OverClocking 101

Status
Not open for further replies.
OP
Skull One

Skull One

Member
Joined
Mar 11, 2010
Messages
759
Reaction score
6
I recently changed P3 kernels from 125-800 to 125-1000 low voltage. I've noticed that "Time in State" notification in SetCPU no longer shows time in state for each of the slots. It is there something I'm missing to get the data?

It is an option that has to be compiled in at the time the kernel is made. You will have to ask P3Droid if he has decided to no longer support full stat collection.

I do know that cvpcs compiles his kernels with that option. You can find them here - sapphire:releases:2.0.0 [CCRoms Wiki]

They are the kernels I use daily and for testing.
 

hrdnhvy

Member
Joined
Mar 13, 2010
Messages
613
Reaction score
0
Location
WI
Haha...so much for my reply on the topic...lol...but Does the option included or not have any affect on overall kernel performance or on battery life?

I recently changed P3 kernels from 125-800 to 125-1000 low voltage. I've noticed that "Time in State" notification in SetCPU no longer shows time in state for each of the slots. It is there something I'm missing to get the data?

It is an option that has to be compiled in at the time the kernel is made. You will have to ask P3Droid if he has decided to no longer support full stat collection.

I do know that cvpcs compiles his kernels with that option. You can find them here - sapphire:releases:2.0.0 [CCRoms Wiki]

They are the kernels I use daily and for testing.
 

jerroedr

Active Member
Joined
Aug 14, 2010
Messages
1,192
Reaction score
0
so what happens with an up threshold of like 20?
 
OP
Skull One

Skull One

Member
Joined
Mar 11, 2010
Messages
759
Reaction score
6
Haha...so much for my reply on the topic...lol...but Does the option included or not have any affect on overall kernel performance or on battery life?

That sir is a damn good question. It is so good, I asked cvpcs to compile a kernel for me with "No stats collection what-so-ever". I had a rough idea it might extend battery life while in sleep/standby mode. Maybe up to 30% longer sleep time. At the end of the first 7 hours of testing, the battery read 20%. Just before that test the same battery, same phone setup and same usage pattern had lasted 19 hours when it hit 20%. Needless to say, I was dumbfounded. I just sat there and asked myself: Why?

So I started the test again and 7 hours later I had another nearly dead battery. It wasn't a fluke. There had to be something there. Then it dawned on me. The battery usage display only showed the Screen time, Wifi, Cell Standby, Android System, Android OS, etc, etc. And that is when it hit me. There were no Apps in the list. Because I asked for all the stats to be turned off. So the core part of the battery code only knew to calculate battery usage off of the "worst offenders" for battery drainage. And I am pretty sure the formula is weighted. Now that the Display usage was reading 60% instead of the normal 30%, it was predicting battery usage at a very high rate. So before I recharged the battery I changed out the kernel (while praying I wasn't wrong since the battery reading was 20% and flashing a Droid at that percentage is damn dangerous) and erased the batterystats.bin file to force the Android OS to poll the battery for it's true charge. Low and behold, it didn't read 20%. It read 50%. Right on schedule with the baseline test.

So the moral of this story is, we need stats collecting for the Android OS to predict battery life properly.

So now the question I am asking is "Which stats are needed and which ones can be turned off?". Well I can't keep bothering others to build things for me. So this year I plan to start building my own kernels and ROM to help facilitate my testing. A few friends are going to get me up to speed on what they know and then I can start narrowing things down.

And before anyone asks: I will NOT be building a custom ROM for public usage. There are already a dozen ROM makers out there and they know a hell of a lot more than I do about making a proper ROM. This will be for my testing only.

The only bad news in all of this is, my original Droid finally became unusable and had to be replaced Dec 31st. The refurbished Droid so far only shows one quirk my first one didn't but hopefully that won't affect testing since the issue is only during booting.
 
Last edited:
OP
Skull One

Skull One

Member
Joined
Mar 11, 2010
Messages
759
Reaction score
6
so what happens with an up threshold of like 20?

Technical answer: You will be in the higher frequencies most of the time and your phone will be really responsive.

Layman answer: You will run the battery down faster. :)
 

jhom

Active Member
Joined
Mar 28, 2010
Messages
406
Reaction score
33
Location
Dallas, TX
Thanks for the response. Now, I'm wondering why you have chosen the Sapphire over the P3Droid kernels? Any particular advantages or disadvantages of one or the other?


I recently changed P3 kernels from 125-800 to 125-1000 low voltage. I've noticed that "Time in State" notification in SetCPU no longer shows time in state for each of the slots. It is there something I'm missing to get the data?

It is an option that has to be compiled in at the time the kernel is made. You will have to ask P3Droid if he has decided to no longer support full stat collection.

I do know that cvpcs compiles his kernels with that option. You can find them here - sapphire:releases:2.0.0 [CCRoms Wiki]

They are the kernels I use daily and for testing.
 

hrdnhvy

Member
Joined
Mar 13, 2010
Messages
613
Reaction score
0
Location
WI
For roms that sapphire kernels work with they seem to rock, i had previously been running an earlier version of CM (not sure which one) and I got a sapphire kernel to work with it & i was very impressed, needless to say now im running UD & I cant get a sapphire kernel to coperate with it what so ever....additionally the sapphire kernel did provide all of the data collection p3 did not.
 

jhom

Active Member
Joined
Mar 28, 2010
Messages
406
Reaction score
33
Location
Dallas, TX
Good information. I'm currently running a stock FRG83D. Any known incompatibilities with it and the sapphire kernels?

For roms that sapphire kernels work with they seem to rock, i had previously been running an earlier version of CM (not sure which one) and I got a sapphire kernel to work with it & i was very impressed, needless to say now im running UD & I cant get a sapphire kernel to coperate with it what so ever....additionally the sapphire kernel did provide all of the data collection p3 did not.
 
OP
Skull One

Skull One

Member
Joined
Mar 11, 2010
Messages
759
Reaction score
6
Thanks for the response. Now, I'm wondering why you have chosen the Sapphire over the P3Droid kernels? Any particular advantages or disadvantages of one or the other?

It is mutually beneficial situation. cvpcs was interested in my work and I was interested in his. So he builds odd ball kernels once in a great blue moon for me and then he uses my test results to determine if the kernel is ready for release.
 

furbearingmammal

Super Moderator
Joined
Jun 16, 2010
Messages
11,081
Reaction score
363
Location
Anywhere you're not
Website
swdouglas.blogspot.com
Current Phone Model
32GB Moto X Developers Edition
Twitter
furryvarmint
For roms that sapphire kernels work with they seem to rock, i had previously been running an earlier version of CM (not sure which one) and I got a sapphire kernel to work with it & i was very impressed, needless to say now im running UD & I cant get a sapphire kernel to coperate with it what so ever....additionally the sapphire kernel did provide all of the data collection p3 did not.
UD has issues, but that's beside the point of this post.

I was running 125-1200 LV P3Droid and it had all the stats I could want.

Sapphire kernels should run fine on FRG83D provided your phone likes Sapphire kernels. That's the thing -- there are no sweeping statements that can apply in all cases because EVERY DROID IS DIFFERENT. However, Sapphire kernels should work well with FRG83D if you can get them to run because the motto of Sapphire seems to be "make the code as stock as possible; change as little as you can" when it comes to what they do.

On the other hand, in day-to-day use, my stock CM 6.1.2 install with stock kernel was/is almost exactly as fast (just not in Quadrant/Linpack) as the 1200 P3 kernel I was running. And talk about battery life improvement!
 

hrdnhvy

Member
Joined
Mar 13, 2010
Messages
613
Reaction score
0
Location
WI
I take back what I had said previously about Sapphire kernels & UD, out of curiosity I decided to give a Sapphire kernel a try again, low & behold I have the SV 125-800 installed & running fine at this time....
 

NetNathan

Member
Joined
Apr 14, 2010
Messages
239
Reaction score
0
Location
Corona, CA
Running BB 6.2....
Since it is not a ROM load...should I wipe data and cache before installing an overclock kernel?

..
 
OP
Skull One

Skull One

Member
Joined
Mar 11, 2010
Messages
759
Reaction score
6
Running BB 6.2....
Since it is not a ROM load...should I wipe data and cache before installing an overclock kernel?

..

No. You should do a backup just in case your phone doesn't like the kernel.
 

EbE404

Member
Joined
Feb 20, 2010
Messages
407
Reaction score
0
Hey, Skull

Not sure if you've had a chance to play with SetCPU V2.1 yet, but I'd be really interested to know what you think about the new version. Seems there have been some pretty significant additions to the code, with some nice GUI tweaks as well.
 
Status
Not open for further replies.
Top