Siyah Gingerbread kernel (v2.x)

[Kuragari]

Well it (2.2 beta 6) gave me the best performance I've had so far, so...having the option to stick with what works I guess...?

 

[Twinn]

so how come u will know without testing that others are not better ? u use 2.2 and latest is 2.5.2 so a lot has changed so far...

 

[Slug]

Try v2.5.2b if you value battery stamina, but make a CWM backup before flashing it in case your CPU can't handle the (mild) undervolting. Unlikely but some have had problems, as witnessed by the whining in Gokhanmoral's support thread at XDA. It's not as if they weren't warned.

 

[ooberandy]

I'm on 2.5.2b, and the battery seems great. I've fine with the settings that gokhanmoral recommended and have around an extra 15% battery at this time of night.

Deep sleep is 81%. Only thing I've noticed is my antutu has dropped about 1800 points. But I'm guessing this is because the processors are maxing at 800mhz.

I'm a proper noob, so don't mean this to be negative, as the kernel is the best I've used so far, for both battery and silkiness

 

[Slug]

Only thing I've noticed is my antutu has dropped about 1800 points. But I'm guessing this is because the processors are maxing at 800mhz.

Spot on!

I'm a proper noob, so don't mean this to be negative, as the kernel is the best I've used so far, for both battery and silkiness

Nothing negative at all - it's a good observation. I'm finding v2.5.2 very forgiving; I've been playing with different frequencies and voltages today and it's taken it all in its stride. It really has got something for everyone.

 

[ironass]

I
Since 100MHz and 200MHz both run at the same voltage there seems very little to gain, if anything.

using 100MHz will increase your battery life A LOT.

If you make a mistake, who better to correct you that the developer himself!

Siyah v2.6 changelog, (not kernel yet), is up!

Looks interesting.

"Changelog:

Moved some big files (which are not frequently used, i.e. CWMManager.apk, ntfs-3g, superuser files) to the end of the kernel image as a payload. They are uncompressed as needed (one time mostly). This change will decrease boot time.
Changed compression method from xz to gz which also decreases boot time.
Re-enabled the init delay of ondemand and some code changes to keep the cpu running at 1200MHz which will also decrease the boot time.
Configurable MALI thresholds
Configurable (more) hotplug parameters such as sampling rate.
New CWM flashable zip files:

 

[Twinn]

this changelog sounds really good! w8ing...

 

[gokhanmoral]

If you make a mistake, who better to correct you that the developer himself!

well, you were right about voltages.
just to clarify that further: decreasing voltage provides more power saving than decreasing the frequency if you decrease them the same PERCENTAGE.
if you could undervolt, for example, a 900mV default to 450mV then you would have 2 times less battery consumption on the core than decreasing the frequency by half.

BUT, you cannot decrease the voltage that much.

cpu's power consumption is not only related with the core freq and voltage.
but if we only consider the core, 100mV undervolting for 200MHz (~10%) can give you 23% saving whereas decreasing the frequency by 50% gives you 50%. even if you undervolt 200MHz to 800mV you will not gain that much. also it is very unlikely that you will have crashes by decreasing frequency if you do not undervolt .

 

[ironass]

Thanks gokhanmoral for taking the time and trouble to explain that.

Even I understood it... just don't make me sit an exam

It would seem that you put a lot of effort into the kernel development for the SGSII and I know I am not alone on this forum in appreciating your efforts as you will see from this thread.

You will also see that we avidly follow your every release in our own, good humoured, way and then discuss it on here, rather than cluttering up your xda thread, as most of us, I guess, are novices in the kernel world. That is to say that we may not fully understand everything you publish in your changelogs but we know what we like.

Three things I would like to ask are:-

1. You have, over the months, tweaked, changed and reconfigured the kernel so much that I was wondering if there comes a time when you are forced to say... I can do no more?

2. Following on from that question, would it irk you if someone like Samsung came along and said, "We like that Siyah kernel, lets use it."?

3. Last but not least, on a personal note, it has been a long running joke on here that if one of your kernels works really well for Sydney99, a fellow forum member, it will not work for myself and some others.
Therefore, is it possible for you to issue a variation of every kernel you release and just add the tag, say, S99. Then we can all encourage Sydney99 to flash that one so that the rest of us, not least of all me, can enjoy your official release without having the Sydney99 K.O.D. (Kiss Of Death).

Oh... and pushing my luck here, when is v2.6 likely to be released so that we may get on the starting grid.

 

[Sydney99]

Right, having established great performance and battery on 200-800, gonna give it a day our 2 on 100-800 and see if it makes much difference.

Don't worry ironass, am loving2.5.2b, will leave you to 2.6 (for now). But I do like the idea of the developer making me a personal kernel

 

[ironass]

Don't worry ironass, am loving 2.5.2b, will leave you to 2.6 (for now). But I do like the idea of the developer making me a personal kernel

You stick on the 2.5.2b for awhile old boy. It didn't work for me but I'm loving the 2.5.2 and am looking forward to trying the v2.6 which does look interesting.

You can wait for the v2.6 S99!

 

[Sydney99]

Surprised by the 2.6 update - gokhanmoral said above he was working ics and not to expect further updates soon.

 

[gokhanmoral]

Thanks gokhanmoral for taking the time and trouble to explain that.

Even I understood it... just don't make me sit an exam

It would seem that you put a lot of effort into the kernel development for the SGSII and I know I am not alone on this forum in appreciating your efforts as you will see from this thread.

You will also see that we avidly follow your every release in our own, good humoured, way and then discuss it on here, rather than cluttering up your xda thread, as most of us, I guess, are novices in the kernel world. That is to say that we may not fully understand everything you publish in your changelogs but we know what we like.

Three things I would like to ask are:-

1. You have, over the months, tweaked, changed and reconfigured the kernel so much that I was wondering if there comes a time when you are forced to say... I can do no more?

2. Following on from that question, would it irk you if someone like Samsung came along and said, "We like that Siyah kernel, lets use it."?

3. Last but not least, on a personal note, it has been a long running joke on here that if one of your kernels works really well for Sydney99, a fellow forum member, it will not work for myself and some others.
Therefore, is it possible for you to issue a variation of every kernel you release and just add the tag, say, S99. Then we can all encourage Sydney99 to flash that one so that the rest of us, not least of all me, can enjoy your official release without having the Sydney99 K.O.D. (Kiss Of Death).

Oh... and pushing my luck here, when is v2.6 likely to be released so that we may get on the starting grid.

I will change the name of v2.6 to v2.6_S99
and it will make sense because S99 is the last init script that is executed and I hope that Siyahv2.6 will be the last version with linux v2.6 base.

there are lots of things to change, fix and tweak in the kernel. there will always be bugs and settings to tweak. what I do is not very hard in fact because once my tweaks doesn't work for 10-20% of people I do not care about it much and I call them unlucky. but samsung has to make it work for everybody. because of that, there will always be new things to tweak once new sources are released. because of the last reason in the first answer, I don't think samsung will use my kernel. this is an open-source world and samsung does not have to notify me that he wants to use anything from the kernel sources (GPL) or even give credits.

for update3 base, there is not much left to change. Although I may be seen as a kernel developer by most of the people, I am not. I am just an embedded systems programmer. I tried to make almost everything customizable and now users can experiment by themselves. There are lots of people who call themselves "kernel developer" but they only put things together. When I was doing that my thread was in android development section. the first big original move was to make a unified kernel for both samsung and aosp based roms and then my thread was moved to original development section. I like to experiment and I am lucky because the user base is open to experiments.

 

[gokhanmoral]

Surprised by the 2.6 update - gokhanmoral said above he was working ics and not to expect further updates soon.

I am trying to solve a problem with v3.1
actually I reverted back to galaxy nexus source base and then I will repatch to v3.1
but meanwhile, I am making non-experimental changes in v2.6
none of those changes affect stability or battery life.

 

[ironass]

I will change the name of v2.6 to v2.6_S99
and it will make sense because S99 is the last init script that is executed and I hope that Siyahv2.6 will be the last version with linux v2.6 base.

Thanks for the explanation gokhanmoral.

There you go Sydney99... your very own kernel.

Of course, that means it probably won't work for me!!!

 

[mcarans]

According to CPU Spy, my phone spends 92% of the time in deep sleep and 5% at 100Mhz, which sounds quite good to me. What would be a normal case?

 

[ironass]

According to CPU Spy, my phone spends 92% of the time in deep sleep and 5% at 100Mhz, which sounds quite good to me. What would be a normal case?

Depends on your usage I guess.

Mines had a bit of a thrashing so far today with calls, internet and connected via Wi-Fi to MyPhoneExplorer.

 

[Slug]

it will make sense because S99 is the last init script that is executed

Ahh, this reminds me of something I've been meaning to research. You're just the person to ask.... in what order are the init scripts executed? Is S01sysctl run before 06tweaks, for example? How is a user-created script best named?

 

[Sydney99]

According to CPU Spy, my phone spends 92% of the time in deep sleep and 5% at 100Mhz, which sounds quite good to me. What would be a normal case?

Sounds a very reasonable set of figures to me

 

[ooberandy]

According to CPU Spy, my phone spends 92% of the time in deep sleep and 5% at 100Mhz, which sounds quite good to me. What would be a normal case?

I'm on 85% at the moment, with 2 % at 100Mhz

 

[Hawker]

Ahh, this reminds me of something I've been meaning to research. You're just the person to ask.... in what order are the init scripts executed? Is S01sysctl run before 06tweaks, for example? How is a user-created script best named?

The sequence is on a purely ASCII value basis slug, and unlike windows is also case sensitive.
So numbers come first, then upper case followed by lower case.
So if you issued an ls command in this folder, the order in which they are executed is the order the ls command returns.
In fact, there is a script that gets run at a higher level that does an ls of the folder and executes each file in turn.

ASCII character set

 

[gokhanmoral]

The sequence is on a purely ASCII value basis slug, and unlike windows is also case sensitive.
So numbers come first, then upper case followed by lower case.
So if you issued an ls command in this folder, the order in which they are executed is the order the ls command returns.
In fact, there is a script that gets run at a higher level that does an ls of the folder and executes each file in turn.

ASCII character set

Slug:
Hawker is right. in android, the approach is different than conventional linux systems. and also there may be some other differences depending on the kernel you are using.
for example,in hardcore's kernels, init scripts are executed by another script but in some kernels they are executed by using runparts command in busybox.
busybox is simple. it just executes everything by sorting and numbers come first.
normally, in linux systems, the init scripts which start with S is executed during boot time and the ones starting with K is executed during shutdown. there are also different run levels but we only have one runlevel in android. stock roms do not have init scripts. since init scripts are very custom in android, the execution order may differ.
I was writing scripts starting with S because of an old habbit but now I am using numbers.
just run ls -1 and the order is the execution order

 

[Hawker]

This is taking me back to my old UNIX programming days.
I'm showing my age now!

 

[Slug]

Hawker is right

Wow, that has to be a first!

just run ls -1 and the order is the execution order

Thank you both for adding a little bit more to my limited knowledge.

 

[turbohobbit]

I'd keep an eye on that old son. You should be getting a lot more Deep Sleep than that. Particularly if you weren't using it whilst you were asleep!

I'm on 72% Deep Sleep at the moment. I'm guessing that you have stuff running in the background but this is not my strong suit I'm afraid.

There are others on here who could advise you better.

I think I may have got to the bottom of this. I checked CPUSpy this morning and saw that my phone had been running at 100Mhz for over 7 hrs overnight. This worried me but then after reading some other threads - co-incidentally - it dawned on me that my wifi was set as "never sleep". I've changed phone to disconnect wifi when the screen is off, so I'll be interested to see what the deep sleep state is tonight.