Don't know if this has been answered yet or not but was just wondering why in the stock kernel and a kernel like chevy's under battery usage it will show the usage for individual apps but with kernels like p3's or slayher's it doesn't show battery usage for anything except display, cell standby, phone idle, etc but not specific apps. Has anyone else noticed this other than me and can someone enlighten me as to why this is?
DroidForums.net | Android Forum & News
This is a sample guest message. Register a free account today to become a member! Once signed in, you'll be able to participate on this site by adding your own topics and posts, as well as connect with other members through your own private inbox!
OverClocking 101
- Thread starter Skull One
- Start date
- Status
I am unaware of a program that captures all three of those data points.
Because the kernel wasn't built with all the options needed to track the data. They could fix that if they want. They probably just haven't been asked or considered it a priority.
The only way I know of is to use SetCPU, go to Info, click the Menu button and select Copy to Clipboard. Then paste it into the the notepad program of your choice. Unfortunately, you have to do it all manually.
I have wanted exactly this for some time. I don't know any app that does all three (or in fact any app that logs historical temperatures). The paid version of SystemPanel will do CPU activity, battery charge, and device usage. JuicePlotter will do battery charge. NetMeter will do CPU usage and net traffic.
I hadn't noticed that different kernels did not have this capability compiled in. I can confirm that the cvpcs kernel I am running also has it available (I see Facebook and Twidroyd in my usage screen). I don't remember the Slayher kernel I was running missing anything here, but I can't confirm this at the moment.
Last edited:
What has been your experience, so far?
Some questions for you, Skull:
I'm new to my phone, my first choice and what I'm currently running is the Slayher v1.6 1.2ghz kernel (LV?) and have been rather happy with it, getting a quadrant score of 1740, and a stress test in performance mode never gets the cpu/battery above 53/37c, but I plan to try some others.
What's the best way to compare battery life between kernels? And for error testing? You said it's hard to tell when ECC kicks in, would I notice a drop in framerate in a benchmark, then? My use is so erratic, that any comparison would be subjective right now.
Do you know what the voltage is in slayher's 1.6 kernels run at, by chance? Lower/higher/the same as Chevy's? I'd like to try his ULV kernels. Who else makes good "ULV" kernels? What kernel are you running, Skull, and why?
If I read your post on voltage multipliers correctly, voltage is not based on clock, but is constant? That a "1.2ghz LV" kernel is always lower voltage than a stock kernel? I think it would be interesting if someone could make a kernel, then, that ramps the voltage up and down with clock.
Are there any disadvantages to using, say, a 7 or 10 step kernel over a 5 step?
What does the "conservative" setting do in SetCPU?
What are some reasons for using a kernel with a higher base clock? I notice some start at 125, some at 250, 300, or even 400.
In regards to the interactive governor, I read that it works differently from on-demand (Slayher's anyway) by jumping to max clock and then scaling down, based on usage, rather than going up, step by step. Does it truly stay at top clock mostly, and not scale down very well? It seems to me that if scaling worked well enough, it would not use much more battery life at all, and give noticeably improved smoothness.
How can I tell if a program is preventing my phone from a full sleep?
Do some kernels play better with certain ROMs than others, and if so, why?
Thanks in advance!
Last edited:
I'm not Skull, but I think I can help you out with some of these based on his previous posts.
I believe that post was specifically talking about the multiplier at one clock speed. There is actually a separate voltage for each clock speed slot in each kernel.
More steps means more scaling options, so I'm assuming more slots should be better. I don't know how much testing has been done to see if there's additional computational complexity in scaling to more slots, but I think it's assumed to be negligible.
Before Froyo, there were issues with clocking down to 125, especially for waking the phone up from sleep. With Android 2.2, this is no longer an issue, but some people don't like change and would rather keep a higher minimum clock speed.
Grab the Spare Parts app from the Market, which gives you access to the hidden Battery History screen. When set to show "Other usage", the "Running" bar will show what percentage of time your phone is not sleeping. Clicking on the bar will give you actual time in minutes. You would generally expect the phone to be asleep somewhere close to the duration that the screen is off.
Regular usage is the only way to compare kernels at present. Some rough math to show why "proper testing" isn't feasible.
5 kernels makers * 6 different slot configs * 3 different voltages * (60 minutes to charge + 120 minutes to test) * 3 different tests (screen, 3g, cpu) = 810 hours of testing. And that is if you only do it using the stock ROM. Now you know why I haven't attempted this yet.
No, no idea, no idea and I don't like to talk about what I run personally because that might come across as an endorsement. This article is meant to only show what I use for testing and how you as the end user can reproduce the test if you are so inclined. There is no way for me to accurately or factually say "This developer is better than all the rest".
Actually that is exactly what all kernel makers already do. A 125 Mhz slot probably is at around 1000 mV, where a 1100 Mhz slot can be anywhere from 1200 to as high as 1400 mV.
Froyo 2.2 pretty much locks us to 7 slots. Anything more or less is now counter to what the Android OS has been setup for.
In the 2.1 days, it kept the CPU in the slower slots until the cpu idle level fell to low and then it would slowly ramp up trying to play catch up.
Under 2.2 it has changed radically and I haven't formed an opinion nor collected any hard data on it. I did test it for a week and with my usage, didn't see any difference in battery life. But, I could have also set it up improperly for my configuration. Which is why I still need to test it.
Ah yes the good old days prior to Froyo 2.2. The Android OS was too sluggish coming out of standby/sleep mode to justify 125Mhz. That stigma has carried over into 2.2. If the user improperly configures their phone for sleep/standby or have the wrong voltage selected, they will experience the same symptoms as the 2.0, 2.01 and 2.1 days.
This is a test I ran with the phone in standby mode: Post #287 - Interactive testing I feel the data speaks for itself
You have to track it manually as well as look at each applications permissions. Battery status, System Panel, Spare Parts can all be used to help you identify what your phone is doing.
The options that are compiled in to the ROM and the Kernel can be conflicting. But the leading cause of issues is the voltage of the kernel on that particular phone with a particular set of apps that are running at the time the phone is switching thru the slots. And one day I will figure out how to identify this issue. But at present, after over a 100 hours of testing I am no closer today than when I was back in April when I started my quest to find these answers.
Darkseider
Senior Member
Hey Skull or for that matter anyone I have a question. I have been running the Slayher 900 Mhz v1.6 kernel with the interactive governor and it has been running quite well. I decided to take a look at the sampling rate a few minutes ago and noticed it is "-1" and cannot be unset or for that matter changed back to default or higher than default. Any ideas?
JohnnyDepth
Member
It doesn't stay at -1....it will change depending on usage.
If you go back and forth from the main screen inside SetCPu then
go back to the settings menu where you see the -1 it will change.
I have set mine to 250 000.
Bilgediver
Member
Howdly, new to the forum, new to rooting and overclocking and all that. I was able to figure out what the different settings for SetCPU does (its actually fairly straightforward to someone who has a pretty basic knowledge of OS's, CPU's and such). I give mad props to the dev of SetCPU for making it fairly straightforward.
Its interesting that I've only just started thinking about downloading/flashing a different ROM, so I'm only running the officially FRG22D ROM with the stock kernel, rooted using this method. SetCPU booted with the following slots 125/250/400/500/550/600/700/800.
So far I've noticed it is smoother when loading, apps run smoother, but I can't comment on battery life, because I'm a really heavy user usually. I need a few weeks of normal usage before I can comment on that.
My profiles:
Its interesting that I've only just started thinking about downloading/flashing a different ROM, so I'm only running the officially FRG22D ROM with the stock kernel, rooted using this method. SetCPU booted with the following slots 125/250/400/500/550/600/700/800.
So far I've noticed it is smoother when loading, apps run smoother, but I can't comment on battery life, because I'm a really heavy user usually. I need a few weeks of normal usage before I can comment on that.
My profiles:
- Status