Database Users

[REF] bedalus' R&D compendium!

Over my time here in XDA I've built up a few studies on the Nexus S.
Here are the links. There are summaries in each thread in the first and second (and sometimes third) posts of the main findings, but I've done some very quick and minimal overviews here also.
How to build a kernel: this thread
-This post takes you through installing Ubuntu right through to managing git, in order to compile your own kernel, and flash it in CWM recovery!
-Includes the tools you need to add MTP capabilities to your kernel.
User eXperience Benchmarks: this thread
-This shows which ROM and kernel threads have the best user feedback. The feedback is processed by an external sentiment analysis API. It is only meant as a rough guide, something of a quick reference particularly if you are looking for a new ROM to try. As of June 2012 this is no longer maintained.
All ICS ROM Benchmarks: this thread
-Freely available benchmark programs were used to determine which ROMs had the best performance. Generally the difference between the best ones is very small.
Battery Drain Benchmarks: this thread
-A collection of tips for getting the best battery life, and dispelling some myths.
Kernel Features and Benchmarks: this thread
-My original study, the one that started it all, the kernel benchmarks! Provides a quick table to compare which kernels perform best in CPU/RAM/IO/Database/Graphics.
CPU Governors and I/O Schedulers: this thread
-A study done to find which schedulers perform best in combination with the most popular governors. My two favourites (ondemand and smartassV2) combine best with noop.
Power Saving Governors: this thread
-A meta study done combining the findings from the battery study and governor study to show which governors will save the most power. Ondemand comes out significantly better than conservative, which surprised me. SmartassV2 is the best compromise for high performance and efficiency.
Does SuperCharging work?: this thread
-This was just a short study to find out if this script is suitable for the Nexus S, but it turned out it had no effect, positive or negative.
Kernel Memory Allocators: this thread
-This is another short study for kernel developers to show that SLUB performs best.

====
This thread is just an index. Our new mod veyka agreed would be a good idea to summarise all my studies for addition into the sticky roundup.
====
EDIT: Feel free to suggest any new things that I might be able to test!

im out of thanks for today.. but, awesome work regardless. thank you

simms22 said:
im out of thanks for today.. but, awesome work regardless. thank you
Click to expand...
Click to collapse
Me too, but thank you too!
Edit: ah just use the app, unlimited thanks!

bedalus said:
Me too, but thank you too!
Edit: ah just use the app, unlimited thanks!
Click to expand...
Click to collapse
really? unlimited? i dont use the app though, just the browser.

Uh, well, I haven't tested the limit... but I never get denied!

All your work was great I battery drain thread was very helpful
Sent from my Nexus S

Thanks for all your hard work.

IF , one of my assassins had done their job ..........
.. Generated via Tapatalk ..

Oodie said:
IF , one of my assassins had done their job ..........
.. Generated via Tapatalk ..
Click to expand...
Click to collapse
When they finally come for me, I will benchmark their performance.

bedalus said:
When they finally come for me, I will benchmark their performance.
Click to expand...
Click to collapse
nice answer
And nice guide for your very instructive benchmarks bedalus

Even when browsing through the app you'd still get that annoying thanks limit.
This is a really good overview for new NS users. (I doubt there are ever some since mostly would go directly for G-Nex)
What's next?
That "Oodie's Assassins' Performance Benchmark" sounds interesting, though.
PS: @Oodie, you should have kanged your assassins before sending them to bedalus' compound. That way, your assassins' would have at least injured bedalus

UPDATE: Memory Allocators studied. SLUB wins!

Great reference for a refresher or for a new comer. Really like the battery drain benchmarks and how you discuss some myths. Governors is great...love Smartass V2 with Matrix V20!!
Thanks @bedalus!! Totally deserves a sticky.
Sent from my Nexus S using xda premium

Thanks. It's in the sticky roundup!

bedalus said:
Thanks. It's in the sticky roundup!
Click to expand...
Click to collapse
Can we vote for the sticky or it is just for the moderators?

There was like ten stickies, I think veyka wants to keep everything in a list in one sticky.
I appreciate the sentiment though!

bedalus said:
There was like ten stickies, I think veyka wants to keep everything in a list in one sticky.
I appreciate the sentiment though!
Click to expand...
Click to collapse
thank you for all your work, love your kernels ... currently been using deadline/ondemand but since you say it is not best I will try smartassv2+noop.. (I've found smartass with other combinations give me a lot of lag)

BLaZeDRas said:
thank you for all your work, love your kernels ... currently been using deadline/ondemand but since you say it is not best I will try smartassv2+noop.. (I've found smartass with other combinations give me a lot of lag)
Click to expand...
Click to collapse
Ondemand for good balance between performance and battery saving. Smartassv2 for gaming. CFQ and deadline are the best governors, but deadline probably has the edge.

Thanks! Now this widens my perspective on choosing my own Governor and Scheduler Combo.

[REF] Memory Allocator Benchmark

This one will be of more interest to kernel developers.
Link to Spreadsheet: Results
RcrdBrt asked me to see which memory allocator performed best, and provided me with four identical kernels except for the allocators being SLQB, SLAB, SLUB, and SLOB.
SLUB turns out to have a slight performance advantage, as RcrdBrt had suspected.
The rank was:
SLUB - 980.56
SLQB - 972.17
SLAB - 967.72
SLOB - 967.72
Not much difference really, but every little helps, as they say at Tesco (Britain's leading super market/Religion).
Thanks to RcrdBrt for trusting me with this nice little study, and to Chainfire for the very useful benchmarking app. (See here)
Methodology
Technical detail (it's not necessary to read this!)
I used CF-Bench to obtain values for:
Native MALLOCS
Native Mem Read
Native Mem Write
Java Mem Read
Java Mem Write
...and used statistical methods to boil 10 passes for each allocator down to one final score: 10 passes is enough to get the mean and median within one percent of each other, i.e. establishes a confidence in the mean, assuming normally distributed data. Three standard deviations are subtracted from the mean, to show the minimum score we'd expect 99.7% of results to be above. This is to penalize variability (high peaks are less significant than a good consistency). The geometric mean is taken of the five end results to provide one final score.

Bedalus can't live without benchmarks. Closed UX benchmark yesterday and opened a new one today
One more instructive and helpful benchmark buddy

Most kernel developers prefer using SLQB. Any thoughts on why this is the case? Do they have a special reason for that or just a matter of not knowing better? Cheers!

RcrdBrt said it was the fashion! Tbh i don't know the differences between them, but some casual googling showed that slub has had success in other Linux arenas

apatal said:
Most kernel developers prefer using SLQB. Any thoughts on why this is the case? Do they have a special reason for that or just a matter of not knowing better? Cheers!
Click to expand...
Click to collapse
Buzzwordism. The other 3 are in kernel.
Sent from my Nexus Prime.

morfic said:
Buzzwordism. The other 3 are in kernel.
Click to expand...
Click to collapse
You mean all those allocators are present but they mention only SLQB because it's the trend?
Sent from my Nexus S

apatal said:
You mean all those allocators are present but they mention only SLQB because it's the trend?
Sent from my Nexus S
Click to expand...
Click to collapse
Correct.
Sent from my Nexus Prime.

Man, bedalus is really fast!
@morfic:
So, in the ICS stock 3.0.8 kernel, how many and which allocators are present?
Am I correct to say that all those four allocators are present, but only one is used?
Sorry for these dumb questions.

glennkaonang said:
Man, bedalus is really fast!
@morfic:
So, in the ICS stock 3.0.8 kernel, how many and which allocators are present?
Am I correct to say that all those four allocators are present, but only one is used?
Click to expand...
Click to collapse
SLAB, SLUB, SLOB are in most (mainline) Linux Kernel, SLQB is patched in.

bedalus, did you run this on GB too?

morfic said:
bedalus, did you run this on GB too?
Click to expand...
Click to collapse
I can do it if i have 4 identical gb kernels (except for the allocator) ... If you'd like it testing you'll have to build them for me?

I/O preformance is horrible

I ran androbench and this tablets i/o benchmarks are terrible!
How is there such a huge gap between the tf300 and the other devices used as references in this benchmark.
I am oc'd to 1.5 and have the sio scheduler enabled using atp tools.
Is this a hardware or software issue? Or am i misunderstanding these benchmarks and making something out of nothing?
Is there any thing that can be done to improve this?
Just figured id post my screenshots and hear some other opinions

can confirm this. Stock I/O is bad. In Quadrant about 300points behind custom Roms.

Well, are you guys happy with the performance of the TF300?
Sent from my ASUS Transformer Pad TF300T using xda app-developers app

This is exactly what I was suspected, really hope its not caused by cheaper/slower integrated SD storage chip :-/
There is a real impact on performance when accessing internal storage, my TF300 goes to a complete halt if I download using Torrent to internal storage. However when I direct downloads to my external class 10 32gb SD-card I can still somewhat use my pad for surfing.
Skickat från min Transformer Pad TF300T via Tapatalk 2

KindaUndisputed said:
Well, are you guys happy with the performance of the TF300?
Sent from my ASUS Transformer Pad TF300T using xda app-developers app
Click to expand...
Click to collapse
Its a nice tablet, and for the most part yes but, a new $400 device should not be dead last in preformance on any test. And its not just a little behind, its so far behind on some of the tests it almost dosent show up on the bar chart. I would just like to know why this one aspect of the tablet is so much worse than other devices when all the others are on par or better then the competition.

This thread is asking the same question about the TF700: http://forum.xda-developers.com/showthread.php?t=1766836
So far, signs are pointing to slow flash storage chips, which can't be fixed. Hopefully, someone can provide a good answer.

This is expected.
See my reply here,
http://forum.xda-developers.com/showthread.php?p=27733776
Sent from my ASUS Transformer Pad TF300T using xda app-developers app

I am very happy with it. As I said its a Stock rom issue. I don't use Stock, so I have better I/O
I don't trust benchmarks so much. It. has to feel good, that's what counts.
Even with the lag of. I/O its a very fast tablet. I Am about 5200 in Quadrand and about 13000 in Antutu

Well im stock rom but using sio not noop. It did help a little. How much of a difference does using cm9 or aokp make? i dont really want to void my warranty untill i use it for a while to make sure its not going to crap out on me.

if you use rom toolbox you can actually boost the I/O i believe up to 4096 kb which should improve it.
Sent from my ASUS Transformer Pad TF300T using XDA Premium HD app

Rom toolbox shows it at 4096 i set it anyway and the benchmarks are worse

This is interesting.. I've noticed accessing the internal laggy at tines too. Like whenever I'm installing apps from Google play it is horribly laggy yet I can go on my galaxy s2 update 10 apps at once and not even notice it in the background
Sent from my GT-I9100 using Tapatalk 2

Would running a custom ROM either help or fix this issue?

Well its either hardware or software. If its a hardware limitation then no, there is nothing that will drastically improve it. If its. Software then drivers or a kernel update could. At least thats how im seeing it.

AFAIK the Prime also suffers from the same crappy I/O performance. Also, I have read that the performance can be boosted up with custom kernels, so far there aren't any kernel devs working on the TF300T.
Using ATP_Tweaks changing the scheduler from noop to SIO has almost doubled my results in Reading and Writing Speeds, still the numbers suck (it went from 300-400 to 500-600).

always run a bench 5 times and get the average.. there is too much that interferes with these benchmarks especially the IO onces ... 1 background write and youll see a big drop in the score of the running bench...

Just ran antutu on stock...
Running in balanced mode.
8177
Data IO= 390
SD write= 7.1
SD read=18.2
Then installed atp_tweaks and changed to sio...
8065
Data IO=405
SD write=7.5
SD read=48.4
Only did a single run for each though.

Just run a quick test with the latest Skank. Just run it one time.
But as I told before Bench is not so useful. For me it all comes to the way the Rom feels. Never felt any lag in one of the Custom Roms. Can't tell anything about Stock, because I only had it on for about 10 minutes

Having your cake and eating it too

Most of life is a choice between one or the other but with the Nexus 7 along with Root, Custom Recovery, cm10.2 unofficial, Faux Clock Kernel Control App and Jassy Release 5 Kernel it might just be possible to have both. Jassy has managed to put together a kernel that combines a 1.9 GHz over clock and a 487 MHz GPU over clock. With a global undervolt of -24 MV I managed to pull off stunning benchmarks, smooth lag free performance and incredible battery time. Just under 13 hours battery time in fact. I did this with a YouTube kung fu movie fest for about 12 and a half hours and a half hour of benchmarks and downloading a few apps. Remember if you game you can not expect this kind of battery performance but it does do great at that as well. I've attached some screenshots of my settings so you can reproduce these results if you would like.
Remember when you Root, ROM and Over clock you do it at your own risk and please thank the fine Developers who gave their time to make this kind of fantasy possible. I in no way had a hand in developing this software, I just took the time to do many benches and battery tests on many Roms and kernels then undervolted a bit.
Sent from my Nexus 7 using xda app-developers app

Benchmarks
Sent from my Nexus 7 using xda app-developers app

Nice, will have to check this out
Sent from my Nexus 4 using xda app-developers app

So to replicate your results, we need to flash cyagen, and jassy release 5, and then adjust the settings to replicate yours, or does jassy's release already have some of those settings in it?
What other apps did you download?
Getting longer battery life and a faster tablet is quite attractive!
EDIT I just installed jassy r5, and my quadrant is only 5679.. is something wrong, or do I need to also do the tweaks you listed here first?
EDIT 2: I've set faux to the same setting now, but am still using cleanroom1.3, and the quadrant results are 6787

Gareee said:
So to replicate your results, we need to flash cyagen, and jassy release 5, and then adjust the settings to replicate yours, or does jassy's release already have some of those settings in it?
What other apps did you download?
Getting longer battery life and a faster tablet is quite attractive!
EDIT I just installed jassy r5, and my quadrant is only 5679.. is something wrong, or do I need to also do the tweaks you listed here first?
EDIT 2: I've set faux to the same setting now, but am still using cleanroom1.3, and the quadrant results are 6787
Click to expand...
Click to collapse
Sorry I never saw your posts. Never looked at the thread again. To replicate my results you would have to download cm10.2 and Jassy release 5 and faux kernel control app paid version and set exactly like in the screen shots. I have since made a few more refinements but nothing you couldn't play with yourself. Make sure you enable mpdecision. Make every screen look like mine.

Fiops as I/o and 62c 82c as intellithermal is the only settings changed. Bout 7 hours of YouTube and 5 hours of heavy browsing. Couldn't ask for better battery life. NO Gaming.
Sent from my Nexus 7 using xda app-developers app

Yep, I did duplicate them, but couldn't get quadrant scores anywhere near yours. (they were better though.. it might be a cleanroom issue).
I upgraded to Jassy rev 6, ran into issues, and then just reflashed the stock kernel.
I was hoping we'd see Jassy rev 6 lockup issues fixed, but its been a week now. I'm also read of a few misc cm issues, and I'e been waiting for them to clear up as well. Cleanrom isn't having any issues at all, so I've stuck with it for now.

conan1600 said:
Fiops as I/o and 62c 82c as intellithermal is the only settings changed. Bout 7 hours of YouTube and 5 hours of heavy browsing. Couldn't ask for better battery life. NO Gaming.
Sent from my Nexus 7 using xda app-developers app
Click to expand...
Click to collapse
How's stability with your set-up?

Gareee said:
Yep, I did duplicate them, but couldn't get quadrant scores anywhere near yours. (they were better though.. it might be a cleanroom issue).
I upgraded to Jassy rev 6, ran into issues, and then just reflashed the stock kernel.
I was hoping we'd see Jassy rev 6 lockup issues fixed, but its been a week now. I'm also read of a few misc cm issues, and I'e been waiting for them to clear up as well. Cleanrom isn't having any issues at all, so I've stuck with it for now.
Click to expand...
Click to collapse
Cleanrom will not bench as high. Its a stock based ROM. Truth is though once your this high your not gonna see a whole lot of improvement in daily performance unless you go looking for it. Cleanrom is one of the best Roms I've ever used and there is absolutely nothing wrong with sticking with it. I go back and forth between the 2 myself. As themes and tweaks come out for nexus 7 2013 you'll find cleanrom will get even better than it is now also. Wish someone would do a tablet ui lol.

dkb218 said:
How's stability with your set-up?
Click to expand...
Click to collapse
I haven't had a single random reboot. Chrome works real well as long as you can change it from beta in lower left of xda page. I haven't overheated but I gotta tell you I'm no hardcore android gamer. Some angry birds from time to time or subway surfers lol. I have probably gamed for about an hour or so on these settings and not gone into a state of panic about temps though. Any longer than that or some other game and I couldn't say. in fact FOR me things are so smooth on this machine I had to order a new HP to tinker with lol. Seriously this is the best android machine I've owned. And buddy, I've had em all. I also use adaway so that may help with chrome as well.

Impressive scores man.
Yeah hesitant to jump to other roms myself at least at this time. Everything is so new finding a "good one" is hard to do at least yet. Though I've used CleanROM's on other devices I've had in the past and love them. Personally if CleanROM comes out with a 1.4 I might consider jumping into it.

Thanks for sharing! My only suggestion would be for you to recommend another kernel tuning app other than Faux's app. I'm just personally not a fan of someone who does nothing but kang, and even then still doesn't credit who he is kanging from. Too many hard working honest developers for me to support a faux developer (yes, pun intended. Maybe that's where the screen name came from?).
Just my $0.02.…
EDIT : For anyone questioning the integrity of my comment, just do some research and you'll see what I'm saying
SwiftKey'ed from my '13 Nexus 7 FHD using XDA Premium HD

stanglifemike said:
Thanks for sharing! My only suggestion would be for you to recommend another kernel tuning app other than Faux's app. I'm just personally not a fan of someone who does nothing but kang, and even then still doesn't credit who he is kanging from. Too many hard working honest developers for me to support a faux developer (yes, pun intended. Maybe that's where the screen name came from?).
Just my $0.02.…
EDIT : For anyone questioning the integrity of my comment, just do some research and you'll see what I'm saying
SwiftKey'ed from my '13 Nexus 7 FHD using XDA Premium HD
Click to expand...
Click to collapse
I have absolutely No interest in politics on my nexus 7 2013. if he did wrong, shame on him however I have tried each and every kernel control app on the market and reported my best results. I have yet to produce scores or battery life comparable even within 10% of fauxclock/Jassy R5 combo. I do not intend to do any research at all about whatever faux123 did or didn't do. I understand a personal grudge and respect your right to hold one. If your looking for something as close as you can get to his kernel control app try trickster. I didn't have a lot of success with it but you might find a combination that works close. If I were in the middle of whatever happened between faux123 and you or whoever I might have formulated an opinion. I'm not chiding you either. You brought an issue before me and I am simply stating I don't want involved as its in my humble opinion not my place or my fight, not to mention it appears it was already fought.
if I simply hated and didn't use a product based on someone else I don't know having a problem with that individual I wouldn't be able to buy anything, at all, anywhere, on earth, or in heaven, ever! If you tell me he harmed a baby and I see he was convicted of it in a trial of his peer's elected in a fair and balanced environment I will most certainly NOT have anything to do with him. But unproven or conflicting information, that may be due to personal interpretations is difficult for me to get involved in. Especially after the fact. And then there is the whole, if a man was punished for his crime do you continue to persecute him after he paid his dues thing.

Are there any deal breaking issues in CM? If not, switching does not bother me at all, and getting better all over performance is worth while.
I just don't want to spend more time tinkering with my tablet than actually using it.

What's your binning?

appreciate you sharing your jassy 5 set up with the fauxclock app. im on jassy 5 as well and have been running the setup that jassy posted but now im gonna try your set up. i just got one question about it: how is your sleep mode stand by time? i lose 1-2% in 8 hours

Gareee said:
Are there any deal breaking issues in CM? If not, switching does not bother me at all, and getting better all over performance is worth while.
I just don't want to spend more time tinkering with my tablet than actually using it.
Click to expand...
Click to collapse
I personally would call the current cm build almost RC material. I have no problems with it at all.
My bin was 14/2 if I'm not mistaken.
I probably lose a bit less than that but can't say for certain as I constantly use it lmao. I do get exceptional screen on time really. There are some screen shots on previous page I think.

Cool. Might have to take the plunge. Do you have to wipe to install it? (I didn't have to wipe for cleanrom.)

Gareee said:
Cool. Might have to take the plunge. Do you have to wipe to install it? (I didn't have to wipe for cleanrom.)
Click to expand...
Click to collapse
You have to wipe on a first install.

conan1600 said:
You have to wipe on a first install.
Click to expand...
Click to collapse
Ugh.

[Experience][Share]My Usage and Testing of Custom kernels for Touchwiz Kitkat.

To the moderators : I am making this thread on the request of members and not trying to promote any rom or kernel.
--------------------------------------------------------------------------------------------------------------------------------------------------------------------------​​
To all the readers : This is just a sharing of Experience of mine, not to be mistaken as a thread for promoting any kernel or rom. Also, all the posts asking which is the best rom/kernel or saying that this kernel is the best and recommending someone just because you like it, shall be reported to the moderators. So be careful of what you post. Also, don't spam the thread by writing just thank you, thanks etc. there is button on the left bottom of each post for that purpose.
---------------------------------------------------------------------------------------------------------------------------------------------------------------------------​​
Post 1 : General discussion of experience, crashes etc.
Post 2 : Link to the kernels
Post 3 : FAQ(to me )
Post 4 : Important Must Read stuff.
Click to expand...
Click to collapse
[size=+2]The Beginning[/size]
After much waiting and ETA posts by users, when DN3 v5+ was launched, it was met with quite some positive reviews, and being a loyal user to that rom since v2(why, you ask. It's because I like stock like looks that come with my phone, but under the hood optimizations and available customizability and features).
And just in time, Samsung had released the kernel sources for Touchwiz kitkat, so everyone including me were waiting for a good custom kernel, which would allow us to tweak our device to a further extent. During some random discussion @singhpratik87 informed me about an upcoming custom kernel called Nadia kernel, which was made by a small group of devs and modders calling themselves as EmoTdroiD Team.
So, as experiment hungry scientist for my device I am, I went ahead and flashed it. After just some time, I saw that it wasn't yet fully mature and still under construction, although it gave me a good amount of tweaking with the Trickster Mod and a usable battery life, so I went on to search for other kernels too, which would be satisfying my needs, and I found AGNI and Phoenix kernel compatible with my device and rom.
When I started using those, I noticed that many users wanted to know of a good custom kernel they could use, and thus kept spamming different threads asking questions, so I got an idea that I will test the kernel thoroughly and do a proper writeup, so users can be better informed of their needs.
[size=+2]The testing begins[/size]
As I had to begin the test and be thorough with everything I quickly bookmarked some guides to reading a logcat, downloaded tools on PC to capture logcats from boot to shut-down, downloaded each and every kernel and recommended fixes(if any).
I thought of beginning with my stock kernel, since it is the one deemed as the most stable. I got a good battery life but not the advanced tweaking. So, at first I decided to test the most near to stock phoenix kernel, which doesn't have much customisability but is quite stable(for me) for my daily use, but I always have wanted advanced tweaking, so I decided to go for the highly acclaimed AGNI kernel, followed by the under development Nadia kernel.
[size=+2]Stock kernel and Phoenix kernel[/size]
Obviously a starting point for testing the rom itself, I started with the stock kernel. My governor for all the time was pegasusq due to it being optimized properly for multicore systems, but I realized that my music used to have a bit of stutters every 15-20 seconds, so I read and read and found the cause, which you can find on the 31st page of the Nadia Kernel thread(don't be lazy, search for it, I have at least provided the link for the thread ), so I set up profiles suing Xposed module performance profiles and had the switch to ondemand governor whenever I wanted to listen to music, and that made for a smooth experience for me.
As the technology on device gets faster and better, there is one thing we crave for - battery life. Although I got 5 and half hours screen time with 24 hours sage on stock, I still wanted more, seeing that people managed to get upto 7hrs screen time with that usage(I don't know whether they listened to music for at least an hour everyday, though). What can be the best way for getting a better battery life? Underclock? not any way, I want my device to perform to it's full potential rather than be a super sluggish long lasting one. Any more guesses? Undervolting, you say? Yes, that's just what I needed, CPU/GPU undervolting and better control on the governors and LMK(low memory killer).
[size=+2]Reading, Researching, Writing and considerations of various possibilities[/size]
Many of the users who have known me from the time I've joined XDA know that I'm of the kind who is willing to spend his time in reading and solving problems on his own, rather than ask questions repeatedly, so that's what I did again - start reading :cyclops:
I read about the problems faced, questions asked, solutions provided, suggestions given and the proper changelog and FAQs for the kernels.
I researched about how users feel with a kernel, I saw their comments on the threads, I read about their problems faced, their attitude towards the developers and fellow members(which can turn out to be a bit-off when someone is unable to find a solution and gets frustrated) and what they have done to obtain something or how they managed to correct their problems.
Now, after compiling all the things I read and researched about into a small 30+ page word document I wrote to myself, it was time for me to flash a good custom kernel.
Again, which do I go for first? I had to spend my time thinking about the possibilities taking place and the issues I read about :
1) AGNI had some problems with downloading of some apps(Fixed now, but was present when I had to test).
2) AGNI doesn't play well with folder-mount(Dev is working on fix).
3) Nadia gave a bootloop when flashed over AGNI ( as reported by @NickVXD).
4) Some users lost root access when flashing Nadia(Rare and fixed in latest R4).
5) Nadia uses Trickster Mod and thus doesn't provide a full fledged GUI for tweaking the parameters( Dev plans to integrate the best app for this, Synapse, in a bit of time).
[size=+2]It begins with AGNI[/size]
After much thinking, I decided to go for AGNI first, because a bootloop and lost root aren't a big issue for me, as I always keep the fixes with me and know how to recover from these small issues.
I made a nandroid backup, in case for a bootloop, flashed the kernel and wiped cache and dalvik cache and rebooted and I was greeted with my lockscreen(yes it booted successfully) and AGNI control app in my app drawer. I began my testing for the lower clocks, conservatively ramping governors, high undervolts( my device can bear upto -80 on CPU and -75 on GPU while still being stable) and lower GPU clocks, because the first thing I wanted was battery life and indeed I got it.
Then, since I was at home only for a long time(many know why) and had thge charging port next to me(lucky me ) so I decided to give a try to other settings as well. The first thing I wanted for my device was a good read speed for my internal and external SD, so I benchmarked and logged performance for each scheduler over an extensive period of time and found Row suited my needs, because I don't write lots of stuff onto my device's memory, but rather need higher priority for reading of the already present memory blocks.
I tested with different governors now and the name of each is self explanatory so I won't tell which gave me the best battery life or performance. Anyways, I tested the smoothness, battery life, framerate in general use to heavy games, device temperature and noted down each and every value thoroughly.
I benchmarked the SD card read write with different cache sizes and noted down each value.
I tested different LMK minfree values and wrote down the results to myself.
I tested the power saving load balance for performance to battery ratio and wrote down the results thoroughly.
I used the 512mb preloadSWAP and checked and wrote down it's results.
I tested the LZ4 ram with different values and wrote down my results. Did the same for Swappiness priorities, and FRandom RNG entropy generator.
I checked the effects of Dynamic file sync and filesystems checking on boot and again, wrote down.
[size=+2]Shifting to Nadia[/size]
I was done with all the possible combinations of settings with AGNI and now it was time for Nadia kernel. I had been active at the thread for quite some time, helping users and even requesting features.
I read some complaints about lost root and bootloop if used after AGNI, and indeed I got a bootloop once, but naturally, as I am not of the types who would just go ahead for anything, I kept a nandroid backup just before flashing, and then I restored my backup, and some user mentioned using the AGNI script cleaning tool for driving away the issues, so I downloaded that and read through the updated script, and then after I was done I decided that I trust my own experienced hands rather than automated tools, so I removed the required init.d scripts from /system/etc/inti.d (Be careful not to remove 00test and 99superSU daemon, as these are needed for proper functioning of root rights).
After I was done, I flashed the kernel, but a bootloop again , silly me realized that I forgot to wipe cache and dalvik cache, which actually I understood from the logcat, so I went to recovery and did that and my phone was up.
After all was up I opened Trickster Mod and went on to customize my kernel and test different settings.
I did the tests with the available settings and wrote down the results like I had done before.
Contrary to as some users had mentioned, I didn't experience lost root, so all was fine for me, and after the R4 release of the kernel some days ago, no one has complained of any such error.
[size=+2] Finalizing [/size]
After I was done with my thorough testing, capturing logcats on hourly basis and for each system crash or reboot(rare) and compiling of a 60-page word document for each kernel's testing(except for stock and phoenix, which gave me no more than 15 pages :silly: ) it was time for me to report everything, as I had promised to various users.
I read through my writings, preserved logcats and memories and collected everything to write up and express this testing in the best manner I could manage(I usually turn out to be short tempered and rude, so it was another challenge for me).
I also had to decide which kernel to keep for my daily driver(I won't report that here, and neither will answer it via PM, though there are exceptions for the PM one, maybe @smeet.somaiya and @singhpratik87, because I trust them for this and I and they both know why).
[size=+2]And this small article is the outcome of all what I had done over the course of the time I had spent[/size]

Links to the Kernels
Phoenix Kernel - http://forum.xda-developers.com/showthread.php?t=2759906
AGNI Kernel - http://forum.xda-developers.com/showthread.php?t=2311094
Nadia Kernel - http://forum.xda-developers.com/showthread.php?t=2760203

FAQ
How did the testing take so long time?
Click to expand...
Click to collapse
Yes I agree that I took a lot of time, the kernels had a lot of settings to explore, but most of it could be done quite quickly and the results logged down, but the main delay was caused due to problems with my lower back which I had in the month of may, some emotional issues also strayed me away from this motive, so I apologize for any delay and to keep you guys waiting.
May I know your settings and any results/logs you have written to yourself?
Click to expand...
Click to collapse
A big NO. It is because the usage pattern and requirement for each of us is quite different and it can be that my settings give you a worse battery life. Also, I won't show those logs/notes to anyone no matter how good friend/relative/pet/neighbour/acquaintance he/she is because those are results for me, and is can also be that your device can take higher or lower undervolting as compared to me.
Any suggestion to which kernel I should use for getting this/that?
Click to expand...
Click to collapse
Again, NO. I won't entertain these kind of questions or spam on this thread, and while none have seen it yet, I am stricter than a moderator with the rules of this forum and any post relating to asking or suggesting a kernel to someone shall be instantly reported to a moderator and I want the readers here to follow this behaviour too.
I have this/that issue with this kernel, any help?
Click to expand...
Click to collapse
Post this in the respective kernel's thread and you'll get a much better and quicker reply, though I may occasionally feel like reply too, if you get lucky(but don't try to by spamming the thread, else instead of you post I will report your profile itself).
You haven't mentioned your battery life in any post
Click to expand...
Click to collapse
Yes, I forgot.
Anyways, Have a look at these screenshots.
Total battery time
Time I listened to music on my battery test days
Screen time
Note : I listened to music for only 1 hour, by setting up a timer, because this provided the ideal conditions for possible calculations of total time I can get on the battery. Also, my battery is quite old(not replaced or checked even once since I bought the device on the 2nd day of launch in India)
I need better battery life/ this feature/ that feature, so what kernel do you suggest me?
Click to expand...
Click to collapse
I can't suggest anything. There are no best kernels, you'll have to check about which one satisfies your needs and is stable too(remember read, read and read thoroughly and be patient).
I tried undervolting but my device gets unstable even on small values or doesn't respond to voltage change and continues to use stock voltages
Click to expand...
Click to collapse
Nearly forgot that one.
Use this Xposed module to disable the touchwiz DVFS, which has the highest priority for voltage and frequency settings, even more than the kernel, and causes more problems than fixes.

Important Notes
Battery life on stock kernel and Phoenix kernel was some, but with just half an hour less of screen time than the one I have shown in my screenshots.
The difference of battery life between was just of 12 seconds between AGNI and Nadia, so what are you going to do in those 12 seconds? Type Nuclear launch codes?
There were some random reboots, shutdowns on both AGNI and Nadia, but extremely rare and because I undervolted too much, both are equally stable for their current stage.
If you are going to post a reply here, post something worth reading, no spam.
Got a suggestion for me, which you think I might not be knowing? I appreciate you posting it here, will try to consider it gladly
There are to be no suggestions given to users for switching kernels just because you use it and like it, I will report that post at the instant I see it.
Don't worry about me being super strict with all this, usually I'm quite soft to everyone and would be willing to help without any hesitation, but don't want unnecessary spam here, neither do I want any fights for which a moderator has to kick in and do the cleanup and also serve you with a warning which you won't like to see.

Thank you for the time you spent on this.
EDIT : Just read through the whole thing. Great guide(?) indeed. It helped me make better choices regarding kernel settings.

Nice info..thanks mate..keep it up..?
Sent from my GT-N7100 using XDA Premium 4 mobile app

nice writeup..
i lack patience for such research as i get no mental energy remaining after development
Sent from F2FS FORMATED I9300 Temasek CM11 Unofficial v98 (4.4.3) + AGNi pureCM v2.8.0 kernel

psndna88 said:
nice writeup..
i lack patience for such research as i get no mental energy remaining after development
Sent from F2FS FORMATED I9300 Temasek CM11 Unofficial v98 (4.4.3) + AGNi pureCM v2.8.0 kernel
Click to expand...
Click to collapse
Thanks
And I have tried making some small mods/these and I know it's really tiring for the brain to have everything done correctly and make sure that users are happy.
Sent from my RPG with auto targeting

Nice thread mate. It really showed that it needed a lot of time and patience to know your best setting to optimize your phone depending the usage. This inspires me not to just stick on a single thing but to take the risk and explore for the other things. Again, thanks you mate. Now I may go and explore my own settings. Cheers!!!!
Sent from my SM-G900F using Tapatalk

i posted thanks by mistake, i edited it
but how can i delete my post ?
Sent by my AWP #Unzoom
---------- Post added at 07:56 AM ---------- Previous post was at 07:45 AM ----------
what does undervolting do ?
Sent by my AWP #Unzoom

amit.dhatrak95 said:
i posted thanks by mistake, i edited it
but how can i delete my post ?
Sent by my AWP #Unzoom
---------- Post added at 07:56 AM ---------- Previous post was at 07:45 AM ----------
what does undervolting do ?
Sent by my AWP #Unzoom
Click to expand...
Click to collapse
Don't worry if you did it by mistake.
And undervolting means controlling of voltages and reducing the power requirement per clock.
Sent from my RPG with auto targeting

@KNIGHT97
Last month I accidentally undervolted my cpu to -100mv
That caused my phone to freeze. I rebooted it. It booted fine but freezed at the lockscreen after boot.
So I decided that I should flash the agni-oc-uv-reset.zip which is saved in your sdcard as soon as you flash agni kernel. But that zip failed to flash. I tried with philz, twrp both.
What could be the reason?
Sent from my GT-N7100 using Tapatalk

smeet.somaiya said:
@KNIGHT97
Last month I accidentally undervolted my cpu to -100mv
That caused my phone to freeze. I rebooted it. It booted fine but freezed at the lockscreen after boot.
So I decided that I should flash the agni-oc-uv-reset.zip which is saved in your sdcard as soon as you flash agni kernel. But that zip failed to flash. I tried with philz, twrp both.
What could be the reason?
Sent from my GT-N7100 using Tapatalk
Click to expand...
Click to collapse
High undervolting has led to freezes that you have figured out. But zip failing to flash? I can't seem to comment on that, becuase I don't know about that issue, was that something like status 6/7 or bad zip error?
Sent from my RPG with auto targeting

KNIGHT97 said:
High undervolting has led to freezes that you have figured out. But zip failing to flash? I can't seem to comment on that, becuase I don't know about that issue, was that something like status 6/7 or bad zip error?
Sent from my RPG with auto targeting
Click to expand...
Click to collapse
Yeah. forgot to mention. status 7 it said. Error executing update-binary.
Flashed stock kernel then and everything was fine.
After that I went on to check the zip file. And shockingly it was of 0 bytes.
So I think there was some error in extracting that zip while flashing agni kernel.
Anyways that has been fixed in the latest version. Was just informing you☺
Sent from my GT-N7100 using Tapatalk

smeet.somaiya said:
Yeah. forgot to mention. status 7 it said. Error executing update-binary.
Flashed stock kernel then and everything was fine.
After that I went on to check the zip file. And shockingly it was of 0 bytes.
So I think there was some error in extracting that zip while flashing agni kernel.
Anyways that has been fixed in the latest version. Was just informing you☺
Sent from my GT-N7100 using Tapatalk
Click to expand...
Click to collapse
Yea, I saw the earluer ones just now, seems to be an typo in updater script which caused status 7 error.
But yes, glad that its fixed now.
Sent from my RPG with auto targeting

KNIGHT97 said:
Yea, I saw the earluer ones just now, seems to be an typo in updater script which caused status 7 error.
But yes, glad that its fixed now.
Sent from my RPG with auto targeting
Click to expand...
Click to collapse
What typo was it?
And yeah one more thing, kindly let me know what does Frandom rng to? It uses random entropy generator but can't figure out what it is exactly☺
Sent from my GT-N7100 using Tapatalk

smeet.somaiya said:
What typo was it?
And yeah one more thing, kindly let me know what does Frandom rng to? It uses random entropy generator but can't figure out what it is exactly☺
Sent from my GT-N7100 using Tapatalk
Click to expand...
Click to collapse
Frandom is a Linux kernel random number generator, which is 10-50 times faster than what you get from Linux' built-in /dev/urandom. And it uses very little (/dev/frandom) or none (/dev/erandom) of the kernel's entropy pool, so it is very useful for applications that require a handy source for lots of random data.
Its most common use seems to be wiping large hard disks with random data before encryption
here you have the link for more info: http://billauer.co.il/frandom.html
Enviado desde mi GT-N7100 mediante Tapatalk

smeet.somaiya said:
What typo was it?
And yeah one more thing, kindly let me know what does Frandom rng to? It uses random entropy generator but can't figure out what it is exactly☺
Sent from my GT-N7100 using Tapatalk
Click to expand...
Click to collapse
I couldn't figure put the typo, but that's the most probable problem for status 7 error.
And your answer for Frandom RNG is already answered below
Pafcholini said:
Frandom is a Linux kernel random number generator, which is 10-50 times faster than what you get from Linux' built-in /dev/urandom. And it uses very little (/dev/frandom) or none (/dev/erandom) of the kernel's entropy pool, so it is very useful for applications that require a handy source for lots of random data.
Its most common use seems to be wiping large hard disks with random data before encryption
here you have the link for more info: http://billauer.co.il/frandom.html
Enviado desde mi GT-N7100 mediante Tapatalk
Click to expand...
Click to collapse
So if the kernel has in impact on memory, would it be possible to allocate the data into random blocks onto the memory, both internal and external, just like it is on 64-bit. Intel has managed a 64-bit kernel so maybe you can look at that to get some idea about it.
Or is this thing completely 64-bit only?
Sent from my RPG with auto targeting

KNIGHT97 said:
So if the kernel has in impact on memory, would it be possible to allocate the data into random blocks onto the memory, both internal and external, just like it is on 64-bit. Intel has managed a 64-bit kernel so maybe you can look at that to get some idea about it.
Or is this thing completely 64-bit only?
Sent from my RPG with auto targeting
Click to expand...
Click to collapse
Didn't even get a single word of what you said. My knowledge on memory allocating is limited. I always make mistakes in sums related to memory mapping in my microprocessor exams??
Sent from my GT-N7100 using Tapatalk
---------- Post added at 11:59 AM ---------- Previous post was at 11:57 AM ----------
Pafcholini said:
Frandom is a Linux kernel random number generator, which is 10-50 times faster than what you get from Linux' built-in /dev/urandom. And it uses very little (/dev/frandom) or none (/dev/erandom) of the kernel's entropy pool, so it is very useful for applications that require a handy source for lots of random data.
Its most common use seems to be wiping large hard disks with random data before encryption
here you have the link for more info: http://billauer.co.il/frandom.html
Enviado desde mi GT-N7100 mediante Tapatalk
Click to expand...
Click to collapse
Okay. Got what you mean. But what is its application on a phone with limited memory?
Sent from my GT-N7100 using Tapatalk
---------- Post added at 12:03 PM ---------- Previous post was at 11:59 AM ----------
@KNIGHT97
What I meant to say is please elaborate. Don't mind was just kidding☺
Sent from my GT-N7100 using Tapatalk

smeet.somaiya said:
Didn't even get a single word of what you said. My knowledge on memory allocating is limited. I always make mistakes in sums related to memory mapping in my microprocessor exams??
Sent from my GT-N7100 using Tapatalk
---------- Post added at 11:59 AM ---------- Previous post was at 11:57 AM ----------
Okay. Got what you mean. But what is its application on a phone with limited memory?
Sent from my GT-N7100 using Tapatalk
---------- Post added at 12:03 PM ---------- Previous post was at 11:59 AM ----------
@KNIGHT97
What I meant to say is please elaborate. Don't mind was just kidding☺
Sent from my GT-N7100 using Tapatalk
Click to expand...
Click to collapse
hi do you know how can I enable frandom in nadia kernel