Related
Okie dokie, simple plan to directly compare clean ROMs from the cooks without opinions or bias.
I believe the general idea was to either add a new table to the Wiki or modify the existing tables in the ROMs section, to display standardised benchmark result, ie all from the same version following a set of instructions
those might be, flash ROM, hard reset, install bench mark programs, softrest, softrest run benches with radio on/off
Could be expanded to include basic battery life i dont think having call time / sms count would be helpful as its too unpredictable. but perhaps standby time and or wifi/bluetooth turned on but not connected.
I dont think its about real world tests since we cant have opinions, its a simple OS comparison.
course, for all i know there will be almost nothing between any of them making it redundant so this is why im making this thread, to talk it out and see if its worth while.
and you are starting this "new" thread as a result of my ideea and proposal
wouldn't have been nice to mention?
Benchmarking will give you a very incomplete picture of how a rom is actually going to perform, and therefore will be a virtually worthless use of your time. The real performance of a rom is gonna very significantly based on what software you use, which versions of the software you use, and sometimes even the order in which you install it. There is no consistent reliable benchmarking tool that i have seen as any useful indicator of actual speed of a rom.
The best way to find out about what rom will suit your needs is to read the frist couple posts, then kinda browse through the thread to see which bugs people are posting most regularly. When i was testing the WM6.5 roms, I had no issues with the fact that some people had exchange issues, because I was not having them, but eventually the fact that the notifications weren't working correctly caused me to change to something else. I tried another rom with some TP2 features, and generally liked it, but MyPhone didn't work correctly on that roms for whatever reason.
The point I'm trying to make is that there is no such thing as an unbiased benchmarking, because if we only test them all with no software installed, running clean then you're gonna find that almost all roms score similarly. Even more to the point, I simply have never seen a benchmarking program which had results which were indicative of any real world performance, and as such, I have disregarded the use of them entirely.
If you're happy with the rom you're on, then keep it. If not then read a few threads, see what is out there and then based on a little bit of research try one. If you don't like it, then try another one or even go back to the old one.
noris08 said:
and you are starting this "new" thread as a result of my ideea and proposal
wouldn't have been nice to mention?
Click to expand...
Click to collapse
Sorry, looking after my 3 year old today, pulling my hair out, damn cheeky monkey!but yes totally based on your idea, just wanted to make the link in the old post before it got the chop ill edit the original post in a min
scotchua said:
Benchmarking will give you a very incomplete picture of how a rom is actually going to perform, and therefore will be a virtually worthless use of your time. The real performance of a rom is gonna very significantly based on what software you use, which versions of the software you use, and sometimes even the order in which you install it. There is no consistent reliable benchmarking tool that i have seen as any useful indicator of actual speed of a rom.
Click to expand...
Click to collapse
amen to that!
just making a little bit of reading and got scared by the possibility of threads about "witch ROM is better" beeing aloud
this is going to end badly as i already know that from the trinity forum and from the diamond forum. the forum is going to be cluttered only because a few lazy sobs are not ready to spend a few moments reading the ROM's threads and drawing their own conclusions. it is enough to read the first page, one of every 3 pages and the last one in a ROM thread to understand what is all about.
not only that but when the thread is going to be too long some smart ass is going to open another (he has no time to read a l l those pages)...and than another. please, stop the madness
scotchua said:
The real performance of a rom is gonna very significantly based on what software you use, which versions of the software you use, and sometimes even the order in which you install it. There is no consistent reliable benchmarking tool that i have seen as any useful indicator of actual speed of a rom.
Click to expand...
Click to collapse
I agree, to compare ROMs completely requires opinions and subjectional review but since ive not tried this idea on different ROMs i couldnt say for sure what the results would hold.
Once upon a time, before the internet age, there were "Informatics" and "personal computers"
We had a problem: every shop sell HIS PC, assembling HW and assessing "My PC is the faster one". Also many TWEAKS were proposed for DOS and applications (This was before Egyptians I believe...)
The solution?
We built a SUITE, using a scripting tecnique, EMULATING REAL LIFE USER and measuring time, memory usage, battery load, CPU load...
Instead of using commercial benchmarks, because we don't care really about the file system speed, scrolling a large DIR matters, or rotating the screen speed.
We emulated a PRO user and a FUN user, opening a standard office file, starting and closing TT7 or video player speed with standard app (es. TCPMP) , opening an IE page, internet speed, scrolling a LOOOONG contact list, anything you consider as normal life usage.
A script could be used to start every single task and measuring the time.
AND IMHO THIS will be the killer application, comparing different ROM, devices, Diamond is faster than HD? and Kaiser? WVGA how slower is in REAL LIFE?
i belive you miss the point
this is supposed to be a tool helping people with less time or experience to choose the rom that is more appropiate to them
and maybe even help the developers to improve their work
as in the original thread is mentioned it isn't ment to say which rom "rules"
but, as i already said, if people will find this useless, or discussion will degenerate, a mod can close the thread anytime
noris08 said:
i belive you miss the point
this is supposed to be a tool helping people with less time or experience to choose the rom that is more appropiate to them
and maybe even help the developers to improve their work
as in the original thread is mentioned it isn't ment to say which rom "rules"
but, as i already said, if people will find this useless, or discussion will degenerate, a mod can close the thread anytime
Click to expand...
Click to collapse
I don't think we missed the point, just trying to share from experience why benchmarking isn't a particularly useful gauge to users trying to pick a rom. If you don't have the time to look around then i'd actually suggest just picking a rom based on what you read in the first couple posts describing it. Also that type of users should just pick a rom that is fuller with more programs installed by default.
Great idea, but will it ever be real?
While it's possible to devise a set of benchmarks and some kind of subjective scoring algorithm, the measurement accuracy will depend on too many factors and the result is likely to be unreliable and inconsistent. Seeing how many various ROMs based on the same OS builds are getting very different ratings in ROM benchmarking threads, real life performance seems to depend on every modification implemented in a ROM. E.g. even if the difference between ROMs is merely in a few files or a few registry keys, you'll probably end up comparing apples to oranges so this won't be very useful.
stepw said:
Great idea, but will it ever be real?
While it's possible to devise a set of benchmarks and some kind of subjective scoring algorithm, the measurement accuracy will depend on too many factors and the result is likely to be unreliable and inconsistent. Seeing how many various ROMs based on the same OS builds are getting very different ratings in ROM benchmarking threads, real life performance seems to depend on every modification implemented in a ROM. E.g. even if the difference between ROMs is merely in a few files or a few registry keys, you'll probably end up comparing apples to oranges so this won't be very useful.
Click to expand...
Click to collapse
I must agree to that! At the end of the day, it all comes down to what started this thread: the personal feeling a user gets when using a ROM in his own particular way of using it, which is totally different and uncomparable with other's.
it seems that i can not make you see my point
i will say it for the last time
my original ideea was NOT to compare one rom to another rom.
that will only lead us to square one - which rom is best
AND THIS IS NOT THE POINT!
whatever!
i'll give up
cheers!
tnyynt said:
I must agree to that! At the end of the day, it all comes down to what started this thread: the personal feeling a user gets when using a ROM in his own particular way of using it, which is totally different and uncomparable with other's.
Click to expand...
Click to collapse
Benchmarking suite....
1) Free memory, storage mem at first start
2) Boot up time (measured with a simple program started at the end of the boot)
3) Search a string (time)
4) Open a complex word document (time)
5) open a complex excel document (time)
5) Active sync connected, open a simple web page, time, open a a Complex web page (success, time) (IE)
6) Open a LOCAL complex web page, scrolling, (time)
7) Restoring 4000 contacts (pimbackup, success, time)
8) Restoring 2000 SMS (pimbackup, success, time)
9) TT7 startup and close (or other BIG software, time)
10) TCPMP standard video player performance (direct draw, accelerated)
3 times each test, % of battery resulting startin with a full charge (these are only examples, just to explain better the point)
noris08 said:
it seems that i can not make you see my point
i will say it for the last time
my original ideea was NOT to compare one rom to another rom.
that will only lead us to square one - which rom is best
AND THIS IS NOT THE POINT!
whatever!
i'll give up
cheers!
Click to expand...
Click to collapse
i agree to noris08!
this thread is not what was the original intention
i doen't want to know if some rom is about 10ms faster than other rom as i have readed in other threads where one published his benchmark score 2635. and a user posted the question if his device fails because he comes "only" to score 2450 e.g.
no, for example i testet 5 ROMs from 4 chefs.
i would be able to tell the reason for choosing the actual ROM on my HD. There where many facts hardly subjective opinions for choosen that! I think if a user would describe his decision respectfully to the hard work of the cockers this will help all of us.
if the guy is wrong other users can reply with their opposite experiences he knows it is worth to look for the reason on his device(configuration).
We all know speed, good feeling depends on many factors. so only a respectful discussion will help users and chefs.
even when user are telling a specific ROM is slower than other specific rom, other users can agree or disagree. if one is telling it is slow and many others reply the opposite this is a useful information for the user and the chef. chef get the response his rom seems to work fine. otherwise the chief get the information his rom leaks in speed and there are further tuning possibilities because other rom look to be faster.
I would find a thread e.g.
"Your experences with different ROMS" or
"what rom do you use and which property or feature do you like extra"
maybe useful.
If someone is posting: "ROM X ist the best" useres are old enough to know that this is not a meaningful post.
I would like to read postings like:
I am using ROM XY - has great looking german keyboard with äöüÄÖÜß - very stable (softreset max. 1 in 2 Weeks ) Battery: without Backligth and no running program and no dataconnection only 60mA i experienced a "normal" value, in suspendmode over night max. 1% Power loss although G-Alarm and phone active ) overall good responses.
are you guys so scared of whos rom will be the fastest??
end of the line the fastest rom will be the one better built. (the one more tweaked)
its like talking to ppl :
"wow its stupid comparing a dell pc to a alienware pc with the same hardware"
why??
its like if someone is a fanboy of ati and never admits a nvidia car will perform better & vice-versa
and of course as like the nvidia & ati cards, not only hardware will make a difference.
do you want to know how it really should look a ROM review?! did you ever imagine how complex a ROM is to build a ROM? do you know what is the difference between a good and a bad ROM? here u have an example from one expert i'll trust with an opinion!
the-equinoxe said:
Cooking a ROM isn't just trowing a few packages, and some registry fixes in a kitchen and press build..
It's knowing why a ROM behaves in a certain way, and fixing unwanted behaviour.
It's knowing what happens when you cook a ROM, what the scripts in your kitchen are supposed to do, and what the are actually doing(!!).
It's knowing this and so much more..
I have seen ROMs released lately where the XIP wasn't rebased, where the rgu's were contradicting the hv files and worse: where RGU from package a was contradicting the RGU from package B. (and even worse: A contradicting B but both contradicting HV..).
Why not simply import the RGU's into the HV files first? and check the rgu files, or for fraks sake, merge them!
Some had added certificates, but the cook didn't knew that the base he was using was already patched to ignore all certificates.
Why on earth clutter the ROM with unnecessary certificates? Really WHY?
I have seen cooks adding XIP of a higher build but using the OS of a lower build, just to get a high build number. (some just plain hexedit the build number :s )
This simply makes me puke, why on earth would you add an UNMATCHING XIP in another OS? You would think that that unmatched part was the cause of some unexpected bugs, wouldn't you..
Or massive amounts of files that are moved from root (=\\windows) to some kind of subdirectory, it seems that the chefs who are doing this are unaware that they aren't MOVING but COPYING the files (jups, that was bad design from microsoft)
Why concentrate on the build number?
A higher build number doesn't make a better ROM.
Focus on making a ROM BETTER, fix those contradicting registry entries, there are plenty of tools out there to ease your work nowadays, it can be done in hours instead of weeks.
Rebase those files that are supposed to be XIP, don't leave it unaltered and most certainly don't make it a PE-file (like dll or EXE), you will have unpredictable occurrences of drivers unloading from memory when doing this, and other crazy bugs that are hard to pinpoint.
And if another chef removes those dsm files and replaces all RGU's with one (or just simply stay with the HV) ,it's mostly not to piss off other chefs so it can't be shared, but to make the ROM better and faster.
Etc etc etc etc.. (I could go on and on.. really ! The curse of the kitchens I called it)
This is not a Flame to a certain Chef, or even directed at this particular Forum, I have many devices, and I have flashed them a lot, sometimes I take the time to analyze a ROM, and I have been amazed what junk has been produced by some..
The main idea of a cooked ROM is to have a better device with fewer bugs, not a fancy picture, nor to have the highest number..
So in some cases the cure is worse than the problem..
I am not going to single out a bad cook, nor a good cook, there are plenty of both..
Some will see this as a personal attack, personally: I don't care, if you claim to be a Top-chef, but don't know the basics of cooking, call yourself what you want, but don't expect me (or others) to be fooled.
I am just saying: instead of focusing on a high build number, or a (bug ridden) beta, focus on what you are actually doing.
Take a stable ROM, and make that one better, most AKUs are intended to support newer devices, and newer hardware, a higher AKU doesn't mean the ROM is better (it could have new bugs to deal with).
Click to expand...
Click to collapse
the whole post is here:
http://forum.xda-developers.com/showpost.php?p=3387292&postcount=76
KukurikU said:
did you ever imagine how complex a ROM is to build a ROM?
Click to expand...
Click to collapse
sorry, but this does not subject this thread and nobody doubts about ROM cooking is not a simple job! thanks to all chefs!!!!
KukurikU said:
do you know what is the difference between a good and a bad ROM?
Click to expand...
Click to collapse
from the user's viewpoint i think its clear and software is made for users. user do not mandatory want to know whats behind the scenery in deepth. user want a device with software with basic properties and naturally many cool features - lets say:
a stable ROM with all features he needs for daily use, cool features for fun and suitable speed. if the userinterface have a appealing design also it's nearly perfect!
and here we are back again:
"what features or properties do you like most"
would help users to find out, what is the best for him!
this is naturaly only my opinion
P.S: Sorry for my suggestion to open a thread, i just found a thread with such a Subject!!!!!
autdev said:
sorry, but this does not subject this thread and nobody doubts about ROM cooking is not a simple job! thanks to all chefs!!!!
from the user's viewpoint i think its clear and software is made for users. user do not mandatory want to know whats behind the scenery in deepth. user want a device with software with basic properties and naturally many cool features - lets say:
a stable ROM with all features he needs for daily use, cool features for fun and suitable speed. if the userinterface have a appealing design also it's nearly perfect!
and here we are back again:
"what features or properties do you like most"
would help users to find out, what is the best for him!
this is naturaly only my opinion
P.S: Sorry for my suggestion to open a thread, i just found a thread with such a Subject!!!!!
Click to expand...
Click to collapse
cool features and appealing design in a rotten ROM means high battey consumption, frequent freezes and soft resets and so on...
i think i wasn't clear enough in my previous post. what i meant (in a nut shell) was that because of it's complexity deciding that a ROM is good, less good or bad is a professional's job. only an expert can give you a trustworthy opinion.
the moment somebody starts a "what is the best ROM for me" everybody will jump in and push the ROM he is using. after a while the thread will be cluttered with hundreds of honest but unprofessional opinions and you'll have on one page 8-9 different opinions. does this make yr decision of choosing a ROM easier?? i don't think so! so, back to square one
in my wet dreams i see a thread filled with posts of "ROM critics" that are analyzing each new ROM. reading such reviews could be very useful indeed.
in short: if u are a chef and own a kitchen than beware of the "food" critics that are visiting yr restaurant. they can kill yr business in tomorrows front page ROM review or make you a very, very rich and famous chef
ok So to conclude,
we cant use benchmarks since the difference between them would be next to nothing
we cant judge a ROM based on its funtionality since its purely opinionated
we could potential use a script to run a series of real world tests that become useless after a few resets and number of programs installed
That pretty much knocks the idea on the head then?
dazza9075 said:
ok So to conclude,
we cant use benchmarks since the difference between them would be next to nothing
we cant judge a ROM based on its funtionality since its purely opinionated
we could potential use a script to run a series of real world tests that become useless after a few resets and number of programs installed
That pretty much knocks the idea on the head then?
Click to expand...
Click to collapse
sure looks that way.
I would all come down to every user deciding on his own. Would you believe that after trying lots of ROMs I've reverted to a certain Stock ROM because I find that it offers close to perfection for my needs? I could praise the speed and stability and I'd advise you to use it, but you 'd most probably find that it would not suit your needs.
Anyways, I've offered to support a thread that is civilized and based on factual opinions, I will keep my word in doing so, if there's be such thread.
Here's another idea for you all: a good factor to take into consideration when judging a ROM is the number of users using it. It's a good indicator of the ROM's value, since X no. of users are hangin' on to that. Why not post and maintain a simple poll with the most common ROMs (stock and cooked) and see where it gets you?
additional details on test:
The first part of the table is to test the rom with the stock kernel.
The second part is to see how well it performs oc/uc.
This is supposed to show how well a ROM performs by itself and how well it works while oc/uced since a lot of us do use custom kernels.
On all of the tests, there should be no widgets or apps installed that didn't come preset with rom(excluding setCPU, fps2d,quadrant, and linpack). This should help eliminate variability of what apps different users have.
On all the test (performance or battery) if the rom has preset profiles they can use them as long as it doesn't require a reboot or any 3rd party apps. all these profile should be mentioned though in additional notes.
I said no profiles on the battery test, meaning no setCPU profile due to variability of profiles.
All ROMs will be using sense except CM6
Update 8/6: Updated the spread sheet and put it up and google docs Niefs Rom Comparison
There are now explanations of the tests and I am going to start testing today. Also, looking for testers.
I have posted my scores for part of Fresh.
update: 8/12
Most ROMs are completely filled out under features
Most performance test have been ran
Few battery tests
http://i996.photobucket.com/albums/af90/nief1313/test.jpg
{
"lightbox_close": "Close",
"lightbox_next": "Next",
"lightbox_previous": "Previous",
"lightbox_error": "The requested content cannot be loaded. Please try again later.",
"lightbox_start_slideshow": "Start slideshow",
"lightbox_stop_slideshow": "Stop slideshow",
"lightbox_full_screen": "Full screen",
"lightbox_thumbnails": "Thumbnails",
"lightbox_download": "Download",
"lightbox_share": "Share",
"lightbox_zoom": "Zoom",
"lightbox_new_window": "New window",
"lightbox_toggle_sidebar": "Toggle sidebar"
}
I think it will be a great idea... google docs might be a good way to let people collaborate, otherwise some kind of wiki.
Might be easier to organize by features such as "stripped of sprint apps" or "added X app" vs things like battery and performance which can sometimes be highly variable by device. Also, if you run performance I would maybe do a baseline of 998mhz (since we all know we can hit this clock) and then another showing oc'd (since some people can barely OC one tick).
I think there are enough people across all the roms that if the spreadsheet were public you'd gather a lot of data. I also have NAND backups of a few of the froyo roms so it wouldn't be hard to boot into recover, reflash a nand, run the benches, then reflash back to whatever i was on.
The battery ones will be a little tough #1 cuz i'm not sure how many people wanna completely lock up their phone for several hours waiting to see how long it lasts. #2, depending on how some people's batteries are calibrated and what they're running in the background, MMV.
Great idea tho!!! I think the ultimate key is thinking of a good way for everyone to easily submit their data. Anyone have suggestions?
u could add it to our wiki and update it like monthly. idk just an idea
Looking forward to it
This is a great idea, and i look forward to seeing this list grow. This could really help someone new to Android like me.
toastcfh said:
u could add it to our wiki and update it like monthly. idk just an idea
Click to expand...
Click to collapse
I think adding it to the wiki would be really nice. I will try to come up with a spread sheet that only a few can edit(I don't want one person to delete or mess up all the data). I rather have were people pm there scores and other data so i can add it in.
I will finish tweaking on how to tests and how the spreadsheet will look tomorrow.
You should add 2 additional categories:
Does WiFi Tethering work with 3G speeds Y/N
if yes, which version of Wifi Tethering
Does WiFi Tethering work with 4G speeds Y/N
if yes, which version of Wifi Tethering
rek410 said:
You should add 2 additional categories:
Does WiFi Tethering work with 3G speeds Y/N
if yes, which version of Wifi Tethering
Does WiFi Tethering work with 4G speeds Y/N
if yes, which version of Wifi Tethering
Click to expand...
Click to collapse
I have just added that to the spreadsheet, the versions will be written down in the additional notes. Also, i just posted the link to the spreadsheet in the OP. I will be looking for testers if anyone is interested to help.
nief, im running EViO 2 V1.0.2
ill post results for you later.
nenn said:
nief, im running EViO 2 V1.0.2
ill post results for you later.
Click to expand...
Click to collapse
Sweet thanks for the help.
Which version of CM6 were you going to test? RC1 is getting pretty old, a lot of us are using nightlys. RC2 should be out soon though.
OgremustCrush said:
Which version of CM6 were you going to test? RC1 is getting pretty old, a lot of us are using nightlys. RC2 should be out soon though.
Click to expand...
Click to collapse
I was planning on testing RC1... at least until RC2 has been released.
if some one wants to do the tests for a nightly build, i will gladly add another row
I noticed you had an over clocked set to test performance but for battery you have it set not to test profiles for example in battery. this can be a huge disadvantage for some roms (ie DC has profiles for performance and battery that can make a huge difference int here DConfig.) also the kernels can greatly affect these. If we wanted to actually test the rom shouldn't we be testing these with either 1 the same kernel for all or their stock kernels?
for example someone using the stock Fresh kernel may have vastly different results then someone using netarchy's kernel. this would provide skewed results for the rom test.
Same goes for Baked Snacks since there are several kernel options. if using the performance kernel for the performance test then switching to battery saving for battery saving test you will get results showing unachievable results.
I think this is a great idea but if this is going to happen i think "guidelines" should be used so that the data is as close to comparable and accurate as possible. what do you think?
something like
1) for testing purpose you should use the stock kernal with the rom for all tests.
2) with the exception of the OC tests you can not be running setCPU or any other under/overclock programs.
3) the following settings should be used. <list some basic settings>
4) if your rom has built in profiles settings that do not require a reboot or flashing anything you CAN use them.
5)etc
6)etc
If we dont have guidelines i think the results will very greatly on settings and kernel.
Even with doing the above you will still have a decent variance from usage/apps/widgets but at least this would make it as close to a level playing field and testing of the rom itself rather then the kernel/settings/who has what in setcpu
omegasun18 said:
I noticed you had an over clocked set to test performance but for battery you have it set not to test profiles for example in battery. this can be a huge disadvantage for some roms (ie DC has profiles for performance and battery that can make a huge difference int here DConfig.)
Click to expand...
Click to collapse
I agree. Using performance profile for the performance test and then battery profile for the battery test will lead to skewed results. Perhaps what should be done is to test each profile against all tests. So you might have DC/Performance Profile and DC/Battery Profile with results for every test. This would tell you how well the battery performs under the performance profile or how well the rom performs under the battery profile.
omegasun18 said:
I noticed you had an over clocked set to test performance but for battery you have it set not to test profiles for example in battery. this can be a huge disadvantage for some roms (ie DC has profiles for performance and battery that can make a huge difference int here DConfig.) also the kernels can greatly affect these. If we wanted to actually test the rom shouldn't we be testing these with either 1 the same kernel for all or their stock kernels?
for example someone using the stock Fresh kernel may have vastly different results then someone using netarchy's kernel. this would provide skewed results for the rom test.
Same goes for Baked Snacks since there are several kernel options. if using the performance kernel for the performance test then switching to battery saving for battery saving test you will get results showing unachievable results.
I think this is a great idea but if this is going to happen i think "guidelines" should be used so that the data is as close to comparable and accurate as possible. what do you think?
something like
1) for testing purpose you should use the stock kernal with the rom for all tests.
2) with the exception of the OC tests you can not be running setCPU or any other under/overclock programs.
3) the following settings should be used. <list some basic settings>
4) if your rom has built in profiles settings that do not require a reboot or flashing anything you CAN use them.
5)etc
6)etc
If we dont have guidelines i think the results will very greatly on settings and kernel.
Even with doing the above you will still have a decent variance from usage/apps/widgets but at least this would make it as close to a level playing field and testing of the rom itself rather then the kernel/settings/who has what in setcpu
Click to expand...
Click to collapse
I agree with a baseline but there should also be a performance test done where you have it maxed to the peak and state the battery life you'd get with those settings and numbers.
......
now that I think about it....
Really in the end we are speaking real world examples here and itd be best to just have the results state what rom, what kernel, and what settings you had to get "X" results and then go form there.
I dont care if DC is better on stock kernal and settings than BS. I'd rather know what settings in BS yield what results for people and the same for any other rom.
With the number of changes the dev's put in the roms doing baselines against them is kinda pointless as they might each take advantage of certain things differently. It'd be best to just notate the results for each post and state the changes made to yield the results so people can get more real world conclusions.
sgt. slaughter said:
I agree with a baseline but there should also be a performance test done where you have it maxed to the peak and state the battery life you'd get with those settings and numbers.
......
now that I think about it....
Really in the end we are speaking real world examples here and itd be best to just have the results state what rom, what kernel, and what settings you had to get "X" results and then go form there.
I dont care if DC is better on stock kernal and settings than BS. I'd rather know what settings in BS yield what results for people and the same for any other rom.
With the number of changes the dev's put in the roms doing baselines against them is kinda pointless as they might each take advantage of certain things differently. It'd be best to just notate the results for each post and state the changes made to yield the results so people can get more real world conclusions.
Click to expand...
Click to collapse
The thing is if you do what you suggest .. you arent comparing each roms performance really. If this is supposed to be used as a sort of guide for people to compare the roms themselves then what you said doesnt make sense since it adds a large a mount of varibales tot he "tests". Now if we are trying to compare complete packages and find what works best sure i agree with you completly. however i would not consider that to be comparing roms. which is what the origonal intent LOOKS to be.
When you are trying to compare things to get the most accurate results you compare them in as close to the same conditions and set up as possible. The more you differentiate the things you are comparing the more aspects can influence the results and the more skewed or inaccurate the results are.
If you have extremely random results based off a wide variety of conditions then the results lose there usefulness. In this case if we are trying to test ROMS then we should try to single out the roms and get them as close tot he same footing as we can so we can compare the ROMS and not the tweaks everyone can do. Maybe having another table or thread comparing different settings and tweaks and kernels would be helpful as well?
By stock kernel i mean the stock kernel that comes with each rom. The devs tweak these and make changes to give better performance/battery. I dont mean the stock as in the original froyo kernel. I also disagree that this isnt real world examples. I think it is exactly what it is however it takes out individual tweaks from the user and tries to keep everythign as close to level as possible to yield the best TEST results
Still going to say these will be nice but not what people will be looking for. So what if BS beats out DC outa the box in terms of bat life and performance. Thats nice and all but if you could show that with a few tweaks with set cpu and other things included in the rom that that DC greatly outperforms BS then its completely different.
What really needs to be done is what we both are saying. one done stock with the included kernel with the rom(or whatever the developer suggests using), and then one where users can make there tweaks and post those results.
If I just knew that one rom had better battery life stock and didn't know that I could possibly use a different rom that with a few tweaks gets better battery life and performance I'd rather use the later. Some roms perform differently than others with regards to tweaks being done by the user.
I'd stress that two tests be done, one with stock rom with stuff loaded form the dev's page only, and then one done with tweaks by the user and state the results so we can really get an idea what rom is best at what performance wise at least.
I was wondering if there would ever be something like this!! Great job!!! ...just on the side note, is there anyway to put the features working on the left side closer to the rom? i know a lot of users who care more about working features than technical stuff (linpack etc). I too would love to see this on the wiki!
**edit: also, maybe consider adding kernel box so we can see which kernel is being used w/ which rom and working features etc.
The first part of the table is to test the rom with the stock kernel.
The second part is to see how well it performs oc.
This is supposed to show how well a ROM performs by itself and how well it works while oced since a lot of us do us custom kernels.
On all of the tests, there should be no widgets or apps installed that didn't come preset with rom(excluding setCPU, fps2d,quadrant, and linpack). This should help eliminate variability of what apps different users have.
On all the test (performance or battery) if the rom has preset profiles they can use them as long as it doesn't require a reboot or any 3rd party apps. all these profile should be mentioned though in additional notes.
I said no profiles on the battery test, meaning no setCPU profile due to variability of profiles.
adeyo said:
I was wondering if there would ever be something like this!! Great job!!! ...just on the side note, is there anyway to put the features working on the left side closer to the rom? i know a lot of users who care more about working features than technical stuff (linpack etc). I too would love to see this on the wiki!
**edit: also, maybe consider adding kernel box so we can see which kernel is being used w/ which rom and working features etc.
Click to expand...
Click to collapse
There is a kernel box already there, its to the very right. THis is the custom kernel used during the oc test.
nief1313 said:
The first part of the table is to test the rom with the stock kernel.
The second part is to see how well it performs oc.
This is supposed to show how well a ROM performs by itself and how well it works while oced since a lot of us do us custom kernels.
On all of the tests, there should be no widgets or apps installed that didn't come preset with rom(excluding setCPU, fps2d,quadrant, and linpack). This should help eliminate variability of what apps different users have.
On all the test (performance or battery) if the rom has preset profiles they can use them as long as it doesn't require a reboot or any 3rd party apps. all these profile should be mentioned though in additional notes.
I said no profiles on the battery test, meaning no setCPU profile due to variability of profiles.
Click to expand...
Click to collapse
Thanks for clearing up what your vision for this is.
sgt slaughter. I agree with a lot of your points. I think 2 sets could be done to better show a variety of results. I was simply arguing the merits of doing a rom specific test comparison vs a complete setup specific comparison. You are right that different roms react differently to different tweaks that was the point. BUT not only do different roms act differently, different evos with the same everything react differently to different tweaks and settings. Thus just because user A get these results doesnt always mean User B will too. that's why i thought baseline would be a better gauge on the roms themselves.
I have yet to find any decent threads anywhere regarding the best settings for the Nexus 7 so I thought I would start and hopefully learn some info from other users and provide a good resource thread for others.
I have not overclocked and haven't added any cutom tweaks to governors and schedulers yet so my experience so far has been with the default kernel options. Since hardware is different and kernel/rom experiences will vary I will just start with Scheduler, Governor, and other general tweaks.
Scheduler:
i have so far had the best experiences with SIO, and NOOP
Governor:
So far the best balance between performance and battery life has been with Pegasusq and Interactive
ADJ and LMK:
I don't use any scripts like Supercharger I like to do everything manually, here are the values I have found that offer the smoothest user experience:
ADJ:
0,1,3,6,9,15
1536,9216,25600,51200,76800,102400 - (Wife's for general use and multi-tasking)
ADJ:
0,1,4,7,10,15
5120,15360,61440,81920,102400,122880 - (Mine for gaming and high performance)
I have boat-loads of tweaks that I have used for roms I have built for other devices, but hesitant to use them with the Nexus 7 due to the fact that I don't want to cause any negative effects. I have added generic linux tweaks (ext4, database defrag, kernel tweaks) but the Adreno and Powervr tweaks obviously are null and void for this. As I discover more I will definitely share.
I have had issues getting custom LMK values loaded on some kernels, but franco and anything based off Motley has worked just fine. I know there is a bug with LMK on some source versions and assume it is related to that, but if anyone has found a workaround i would be very interested as I would love to give the other kernels a shot.
I am intrigued about lulzactive but the parameters need to be tweaked as the default settings provide horrible battery life. I am interested to know what other people have been using and any info you want to provide is welcome and greatly appreciated!
The problem with such threads and people sharing all this info is everyone uses something different, meaning what tweaks and settings you use for your given device are dependent also on the ROM and the kernel you're using so...
You can see how that basically ends up being a thread full of info where people just end up saying "oh yeah, well my <whatever device> can get <however many hours> of battery life..." and so on.
It's a good idea I suppose but it only ends up providing benefits to those people that a) end up using the same hardware and b) end up using the same ROMs and c) end up using the same kernels too. And yes I know this is the Nexus 7 so that covers the device aspect right there...
Suffice to say this: none of the tweaks or info you provided in your post are relevant to me at all with my Nexus 7 as the kernel I use (M-Kernel a37) doesn't support the kinds of governors that are considered the more esoteric ones (it has interactive, ondemand, touchdemand, and performance) nor the same class of schedulers (it has noop, deadline, row, and bfq). Would be nice but, I look at it this way:
Both the Paranoid Android developers (any and all of them) as well as Metallice who's responsible for M-Kernel, have done a metric frak-ton of work before I even got a Nexus 7 with testing, builds, more testing, more builds, testing, more testing, even more testing, and they continue to develop and test even to this moment and my Nexus 7 runs fabulously, I honestly can't ask for more considering the performance (even with the default of 2 cores enabled with my choice of kernel) and battery life (which is damned amazing as well).
But who knows, maybe something useful will show up... I've just gotten to a point where "It just works" and that's good enough for me. Tweaking is just boring nowadays, but that's just my opinion.
<all this is coming from a guy that's been tweaking computers and consumer electronics of all kinds, shapes, and sizes for several decades now, and yes I can honestly say "I've pretty much seen it all...">
br0adband said:
You can see how that basically ends up being a thread full of info where people just end up saying "oh yeah, well my <whatever device> can get <however many hours> of battery life..." and so on.
It's a good idea I suppose but it only ends up providing benefits to those people that a) end up using the same hardware and b) end up using the same ROMs and c) end up using the same kernels too. And yes I know this is the Nexus 7 so that covers the device aspect right there...
Click to expand...
Click to collapse
Given that this is the Nexus 7 forum, <whatever device> = <Nexus 7>
richardorvince said:
Given that this is the Nexus 7 forum, <whatever device> = <Nexus 7>
Click to expand...
Click to collapse
I covered that in the last sentence you just quoted.
I know I only have 9 posts here and that makes me a "noob" in the eyes of most people here, but I am by no means a noob in general, just a noob when it comes to tegra. I have boosted performance and battery life (up to 50%) on other devices and don't see any reason why it can't be done on this one.
I already touched on the fact that opinions are just that, experiences may vary, but when it comes to governors there is a discernable difference when it comes to battery life and performance and that is what I am inquiring about. There are also tunables which can change a governor from completely horrible to exceptional. THis is why I am inquiring. Don't bother yourself with this thread if it is of no use to you.
If you'd take the time - if you haven't already - to read with comprehension the threads for kernels like franco.Kernel and M-Kernel you'd see a massive if not overwhelming amount of information that you're actually looking for that's already been done. Those two kernels - and probably most of the other popular ones - have extensive levels of research and testing already done on them and in their most currently available forms they are the epitome of kernel development on the Nexus 7, and I stand by that statement even in spite of me not being a developer myself. I have read thousands upon thousands of posts and done my own little amounts of experimentation just to prove to myself that yes, those two kernels (and probably others) provide exactly what their respective developers have worked so hard to provide:
Exceptional performance overall coupled with exceptional battery life as well. Yes you can do your own tuning for your own particular needs and wants and even your own requirements as needed which is pretty much what you're apparently hoping to discover from other people and their own experiences.
I'm not saying this thread is a bad idea, I'm saying it's been done before: hundreds if not thousands of times in the very threads that exist for each of the respective kernels (and ROMs by association). The kernel threads have extreme amounts of info on basic tunables, governors, schedulers, pretty much every last thing you or anyone else could ever do to squeeze out the best they offer and the the developers implement that stuff into the defaults of their kernels based on the results of testing and reports from users implementing them.
The info is out there, you just have to go find it - and yes I note that you said you'd never found any decent threads about the Nexus 7 which is precisely what prompted my first reply: the info is already out there in the kernel and ROM threads.
There can't be a "best of" type thread like this because there's just so many different variations of kernels, ROMs, etc even if the device is the same amongst all the people.
Also, one aspect that most people don't realize: there are different revisions of the Nexus 7 hardware and they tend to perform differently. I, for example, have an original (and I mean first production run) Nexus 7 8GB model that has totally different performance characteristics than the later revisions of the same 8GB model, and then the 16GB and 32GB models are even different on another level because they use faster Flash-RAM onboard than the 8GB models ever did.
So again, it's a great idea and I figured if nobody else ever created one I'd do it myself, but in the long run every device is going to perform differently. Even if you had two Nexus 7s that came off the production line one right after the other and you tested them with the exact same ROMs, kernels, and tweaks, you'd get different results from each one. I'm not being facetious when I say that, I'm being totally absolutely dead serious.
Also, there is no really good overall benchmark for Android, probably never will be. You have to use different ones geared towards specific purposes to get any useful data, like AndroBench which tests the storage specifically and Vellamo which can test the HTML/surfing performance specifically. Antutu is probably the best actually useful overall system benchmark but unfortunately more people use Quadrant which is a shame because that one is so useless the results can vary rather dramatically even if you run it several times in a row.
Believe me, I've been tweaking for decades, so there's nothing I can say or do that'll stop anyone from doing it - I know the bug of "Upgraditis" quite well and have suffered from it forever, right alongside the companion condition known as being a Tweakaholic...
br0adband said:
If you'd take the time - if you haven't already - to read with comprehension the threads for kernels like franco.Kernel and M-Kernel you'd see a massive if not overwhelming amount of information that you're actually looking for that's already been done. Those two kernels - and probably most of the other popular ones - have extensive levels of research and testing already done on them and in their most currently available forms they are the epitome of kernel development on the Nexus 7, and I stand by that statement even in spite of me not being a developer myself. I have read thousands upon thousands of posts and done my own little amounts of experimentation just to prove to myself that yes, those two kernels (and probably others) provide exactly what their respective developers have worked so hard to provide:
Exceptional performance overall coupled with exceptional battery life as well. Yes you can do your own tuning for your own particular needs and wants and even your own requirements as needed which is pretty much what you're apparently hoping to discover from other people and their own experiences.
I'm not saying this thread is a bad idea, I'm saying it's been done before: hundreds if not thousands of times in the very threads that exist for each of the respective kernels (and ROMs by association). The kernel threads have extreme amounts of info on basic tunables, governors, schedulers, pretty much every last thing you or anyone else could ever do to squeeze out the best they offer and the the developers implement that stuff into the defaults of their kernels based on the results of testing and reports from users implementing them.
The info is out there, you just have to go find it - and yes I note that you said you'd never found any decent threads about the Nexus 7 which is precisely what prompted my first reply: the info is already out there in the kernel and ROM threads.
There can't be a "best of" type thread like this because there's just so many different variations of kernels, ROMs, etc even if the device is the same amongst all the people.
Also, one aspect that most people don't realize: there are different revisions of the Nexus 7 hardware and they tend to perform differently. I, for example, have an original (and I mean first production run) Nexus 7 8GB model that has totally different performance characteristics than the later revisions of the same 8GB model, and then the 16GB and 32GB models are even different on another level because they use faster Flash-RAM onboard than the 8GB models ever did.
So again, it's a great idea and I figured if nobody else ever created one I'd do it myself, but in the long run every device is going to perform differently. Even if you had two Nexus 7s that came off the production line one right after the other and you tested them with the exact same ROMs, kernels, and tweaks, you'd get different results from each one. I'm not being facetious when I say that, I'm being totally absolutely dead serious.
Also, there is no really good overall benchmark for Android, probably never will be. You have to use different ones geared towards specific purposes to get any useful data, like AndroBench which tests the storage specifically and Vellamo which can test the HTML/surfing performance specifically. Antutu is probably the best actually useful overall system benchmark but unfortunately more people use Quadrant which is a shame because that one is so useless the results can vary rather dramatically even if you run it several times in a row.
Believe me, I've been tweaking for decades, so there's nothing I can say or do that'll stop anyone from doing it - I know the bug of "Upgraditis" quite well and have suffered from it forever, right alongside the companion condition known as being a Tweakaholic...
Click to expand...
Click to collapse
Blah Blah Blah
I mean seriously dude, that is a lot of words for saying absolutely nothing.
I am a developer on other devices, and I will be releasing roms for this one as well. I was looking for personal experience, not an opinion from someone who has absolutely nothing to add. I have been a member of this site for over six months, and I am just now hitting the "10 post requirement" to post in the developer forums. Why?? Because I am not out to make a name for myself on this website, I already have a reputation - a good one - on others, places where post counts and titles mean nothing unless you have actually developed something.
I work and have a family so I don't have tons of time to test out schedulers and governors so I was looking for some info from others and gave some of my own in return. I am not a tweakaholic, but you will thank me when I release an optimized ROM with extended battery life and a fluid and smooth user experience.
This site is great in a lot of ways, but I have noticed there are a lot of egotistical holier than thou people on the forums who like nothing better than to put other people down and be (for lack of a better word) ***holes. If your hobby is to crap on other peoples threads then it is time for a life change, get out and find a girl-friend, or friends in general.
You're a ROM developer and a member at the biggest most popular site on the entire Internet for ROM development for mobile devices and it took you six months to get to 10 posts? Guess I'm missing something but whatever, you go right on going on, son, I'll keep an eye out for your progress.
As for thanking you for ROMs, no, that won't happen since you're too far behind the curve now and anything you come up with will basically be copied work that others before you have long since created and improved upon (even if you don't realize it). If you're that busy with 'real life' and whatever - which always seems to be the reason, go figure - then perhaps you don't have the time to focus on such tasks in the first place. Priorities, man, priorities.
Ad hominem sure seems to be the norm for people online these days, wonder why.
Fact - 95% of the battery tweaks are placebo at best.
The remaining come from kernel developers who have a firm grasp on undervolting and can actually get a little longer battery life out of your device.
The best blanket solution to increasing your battery.. Turn off Location Access and Google Now.
Yes my job and family are my priority. As they should be. Also my point was I am not trying to boosty my post count to show how "cool" I am, I could care less. As it should be.
And no battery tweaks are not placebo, unless you are using placebo battery tweaks of course. There are Build.prop, init.d and sysctl.conf tweaks that can boost battery life 20-30% easy, and it can be proven. No placebo. If you choose to believe that then I have no problem with it.
I thank both of you for littering this thread with crap. I will not bother with this one I have much better things going on other forums. I truly wish I knew as much as you guys, It must suck to know everything because life must be truly boring for you.
Hi there,
I was unsure if I should ask this in the Oppo Forum or here, if I am in the wrong place just tell me!
As for my question: I just wanted to know what kernel other user owning the Find 5 and running Omni Rom are using.
I was kind of unable to find a fit when i searched in the Find 5 Forum in "Android Original Development" (this was the right forum right?). So I am happy about any reply. If you could drop a few words on why you use this very kernel, even more!
If I was just plain dumb/blind asking this question cause it was answerd a thousand times before - please give me directions, I really didn't find what I was looking for.
chuSmu
Every oppo find 5 kernel for 4.3/4 is a custom one since coloros is still at android 4.2
Which is what oppo shares on their github.
For 4.3/4 there are AFAIK only two kernels
-based on the omnirom one
-from cfxe http://www.oppoforums.com/threads/a...-y-codefirexperiment-nightlies-weeklies.5630/
Dont know about CM
maxwen said:
Every oppo find 5 kernel for 4.3/4 is a custom one since coloros is still at android 4.2
Which is what oppo shares on their github.
For 4.3/4 there are AFAIK only two kernels
-based on the omnirom one
-from cfxe http://www.oppoforums.com/threads/a...-y-codefirexperiment-nightlies-weeklies.5630/
Dont know about CM
Click to expand...
Click to collapse
Ok, thanks a bunch - that explains it. I concluded from your post that 4.3 kernels work on 4.4 too? Is that correct?
But i thought that there were only two major differences between kernels for the same device and the same
android version - sense and aosp. Your words were "the only ones based on omni rom..." does that mean there are more
differences i have to look out for? Sorry.. i really still am a noob when it comes to kernels and stuff.
Keep in mind that modified system configurations (such as changing the kernel) are NOT supported by Omni.
If you are running a kernel other than the included one, your configuration becomes invalid for any bug reports.
Entropy512 said:
Keep in mind that modified system configurations (such as changing the kernel) are NOT supported by Omni.
If you are running a kernel other than the included one, your configuration becomes invalid for any bug reports.
Click to expand...
Click to collapse
Thanks for the info - I did not know this..it might actually keep me from installing another kernel. It was just that i wanted to toy a little around wih it (after reading some more ^^) simply cause I have never done it before and wanted to experience firsthand how it can affect performance and battery life.
chuSmu said:
Thanks for the info - I did not know this..it might actually keep me from installing another kernel. It was just that i wanted to toy a little around wih it (after reading some more ^^) simply cause I have never done it before and wanted to experience firsthand how it can affect performance and battery life.
Click to expand...
Click to collapse
Many of the tweaks that people install custom kernels for can be simply achieved by altering the tuning settings of their existing kernel.
For example, those who really like battery life may wish to choose the conservative governor and set aggressive up/down thresholds. (up threshold 90 and down of 60 is what I use)
Entropy512 said:
Many of the tweaks that people install custom kernels for can be simply achieved by altering the tuning settings of their existing kernel.
For example, those who really like battery life may wish to choose the conservative governor and set aggressive up/down thresholds. (up threshold 90 and down of 60 is what I use)
Click to expand...
Click to collapse
Yeah that's what I was looking for. Though i failed to realize that these options are provided ^^. It's kinda embarassing cause i only noticed
them when i wanted to switch apps with switchtr and pulled out the sliding bar by accident.
So yeah i had what i wanted all along .. Well anyways ... thanks for providing the settings you use. I'm gonna try them!
chuSmu
Removed
Maybe this will answer your question
I had read that at his end, but I was curious why I never saw it discussed anywhere by anyone. He has a good thing going there.
Megaflop666 said:
I had read that at his end, but I was curious why I never saw it discussed anywhere by anyone. He has a good thing going there.
Click to expand...
Click to collapse
I was intrigued as well. There are rules pertaining to kernels here in xda. If a dev won't comply, the thread gets to be closed down.
saywhatt said:
I was intrigued as well. There are rules pertaining to kernels here in xda. If a dev won't comply, the thread gets to be closed down.
Click to expand...
Click to collapse
Yes I know the rules, but I figured someone would've started a discussion in the general area. Oh well, guess I just did.
I don't know what I expected actually, my experience on any forum is that most people don't research on their own, they come to forums and if the answer isn't there, they give up. I love googling for new undiscovered gems. :thumbup:
Hi!
First of all, I would like to thank you wholeheartedly for the interest in ShinTo Kernel and your time in testing it.
I am sincerely blown away by the spike in interest... Just wow!
In the space of two months I have moved twice across Europe and am currently getting settled in my new home (and waiting for the darn broadband).
I can assure you two things; ShinTo Kernel & PreCog.me are not a hobby.
They are my daily work focus (as in my daily job).
I could extend myself talking about the great projects I have but I am the kind of person that prefers to prove with facts so... Keep an eye on PreCog.me
Ps: i don't want to monopolize the thread so... If you have any specific questions or I can help out in any way, then please do ask
Ps2: I am eternally grateful to xda, civato and all the great ppl here...
Note3 N9005 running the insanely brutal [ShinTo Kernel] v028a005 yeaaaah!
For some interesting stuff why not head to PreCog.me
CekMTL said:
Hi!
First of all, I would like to thank you wholeheartedly for the interest in ShinTo Kernel and your time in testing it.
I am sincerely blown away by the spike in interest... Just wow!
In the space of two months I have moved twice across Europe and am currently getting settled in my new home (and waiting for the darn broadband).
I can assure you two things; ShinTo Kernel & PreCog.me are not a hobby.
They are my daily work focus (as in my daily job).
I could extend myself talking about the great projects I have but I am the kind of person that prefers to prove with facts so... Keep an eye on PreCog.me
Ps: i don't want to monopolize the thread so... If you have any specific questions or I can help out in any way, then please do ask
Ps2: I am eternally grateful to xda, civato and all the great ppl here...
Note3 N9005 running the insanely brutal [ShinTo Kernel] v028a005 yeaaaah!
For some interesting stuff why not head to PreCog.me
Click to expand...
Click to collapse
Already a member at precog and get your email updates. I've been watching your progress and am currently test driving your most current version. :thumbup:
Thank you so much!
Expect v28a005 by mid next week if all goes well.
Note3 N9005 running the insanely brutal [ShinTo Kernel] v028a005 yeaaaah!
For some interesting stuff why not head to PreCog.me
CekMTL said:
Thank you so much!
Expect v28a005 by mid next week if all goes well.
Note3 N9005 running the insanely brutal [ShinTo Kernel] v028a005 yeaaaah!
For some interesting stuff why not head to PreCog.me
Click to expand...
Click to collapse
ooooh i can't wait! :highfive:
ShinTo Kernel v028a005 has over 50 updates/enhancements/fixes and it is compiled with the latest bleeding edge Linaro Toolchain GCC 4.9.2
I also fixed the cosmetic issue of not displaying the kernel name correctly in the TW settings app.
If there are no unforeseeable issues, it should be released around the middle of the week
CekMTL said:
ShinTo Kernel v028a005 has over 50 updates/enhancements/fixes and it is compiled with the latest bleeding edge Linaro Toolchain GCC 4.9.2
I also fixed the cosmetic issue of not displaying the kernel name correctly in the TW settings app.
If there are no unforeseeable issues, it should be released around the middle of the week
Click to expand...
Click to collapse
And it runs smooth as silk....
Just to update you guys; there's a slight delay with QA, but it will be released this week, probably tomorrow or friday.
Thx!
Is Faux Sound compatible with this kernel?
B3311 said:
Is Faux Sound compatible with this kernel?
Click to expand...
Click to collapse
Yes it is
ShinTo Kernel for note 3 has been released!
Can anyone post a cycle of their battery life with this kernel? And a benchmark would be nice too(I know what you'll say but please one would be appreciated thanks!). Thanks!
Graffiti Exploit said:
Can anyone post a cycle of their battery life with this kernel? And a benchmark would be nice too(I know what you'll say but please one would be appreciated thanks!). Thanks!
Click to expand...
Click to collapse
Why?
Your setup and phone use is completely unique. Anyone else's battery life will not give you the slightest clue as to what yours might be.
Flash it and see for yourself.
Graffiti Exploit said:
Can anyone post a cycle of their battery life with this kernel? And a benchmark would be nice too(I know what you'll say but please one would be appreciated thanks!). Thanks!
Click to expand...
Click to collapse
Benchmarks are different from device to device. They're not a true statement about a kernels performance. I get insane battery, I get lightning fast speeds and I love using the kernel. This is a discussion of the kernel in this thread, that's why I started it. This is not an official Dev thread, that will come later from the Dev. Of you're interested in it, head over to precog.me and read the official kernel page. :thumbup:
B3311 said:
Why?
Your setup and phone use is completely unique. Anyone else's battery life will not give you the slightest clue as to what yours might be.
Flash it and see for yourself.
Click to expand...
Click to collapse
Actually it can serve as basis. For example if that user is able to achieve 10hrs SOT, then at least I know that the kernel is capable of going that far. No matter how you force a kernel, if it has bad battery life then it can never achieve high SOTs. It actually us very helpful for one to post battery stats because I can then adjust and try to get close or even surpass the user's settings.
With all due respect, I think you are wrong in saying it won't give me the slightest clue of what the battery life of mine will be. It actually can tell a lot(if I'm doing things wrong or right). That's how users learn how to improve their devices by getting tips from others. And for that reason I'm refraining from flashing it because I have a very good set-up. I know I can go back just with flashing the stock kernel, but we all know that sometimes it doesn't really work 100%. "Something" gets broken somehow at a random chance.
Graffiti Exploit said:
Actually it can serve as basis. For example if that user is able to achieve 10hrs SOT, then at least I know that the kernel is capable of going that far. No matter how you force a kernel, if it has bad battery life then it can never achieve high SOTs. It actually us very helpful for one to post battery stats because I can then adjust and try to get close or even surpass the user's settings.
With all due respect, I think you are wrong in saying it won't give me the slightest clue of what the battery life of mine will be. It actually can tell a lot(if I'm doing things wrong or right). That's how users learn how to improve their devices by getting tips from others. And for that reason I'm refraining from flashing it because I have a very good set-up. I know I can go back just with flashing the stock kernel, but we all know that sometimes it doesn't really work 100%. "Something" gets broken somehow at a random chance.
Click to expand...
Click to collapse
Fair enough, but I still disagree.
For example - on my last cycle I got nearly 8 hours SOT and about 1 day 13 hours on one charge. What can one infer from that? I say another user can infer very little, there are way too many variables.
I use Alliance ROM. This was my first cycle with this kernel. I use only 100% black wallpapers, system, desktop, and lockscreen are all black. One homescreen. Brightness about 40% except for videos. Tasker is set so that screen off=WiFi and data off, and I use a flip cover. I've deleted over 100 apps. I've removed about 25 autostarts with SD Maid, usually only use about 750 mb at startup. I never sync anything whatsoever. Two widgets, clock/weather and music. Location is always off unless I'm navigating.
I could go on, there's more, but I hope you see my point. And that's before the vast amount of kernel tweaks I've set up with Faux Clock, which I'll probably tweak further. And I haven't even mentioned my usage....
With further cycles I'll probably squeeze more juice out of it, who knows?
If you think this will give you any idea about your battery life, fine. But I can't see how it can.
As for benchmarks - a waste of time if you're comparing with another device, only remotely useful if you're comparing performance with different setups on the same phone, IMHO.