[Q] Best governor Scheduler combo ? - Galaxy Note GT-N7000 Q&A, Help & Troubleshooting

Which combination of governor and scheduler gives the best performance for ICS Stock ? And which one for better battery life ?

Akilesh said:
Which combination of governor and scheduler gives the best performance for ICS Stock ? And which one for better battery life ?
Click to expand...
Click to collapse
Other said it is pegasus and devs say it is all the same just a little difference.

jonpaslim said:
Other said it is pegasus and devs say it is all the same just a little difference.
Click to expand...
Click to collapse
I was told by the majority it goes in this order:
1) Pegasusq
2) Lulzactive
3) SmartassV2
4) Ondemand
5) Hotplug
6) Interactive
... and the rest.
And for IO Scheduler:
1) VR
2) CFQ
3) SIO
The truth is that, this is a Dual Core Phone with 1GB RAM, and improved Android 4.03+ kernel. So the previous hardware and software limitations NO LONGER applies. Coming FROM the DEVELOPERS of governors and schedulers the best ones seem to be:
1) InteractiveX2
2) HotplugX
3) Ondemand
... then lulzactive ~ pegasusq > smartassV2
And for IO Schedulers:
1) SIO
... then the rest
The truth is you should use the DEVELOPER recommended governor and sio scheduler, for an extensive period (1-2 days) and compare them. Finding what works best for you. After that's done, then you should go further stability testing with Voltage Control. Eg, UC your lowest frequency, OC your highest frequency and UV each step to the NEEDS of your device, each device is UNIQUE and functions DIFFERENTLY.
I think doing so you should extend your battery life by 1 or 1.5hour per cycle, while maintaining or increasing performance.

it all depends on the ROM + Kernel and more importantly the user and how they use their Note.
For me, I use Performance + Noop.. But ive had good results with pegasusq + sio
The only way to get the best combination is to try them all, But give them time to perform, Dont make a decision after 5 mins of usage, Test them heavily, Do some serious Gaming + Multitasking and chose one that outperforms the rest !

I'll start testing and post results from time to time.
Sent from my GT-N7000 using xda premium

azzledazzle said:
it all depends on the ROM + Kernel and more importantly the user and how they use their Note.
For me, I use Performance + Noop.. But ive had good results with pegasusq + sio
Click to expand...
Click to collapse
But the performance governor never scales down but always runs at maximum frequency. How is battery life with that? Must be horrible?!

Kangal said:
I was told by the majority it goes in this order:
The truth is you should use the DEVELOPER recommended governor and sio scheduler, for an extensive period (1-2 days) and compare them. Finding what works best for you. After that's done, then you should go further stability testing with Voltage Control. Eg, UC your lowest frequency, OC your highest frequency and UV each step to the NEEDS of your device, each device is UNIQUE and functions DIFFERENTLY.
I think doing so you should extend your battery life by 1 or 1.5hour per cycle, while maintaining or increasing performance.
Click to expand...
Click to collapse
I agree, governors even if they have the same stock names(i.e ondemand) has been heavily tweaked by the kernel dev so that you maximize performance and batt life. Start off by using the kernel/scheduler suggested by the kernel dev.
The only thing I care about is running out of juice when I don't have access to something that can charge my phone, as long as i get 10-15hrs off batt life and solid performance I'm good. Still I always bring a wall charger/usb cord with me if i really needed the extra juice.
Remember you can charge the battery but you can't charge performance

Coxeroni42 said:
But the performance governor never scales down but always runs at maximum frequency. How is battery life with that? Must be horrible?!
Click to expand...
Click to collapse
I get around 24 hours from my battery and I'm happy with that, besides I'm always somewhere where I can charge my phone should by battery run out
Sent from my GT-N7000 using Tapatalk 2

chickentuna said:
Remember you can charge the battery but you can't charge performance
Click to expand...
Click to collapse
This sentence doesn't make sense at all.
Unless you mean you can change or recharge the battery, which is straight-forward.
But you can "change" (increase) the performance. Just clear all background tasks, go to Voltage Control and overclock the cpu frequency upto 1.704GHz (with +1400mV) and keep it at that frequency by using the Performance governor. Also overclock your gpu upto 400MHz (with +1,000mV).
Though I don't recommend that, stability issues, warming issues and possible physical damage to your processor. But use the same concept, just not Overclock too much!
---------- Post added at 02:09 PM ---------- Previous post was at 02:08 PM ----------
azzledazzle said:
I get around 24 hours from my battery and I'm happy with that, besides I'm always somewhere where I can charge my phone should by battery run out
Sent from my GT-N7000 using Tapatalk 2
Click to expand...
Click to collapse
That's simply NOT true.
Otherwise you have your phone in deep sleep/idle far more than the average user, barely using your phone.
Other possibility is that you've severely underclocked your maximum frequency, and have a long time in idle.
Performance governor = Always on maximum frequency.
NOTE's maximum frequency uses 1300+ mV. The NOTE's battery size is 2,500 mAh. Simple maths will tell you 24hrs is impossible ; )

But it is true I'm not saying my phone will be awake for 24 hours I'm saying I charge it roughly every 24 hours, I know performance keeps the phone at max speeds but I am human I will sleep for about 8-9 hours, so in result so will my phone.
And I don't use my phone every minute of the day I do have work to do so my battery life is what I say it is. What I consider 'average use' might be absolute minimal use to you so maybe that's why you doubt me,
I personally don't mess with voltages and things cause I don't understand them but I use note core kernel which is highly tweaked and one of the best kernels for battery life
Sent from my GT-N7000 using Tapatalk 2

My recommendation is lulzactive.

i use hotplug+sio on maclaws slim bean,, awesome battery life

Related

OC and Battery questions

hello,
im just wondering first of all whats the best speed to OC without permanent damage. I have done over 900 Mhz but I will get maybe 1 or 2 reboots a day so I went down to 864Mhz and its been smooth no problems. but what do you think is best to use.
also taking the first question to mind, what governor and i/o settings should I use to give the best balance between battery life and performance? I have tried smartassv2 and the battery like suffers.
Please give me your opinion for what is the best combination for my phone
Well first of all you can't have both performance and battery life.
If you are after performance go for 883 ( I think. I don't exactly remember the freq so forgive me for that) or what you suggested yourself (864) and set the governor to interactive and min frequency to 245. Personally I would set the governor to performance and Max frequency to 864 min 245 though it may sound crazy but that's my settings, phone doesn't get as hot as it used to battery life has kinda increased actually well if you think about it you will see that your phone will complete a given task in a shorter time and therefore ...
If you are after battery life then don't think about over clocking. In fact you might wanna under clock to certain level.
Sent from my GT-S5830 using xda premium

Overclocking the Nexus 4 - Benchmark of application load time = BIG difference

A lot of people say that overclocking the Nexus 4 is useless as it's so fast to begin with, but that turns out to be false, yes, it's useless for making e.g. games run smoother, as they all already do, but it's very useful for reducing loading times, and it turns out that the % of time saved is close to the % of increased clock speed.
My setup: Nexus 4, franco.Kernel updater v9.3.3, Franco Kernel #666, Jelly Bean 4.2.2 JDQ39, Governor Interactive. Angry pigs HD 1.1.0 app loading. Fresh reboot for each speed. I did three measurements for each clock speed and left voltages at Franco's defaults. I would suggest looking into the undervolting threads to figure out how low you can go with voltages, undervolting is good underclocking is usually not.
1026 Mhz 1025mV default Franco
15.4s
15.9s
15.7s
1512 MHz 1150 mV default Franco
11.5s
11.3s
11.3s
1620 MHz 1175mV default Franco
11.1s
10.6s
10.6s
1674 MHz 1175mV default Franco
10.3s
10.6s
10.7s
1728 MHz 1200mV default Franco
10.5s
10.3s
10.3s
Conclusion
12.5% faster clock speed = 8.85% faster load time, which means increasing clock speed is 70% effective at reducing load time going from 1512 Mhz and upwards.
40% faster clock speed = 35% faster load time, meaning that people who think running their device at 1026 Mhz is as good as 1512 Mhz are wrong. Nearly doubling clock speed nearly halfens load time.
The reduction in loading time is reduced a bit as you reach higher frequencies, but it's still 70% effective so every extra Mhz helps cut down the wait. Overclocking the Nexus 4 is meaningless in terms of increasing smoothness IN APP, but meaningful in reducing load times significantly and in increasing the responsiveness of your system.
please tell me for saving the battery and still maintaining a little smoothness
the settings to use
i use faux123 kernel.. i just need the voltage,cpu clock n governer settings
so overclocking makes things faster.. really? who would have thunk. and you just found this out?
my brother uses a galaxy s3, and after using it for a few months he tried my n4... he was amused how fast things open and how smooth everything was, and it was all stock frequencies. i think it makes no sense overclocking this beast, it is already fast enough. +/- a few seconds doesnt make that much of a difference but drains the battery significantly more. just sayin'.
zakoo2 said:
my brother uses a galaxy s3, and after using it for a few months he tried my n4... he was amused how fast things open and how smooth everything was, and it was all stock frequencies. i think it makes no sense overclocking this beast, it is already fast enough. +/- a few seconds doesnt make that much of a difference but drains the battery significantly more. just sayin'.
Click to expand...
Click to collapse
Sorry but little OC can help you save more battery in day to day usage. (If you are not playing only games on your phone).
Sent from SpeedMachine i9100
alen1901 said:
Sorry but little OC can help you save more battery in day to day usage. (If you are not playing only games on your phone).
Sent from SpeedMachine i9100
Click to expand...
Click to collapse
not according to my experiences, and i did play with oc/uc and voltage quite a bit. i always ended up going back to stock on the n4 (not on my old motodefy though) because of the drainage.
So, a 50% increase in clock cycles pushes the performance by a bit.
Who would've thought...?
well that's normal, overclocking = more power draining, more speed; just like with CPU or ram on pc, no rocket science here.
alen1901 said:
Sorry but little OC can help you save more battery in day to day usage. (If you are not playing only games on your phone).
Sent from SpeedMachine i9100
Click to expand...
Click to collapse
OC? Don't you mean Downclocking or Undervolting?
sigma392 said:
OC? Don't you mean Downclocking or Undervolting?
Click to expand...
Click to collapse
Nope , also OC can (in theory) improve battery life because the CPU finishes its task faster.
a pigeon delivered this message.... ** your welcome **
gohan040 said:
Nope , also OC can (in theory) improve battery life because the CPU finishes its task faster.
a pigeon delivered this message.... ** your welcome **
Click to expand...
Click to collapse
But er... Wouldn't it would theoretically need more power and therefore generating more heat than stock frequencies which in turn will have a negative effect on battery life?
sigma392 said:
But er... Wouldn't it would theoretically need more power and therefore generating more heat than stock frequencies which in turn will have a negative effect on battery life?
Click to expand...
Click to collapse
It REALLY depends. I'd say most day-to-day usage, an overclock can actually help save battery by 'racing' to sleep (hurry up, finish, drop to deep sleep on cpu AND radio). I think being on 3g makes it even more pronounced because you're letting the radio drop back down to a lower power state as well as the CPU. So if you need to quickly open an email attachment, and cpu is the limiter, it can help it finish, get off the network, and back to idle state faster and save some battery. Now if you go on running benchmarks instead where you just keep throwing more work at it that's a seriously different story. There is also a point of diminishing returns (usually 1 or 2 steps before the absolute limit, in our case about 1670mhz) but in general, a moderate overclock can help.
sigma392 said:
But er... Wouldn't it would theoretically need more power and therefore generating more heat than stock frequencies which in turn will have a negative effect on battery life?
Click to expand...
Click to collapse
I agree with u, but remember its a theory.. U finish faster so the time it uses more power is less....
^^Look at the post above me, he got a good example^^
The best of both worlds (if your phone is capable to do this) is when u UV and OC. U would use less power on a higher frequency. Win-win in theory.
But I know what u mean and its hard to believe, but I think that's what @alan1901 wanted to say.
a pigeon delivered this message.... ** your welcome **
If you are lucky, you can overclock using STOCK voltage for the maximum clock.
IE, i see many undervolting their phones by alot. I would just try 1150mV (default voltage for 1.5ghz) for your highest speed (OC as high as you can with this voltage), which means you are not using more power, but still making your phone faster.
Haven't tried on mine, but say 1.6-1.7ghz should be possible with 1150mV for a few at least.
No, 50% OC is not = 50% reduced load time
Let me summarize:
OVERCLOCKING > UNDERCLOCKING FOR THE AVERAGE (POWER) USER
You load a lot of apps, you don't play games or watch videos for hours, productivity is your focus. Depending on use you could save probably 1 - 6 hours a year (see assumptions below). Battery gains won't be significant, but spending less time with the LCD/Wifi etc on could give you some gains as the CPU is one of the lower power drain components.
UNDERCLOCKING > OVERCLOCKING FOR THE 3D GAMER
You spent a lot of time playing CPU intensive games, keeping the CPU cores at max clock speed for extended periods of time. In that case you could get some battery life gains at the cost of responsiveness and app loading times - your call.
IN MORE DETAIL
-------------------------------------------------------------------------------------
UNDERCLOCKING + UNDERVOLTING PROs/CONs
+ Reduces CPU temperature, if that matters to you.
+ IF keeping the CPU at max clock speed for extended periods of time, a lower max voltage could give you enough extra battery life to negate the time lost waiting for apps to load - your call.
- Increased load times - even on the Nexus 4 a lot of apps take > 10 seconds to launch, and don't forget in-app load times, reboots etc.
- No significant battery saving in most cases, potential for loss (if doing a lot of app loading you could even get worse battery life as the screen, wifi etc has to stay on for longer, and the screen is the main battery drain).
- Time spent finding stable voltages.
OVERCLOCKING PROs and CONs
+ Potential small battery life gain, as your device spends less time with the screen/wifi etc ON.
++ SAVES YOU TIME. I'm estimating 1 - 5 hours a year depending on usage.
- Times spent finding stable voltages and clock speeds.
-/- Apps does not RUN smoother.
Why 50% CPU does not = 50% reduced load time and why you test things
---------------------------------------------------------------------------------------------------------------------
It'd be more like 40-30% depending on relative clock speed. I did a simple test as we did NOT know that app loading was mainly CPU limited, could have been memory. And If load times didn't go down there wouldn't be ANY reason to OC and underclocking would be a good idea.
Assumptions behind an estimated 1 - 6 hours saved a year by overclocking
---------------------------------------------------------------------------------------------------------------------
Say that you're on average waiting 4 minutes a day waiting for apps to load, rebooting your phone, waiting for web pages to load, waiting for in-app content to load etc.
Overclocking can cut down that wait time up to 10%, that's nearly 1/2 a minute saved every day, 3 1/2 minutes a week, 3 hours a year. This means you can get a good return on the time it took you to overclock your device, assuming you'll keep it for a couple of years. So if that's the case, why not do it? You'll save some time and you'll enjoy using your device a bit more.
Currently I'm OCd at max 1674 Mhz @ 1225mV, otherwise using default Franco voltages and interactive governor. Stable and very snappy Might be able to reduce voltage and retain stability but this is fine.
zakoo2 said:
not according to my experiences, and i did play with oc/uc and voltage quite a bit. i always ended up going back to stock on the n4 (not on my old motodefy though) because of the drainage.
Click to expand...
Click to collapse
sigma392 said:
OC? Don't you mean Downclocking or Undervolting?
Click to expand...
Click to collapse
I said little OC. Example. Stock cpu clock is 1000mhz and 1000mv. Now, you OC your phone to 1200,1300mhz without changing voltage, or by increasing it a little bit(+25, +50mv, to make phone stable), it will use higher cpu frequency with the same voltage like on lower frequency or with a little higher voltage. So it would finish tasks faster, going to idle faster without getting any heat and give you more battery. On my nexus s and galaxy s3, galaxy s2, i tested it, and it helped (about 30min more screen on time for me). But if you play games a lot and doing some heavy tasks then underclock and undervolt are best for you. But on that beast of phone, i woldnt underclock it beyond 1.2, 1.3, not sure what frequencies are there. If you UC that phone a lot, you can easyly go with some weaker phone.
Sent from SpeedMachine i9100
is there any kernel that can overclock with 4.3?
Andre_Vitto said:
is there any kernel that can overclock with 4.3?
Click to expand...
Click to collapse
yes, most of them. some even overclock up to 2052mhz :victory:
simms22 said:
yes, most of them. some even overclock up to 2052mhz :victory:
Click to expand...
Click to collapse
like?
Check the Development sections.
This thread makes my head hurt.

Question about CPU governors

I am looking for a CPU governor I can use with just the root that our AT&T Note 3's currently have. What I would like to do is have a governor so when certain games that are really draining my battery are running that it will lower the CPU and give me better battery life. Is this possible with our current root only? I read that SetCPU had this feature with app based profiles but the recent reviews are stating that the app based profiles are not working correctly. Does anyone have any suggestion of what I could use?
Set cpu is working for me, I have the paid version though. I put it on performance governor and it held 1 core wide open at 2265. I then changed it to ondemand and the core dropped to 300. There's also a conservative, interactive, user space and power save governors. Search governors for explanation of each governors settings/purpose.
Sent from my SAMSUNG-SM-N900A using xda app-developers app
kitsinni said:
I am looking for a CPU governor I can use with just the root that our AT&T Note 3's currently have. What I would like to do is have a governor so when certain games that are really draining my battery are running that it will lower the CPU and give me better battery life. Is this possible with our current root only? I read that SetCPU had this feature with app based profiles but the recent reviews are stating that the app based profiles are not working correctly. Does anyone have any suggestion of what I could use?
Click to expand...
Click to collapse
You can set a profile to lower CPU when your battery gets above a certain speed. That would be your best. Another alternative is to set a max speed for x condition or y. Doing all of the above will lower the performance of your game. It would be best to try different max speeds that still allow you to play your game enjoyably and go from there. Games are just a battery hog, even if you lower your speed it will still take milliamps than you'd prefer. I think off hand to just try, is set max speed at the 15xx speed and just use on demand or interactive. Power save will lower performance too dramatically and possibly kill more battery. Performance is really for the most demanding of games or benchmarking.
Sent from my SM900T.
@rbiter said:
You can set a profile to lower CPU when your battery gets above a certain speed. That would be your best. Another alternative is to set a max speed for x condition or y. Doing all of the above will lower the performance of your game. It would be best to try different max speeds that still allow you to play your game enjoyably and go from there. Games are just a battery hog, even if you lower your speed it will still take milliamps than you'd prefer. I think off hand to just try, is set max speed at the 15xx speed and just use on demand or interactive. Power save will lower performance too dramatically and possibly kill more battery. Performance is really for the most demanding of games or benchmarking.
Sent from my SM900T.
Click to expand...
Click to collapse
so this may be a dumb question, but is 300MHz the lowest clock speed it can reach? I just installed Better Battery Stats on my Nexus 7 (2012) and it can get down to 50MHz. Just looking to conserve more battery. Thanks!
Just get a Zerolemon 10000 mah battery and never worry about what is using your battery again. They are only 50 bucks and that includes the special case/protector to house the phone and the battery.
Solarenemy68 said:
Just get a Zerolemon 10000 mah battery and never worry about what is using your battery again. They are only 50 bucks and that includes the special case/protector to house the phone and the battery.
Click to expand...
Click to collapse
That's not really an option for me. I often rock the phone with no case at all and like it in its 168g form. I already have 30-40 hour batterly life with 3-4 hrs of screen time. But I just noticed the lowest clock speed it 300Mhz and I think it can be lower!
Solarenemy68 said:
Just get a Zerolemon 10000 mah battery and never worry about what is using your battery again. They are only 50 bucks and that includes the special case/protector to house the phone and the battery.
Click to expand...
Click to collapse
Yes! This is my solution. No worries when mpdecision gets stuck on 4.4.2 and stops sleeping cores, who cares?! Pry-Fi running continuously in War mode, booyah.

Undervolting vs Underclocking vs Hotplugging for battery life

Can anyone explain to me what is better?
I know that underclocking reduces cpu frequency, which means lower temperatures and also better battery, but performance is worse, so it takes longer to complete tasks...but if it takes longer to complete tasks, it means it will also have screen on for a longer time right?so it saves power from the cpu, but keeping the screen on for that extra time also means it will consume extra power right?
About undervolting, is it better to undervolt or to underclock when it comes to battery life?i know that with undervolting there is less heat produced because of the reduced voltages. From what i have read (i dont know if it is correct) the power consumption is given by this equation: P = f*c*(V^2) where f is frequency, c is capacitance and v is voltage. It makes sense that reducing voltage means less heat, but if i reduce CPU frequency it also produces less heat because it reduces power. Most of the time i read that underclock is better for battery because it uses less power, but like i said earlier, it takes longer to complete tasks and in result i have to keep the screen on for a longer time.
About hotplugging, i have seen some users with good battery life screenshots, and mentioning that they disabled hotplugging, because they said that turning cores on and off also wastes energy. So is it better to hotplug or not?
would really like to get some answers because i know nothing about this..I also know that i could test each setting and and see which is better, but i dont use my phone the same way everyday so its kind of hard to determine...
Short version and after extending testing (2 cores max, -175Uv, underclocking at 1000 or 1300mHz) with different kernels and always on stock Rom,
the result was that there was no noticeable difference at the battery life with a normal setup of using of 4 cores at normal max freq of 1512mhz.
A custom kernel in comparison to the stock kernel makes more sense cause of the optimization they offer for performance and battery life.
Unleashed from Onda v957m on TDT
RASTAVIPER said:
Short version and after extending testing (2 cores max, -175Uv, underclocking at 1000 or 1300mHz) with different kernels and always on stock Rom,
the result was that there was no noticeable difference at the battery life with a normal setup of using of 4 cores at normal max freq of 1512mhz.
A custom kernel in comparison to the stock kernel makes more sense cause of the optimization they offer for performance and battery life.
Unleashed from Onda v957m on TDT
Click to expand...
Click to collapse
nice to have some feedback from someone that tested=p. what about hotplugging? Today i disabled hotplugging and had always 4 cores on and it does not seem to be wasting more battery than with hotplugging enabled
These studies were done by bedalus on the Nexus S, but alot pertains today. Alot of useful information here
http://forum.xda-developers.com/nexus-s/general/ref-battery-drain-benchmarks-t1478406
Sent from my Nexus 4
n2d551 said:
These studies were done by bedalus on the Nexus S, but alot pertains today. Alot of useful information here
http://forum.xda-developers.com/nexus-s/general/ref-battery-drain-benchmarks-t1478406
Sent from my Nexus 4
Click to expand...
Click to collapse
Also lot of discussion about Uv here
http://forum.xda-developers.com/showthread.php?t=2137034
Unleashed from Onda v957m on TDT
n2d551 said:
These studies were done by bedalus on the Nexus S, but alot pertains today. Alot of useful information here
http://forum.xda-developers.com/nexus-s/general/ref-battery-drain-benchmarks-t1478406
Sent from my Nexus 4
Click to expand...
Click to collapse
totally forgot about bedalus experiments!i had a Nexus S at the time and saw that post, but back then i didnt mess around with anything on my phone so i completely forgot=p.thanks!
The only thing he doesnt adress is hotplugging(the Nexus S was single core so he couldnt even if wanted=p). Anyone care to shed some light on hotplugging?is it really worth it or there are minimal gains?
migueldbr said:
totally forgot about bedalus experiments!i had a Nexus S at the time and saw that post, but back then i didnt mess around with anything on my phone so i completely forgot=p.thanks!
The only thing he doesnt adress is hotplugging(the Nexus S was single core so he couldnt even if wanted=p). Anyone care to shed some light on hotplugging?is it really worth it or there are minimal gains?
Click to expand...
Click to collapse
There are different hotplugging methods.
The stock one as a boost feature, as soon as u touch the screen ur decvice goes to dual core 1026mhz for a couple seconds even if the load is super low.
Custom kernels have many variants on the hotplugging "style". Most of them dont have the touch boost included to save battery (but u can have the same kind of touch boost enabled by the governor, ex: franco kernel). Others simply advise u to turn off hotplugging == less calculation of the load to decide if the device needs to plug it or not + no waiting time to get the performance boost of many cores online (since all 4 are already online) + somewhat more heat since all cores are allways draining battery.
Im no expert and i hope all i said is right, at leats its what i know.
What i personally look for is:
Min core 1
Max cores 4
No touch boost (no heat while u are simply texting via sms on 2G with data and wifi OFF)
sent from my diabetic Nexus 4 (too many KitKats).
C4SCA said:
There are different hotplugging methods.
The stock one as a boost feature, as soon as u touch the screen ur decvice goes to dual core 1026mhz for a couple seconds even if the load is super low.
Custom kernels have many variants on the hotplugging "style". Most of them dont have the touch boost included to save battery (but u can have the same kind of touch boost enabled by the governor, ex: franco kernel). Others simply advise u to turn off hotplugging == less calculation of the load to decide if the device needs to plug it or not + no waiting time to get the performance boost of many cores online (since all 4 are already online) + somewhat more heat since all cores are allways draining battery.
Im no expert and i hope all i said is right, at leats its what i know.
What i personally look for is:
Min core 1
Max cores 4
No touch boost (no heat while u are simply texting via sms on 2G with data and wifi OFF)
sent from my diabetic Nexus 4 (too many KitKats).
Click to expand...
Click to collapse
actually touch boost and hotplugging are different things...touch boost is a feature where as soon as you touch the screen, the cpu ramps up the frequency to the specified touch boost frequency. Hotplugging is a different thing, where cores of the cpu are turned on and off when the phone does not need them to be on all at the same time. I know that touch boost drains more battery, but i dont know about hotplugging...would like to see some tests/benchmarks, but i dont think there are any...
migueldbr said:
actually touch boost and hotplugging are different things...touch boost is a feature where as soon as you touch the screen, the cpu ramps up the frequency to the specified touch boost frequency. Hotplugging is a different thing, where cores of the cpu are turned on and off when the phone does not need them to be on all at the same time. I know that touch boost drains more battery, but i dont know about hotplugging...would like to see some tests/benchmarks, but i dont think there are any...
Click to expand...
Click to collapse
Different things but on stock kernel touch boost is implemented on the hotplug and not the governor.
Any difference btw 4 cores online or hotplugging must minor, and have draw backs on heat wich affects the battery capacity.
Eventhough u can say that hotplugging may drain an amount of battery to plug and unplug cores, i would say its minor.
Talk to @simms22 , he is the "trinity kernel guy", four cores online is a must for him
sent from my diabetic Nexus 4 (too many KitKats).
From my experience, under clocking as well as running a 2 core setup doesn't change anything for the better. Most times for the worst due to a worse user experience.
Undervolting is something that, solely from the physical side, cannot make things worse (unless you under clock too much). Undervolting will make your CPU cores use less current. Less current running through an electric circuit always means less heat, too.
To put it simple:
If my regular voltage at 384mhz is 950mv and I lower the voltage to 800mv and after testing it proves to run stable my device now uses 150mv less on that frequency.
Sent from my Nexus 4 using XDA Free mobile app
Oxious119 said:
From my experience, under clocking as well as running a 2 core setup doesn't change anything for the better. Most times for the worst due to a worse user experience.
Undervolting is something that, solely from the physical side, cannot make things worse (unless you under clock too much). Undervolting will make your CPU cores use less current. Less current running through an electric circuit always means less heat, too.
To put it simple:
If my regular voltage at 384mhz is 950mv and I lower the voltage to 800mv and after testing it proves to run stable my device now uses 150mv less on that frequency.
Sent from my Nexus 4 using XDA Free mobile app
Click to expand...
Click to collapse
no matter what you do , its still going to be a mediocre battery life , best leave it as it is because the way nexus 4 is designed its still going to be getting annoyingly warm on games
Well, from my own experience undervolting definetly reduced heat and made the battery last longer. Don't expect 2 hours more sot, though. Its most notably while the device is idle.
Sent from my Nexus 4 using XDA Free mobile app
nice man
very helpful...
Help pls..
Can any1 help me undervolting Unleashed kernel.. providing tips or the link ll be vry helpfull..

HUGE RENDERING, FPS and SMOOTHNESS improvement on OOS.

I don't know if it works on OOS versions but it does, on the OB12. You need to be rooted to be doing this.
Lower the input boost to 1036 on the big and 556 on the small first. Now, change the governor of the big cluster to conservative and the small to ondemand or conservative , benefit!
To see the improvement in FPS, you need to go to developer options and switch on profile GPU rendering. For those who didn't understand this: lower the graph is, higher the fps. Hit thanks if helped
Where to change these settings?
Any kernel manager would do. I'd recommend kernel aduitor
chinmai560621 said:
Any kernel manager would do. I'd recommend kernel aduitor
Click to expand...
Click to collapse
Just change the option of GPU GOVERNER to simple_on demand one from the above mentioned app. It will work smooth.. works for me..:good:?
ashokspidy said:
Just change the option of GPU GOVERNER to simple_on demand one from the above mentioned app. It will work smooth.. works for me..:good:?
Click to expand...
Click to collapse
I'd recommend to leave it on msm-adreno.
What if I'm on RR 5.8.2 and ElementalX? I have ondemand governor option for little but don't have conservative for big cluster.
karliyo said:
What if I'm on RR 5.8.2 and ElementalX? I have ondemand governor option for little but don't have conservative for big cluster.
Click to expand...
Click to collapse
This is exclusive for OOS. In fact, the interactive is supposed to be better than the rest but it's not the case on OOS which is why I made this thread.
chinmai560621 said:
This is exclusive for OOS. In fact, the interactive is supposed to be better than the rest but it's not the case on OOS which is why I made this thread.
Click to expand...
Click to collapse
Oh alright. Little governor ondemand seems quite fine right now.
By the way, are both governors interactive by default?
karliyo said:
Oh alright. Little governor ondemand seems quite fine right now.
By the way, are both governors interactive by default?
Click to expand...
Click to collapse
Has to be. Unless you use blu_spark kernel. It uses blu_activ.
How do these changes affect the battery life?
chinmai560621 said:
Has to be. Unless you use blu_spark kernel. It uses blu_activ.
Click to expand...
Click to collapse
You mean interactive and not blu_active?
I do recommend you guys taking a look at AKT.. There are alot of peformance oriented and balanced profiles that are already better with smoothness then stock.. Also Franco has some magic with stock settings!
chinmai560621 said:
I don't know if it works on OOS versions but it does, on the OB12. You need to be rooted to be doing this.
Lower the input boost to 1036 on the big and 556 on the small first. Now, change the governor of the big cluster to conservative and the small to ondemand or conservative , benefit!
To see the improvement in FPS, you need to go to developer options and switch on profile GPU rendering. For those who didn't understand this: lower the graph is, higher the fps. Hit thanks if helped
Click to expand...
Click to collapse
Hi friend, shall I change both values of big and small? I mean maximum and minimum?
Same, cant find where to change the input values for those two.
rOck4n said:
Same, cant find where to change the input values for those two.
Click to expand...
Click to collapse
You'll see input boost in the bottom section of CPU in kernel adiutor. Core 1- 1036, core 3 -556 is what you've to do. It'll be 1324 default on both. I'm suggesting that for slightly better battery life
MarcTremonti said:
How do these changes affect the battery life?
Click to expand...
Click to collapse
Lower SOTs are obvious, but to what degree is what matters. As far as I see, it's not that much.
GUYS, I recommend not to use this anymore. Oneplus has a reason to not release the kernel with conservative on . This is because conservative is not optimized for battery while interactive is. If interactive weren't optimized for battery, it'd be better than conservative for sure. I get only 4 hr sot on conservative but 5.5 on interactive. Though this is not placebo, it has a negative impact on SOT .
When I change big cluster to conservative, those cores will always stay above 2 GHz, as shown in Kernel Adiutor.
chinmai560621 said:
GUYS, I recommend not to use this anymore. Oneplus has a reason to not release the kernel with conservative on . This is because conservative is not optimized for battery while interactive is. If interactive weren't optimized for battery, it'd be better than conservative for sure. I get only 4 hr sot on conservative but 5.5 on interactive. Though this is not placebo, it has a negative impact on SOT .
Click to expand...
Click to collapse
The thread is about more performance and you're discussing battery life....of course it will have a lower battery life, the question is are you happy when your quad-core phone stutters....
chinmai560621 said:
You'll see input boost in the bottom section of CPU in kernel adiutor. Core 1- 1036, core 3 -556 is what you've to do. It'll be 1324 default on both. I'm suggesting that for slightly better battery life
Click to expand...
Click to collapse
Got it, thank you. I did notice a slightly battery improvement after two cycles.
Little investigation..
chinmai560621 said:
GUYS, I recommend not to use this anymore. Oneplus has a reason to not release the kernel with conservative on . This is because conservative is not optimized for battery while interactive is. If interactive weren't optimized for battery, it'd be better than conservative for sure. I get only 4 hr sot on conservative but 5.5 on interactive. Though this is not placebo, it has a negative impact on SOT .
Click to expand...
Click to collapse
Good you realised that.. I did a little investigation on your tweak.. Apparently setting governor to conservative locks the frequency of big cluster to max possible, usually above 2 ghz and it will step down slowly ( by slow I mean really slow like 1-2 seconds). Thus causes you to have very less render times on gpu profile as shown in the xda portal post.. I opened up ex kernel manager and observed same freq in higher range.
After some time I just changed governor to interactive and changed minimum freq of big cluster to 2 ghz.. I observed the same render times as situation as time when conservative was set as governor...
So this tweak is nothing but type of placebo since ramping up freq can lead to faster processing
Edit.: Couple of people on xda portal have said the same.. I didn't check it..

Categories

Resources