got an honest response from you. Sorry if I'm pushing it a little hard here, but if you take a look at your responses on the last few pages to other skeptics, you put them down and really blew off any arguments that they had, telling them all that they were "missing the point" (1, 2, 3, 4). Well, that's not nice. Also, you had the support and thanks for a great deal of people (over 20, then over 100!!!), which I think got to your head (1, 2). The way you're talking now is a lot different than the way you talked to these people who were skeptical of the method. As someone that wants to find the truth, I can't help but speak up.

Let me apologize for digging into this thread so much, but I don't want people reading my accusations against you and thinking that I'm a total douche who has nothing better to do. You have a million people thanking you and you even have that nifty green GUIDE sign. This is my way of letting people know why I'm even debating in the first place.

Now, onto the issue at hand. You claim (and this method depends on the fact that) OverclockWidget forces the processor to scale down when the screen is off (1, 2, 3). If we had to prove otherwise, we'd have to somehow do a benchmark when the screen is off. Impossible, right?


Well I took a second just now to think about it, and it's actually quite easy to do. SetCPU has a LONG BENCH benchmarking feature that takes long enough for us to see if turning off the screen does indeed trigger the CPU to scale down. You may argue that the benchmark might still be too short, but at least we can all agree that OverclockWidget shouldn't have a NEGATIVE effect right? Well that's exactly what it turns out to be. I have a video that's uploading to Youtube that should clear this up, but I'll describe it in text here.
  1. Uninstall OverclockWidget, Install SetCPU
  2. Set MAX and MIN in SetCPU down to 245. Test that SetCPU has in fact done nothing by hitting REFRESH a bunch of times (the # on the upper left goes up and beyond 245)
  3. Go into System Info and do some LONG BENCH with the screen on. Should get around 1000.
  4. Now do some more LONG BENCH, but immediately turn off the screen when the benchmark starts. You may get around 1000, or something even HIGHER. This means that the CPU is DOWNSCALING itself and the test is taking LONGER to complete. This is what you'd expect Android to do.
  5. Install OverclockWidget and set it up according to the OP. Make sure you save.
  6. Go into SetCPU and do some LONG BENCH with the screen on. You should get around 1000 like usual.
  7. Now do some LONG BENCH, but turn the screen OFF immediate when the benchmark starts. Wait for it to finish, and look at your score. You should get around 1000 still. However, in my tests, I have NEVER gotten a HIGHER number, meaning the CPU NEVER downscaled when OverclockWidget was active and the screen was off. This actually runs counter to what you claim, since if OverclockWidget some how did magically FORCE the CPU to downscale, then the number should be MUCH higher.
  8. I hope this has proved that OverclockWidget indeed does NOT force the CPU to downscale when the screen is off. Now people may argue that it does SOMETHING else to save battery, but until that happens, I think it is wrong to think that skeptics are just on the wrong track or "missing the point". If you've found these steps confusing, a Youtube video is on the way.
Hey I realize you were just trying to help at first, but it turned into a thing where "hey, if you're a skeptic, you're missing the point". Plus you've managed to convince soooo many people that this works, I think it's about time we figured out why.

Thanks for listening and I truly do appreciate your hard work. But I just totally believe that your method does not work. If OverclockWidget is in fact doing something, heck let's find it and use it, instead of pushing the side affect
Click to expand...

If nothing good is happening, I guess we're done here. Enough people say it's not possible, it must not be.

Just seems odd, with all these results however poorly generalized or graphed in perfect detail showing in favor.

Is there a single line of code in the widget having an effect on the system?
Is it deeper than whether or not it can change the cpu speed?
 
nick325i said:
If nothing good is happening, I guess we're done here. Enough people say it's not possible, it must not be.

Just seems odd, with all these results however poorly generalized or graphed in perfect detail showing in favor.

Is there a single line of code in the widget having an effect on the system?
Is it deeper than whether or not it can change the cpu speed?
Click to expand...

There could be, definitely, but it deserves to be discussed. Now I feel stupid saying all that now that you've said this. But if you look at your post at the top of the page, it's hard not to argue with that.
 
pingpongboss said:
There could be, definitely, but it deserves to be discussed. Now I feel stupid saying all that now that you've said this. But if you look at your post at the top of the page, it's hard not to argue with that.
Click to expand...

Hard to imagine someone coming along and giving a clear answer from this point. At least about the one line of code part...

The author of the of the widget would be a good start but since he hasn't returned my e-mail. I got nothing.
 
nick325i said:
Hard to imagine someone coming along and giving a clear answer from this point. At least about the one line of code part...

The author of the of the widget would be a good start but since he hasn't returned my e-mail. I got nothing.
Click to expand...

I think it'd be really cool if we could figure out why. Hope they get back to you.
 
Ok guys, I've looked and haven't seen anyone w/ the same problem I'm having. I have both evorevoked and overclockwidget installed and I've set the parameter as seen in the pictures posted by users. Now when I click save I get a message saying "Operation failure, please check the su command!" I'm very new to this but assume that su stands for superuser. I had talked to someone saying that it made them check a box upon installation allowing for superuser access. I never got this option. Can I get to it somehow? I also deleted and re-installed and still nothing. Please help!
 
IMO Google should and will implement this in Android, sooner or later. Its like any OS, it uses less resources when doing less. Take Windows, when its sleeping (like phone being off), it shuts almost everything down. But the core OS scheduler will still wake up for any scheduled tasks (e..g check email, downloads etc on phone) and then go back to sleep.

This is not exactly rocket science. But a phone (even a powerful one like Evo) has much higher performance constraints so the effects become noticeable, which is why I suspect we haven't seen it. Trust me, there is a lot of scope for power optimization in Android.
 
You must log in or register to reply here.
Back
Top Bottom