Database Info

[WIP] [7/7] Higher framerate on Novatek panels | Kernel inside | Now for Froyo!

*** THIS IS EXPERIMENTAL ***
Thanks to toastcfh, maejrep, AssassinsLament, Estimated Prophet, Cicatrize, joeykrim, geyes30, kpkimmel and all the others from the kernel & display cap threads who have been working tirelessly, and continue to do so. This isn't over until we can find the real cap! But I think this avenue has progressed enough for it to have its own thread, if only to give hope, since we now know both Epson and Novatek panels physically can do it!
What Doesn't Work
Nothing. Now based on the official kernel with everything working.
About This Kernel
This is just a dirty hack. We haven't yet figured out where the real cap in framerate is coming from. But until we do, this is the best we have. At least it shows the display itself can handle a higher framerate and isn't physically incapable like HTC would have us believe.
Novatek panels can now get up to 46fps pretty reliably. Run fps2d a few times.
Stdev depends on system activity. If you're going to be judging stdev, DON'T measure it immediately after booting, set your CPU throttle mode to "performance", be sure to run fps2d several times, then average those stdev numbers in your head.
The changes are now immediately applied upon boot. There is no need to cycle your display anymore.
This kernel has support for overclocking/underclocking, up to 1.119 MHz. Also includes my own undervolting and overvolting for higher speeds to increase stability.
You can get dramatically tighter stdev numbers by setting your CPU governor to "performance" in SetCPU. It seems scaling really messes with the framerate. Obviously this will drain your battery faster, as your CPU will constantly be running at full speed. Set it back to "ondemand" if you don't need it.
Some people are reporting their touchscreen grounding issues have been cured.
Support to change the T2 value on the fly [thanks to maejrep]
Support for 5 simultaneous touch points! Use Dotty (in the Market) to see it in action. It'd be cool if someone would make a simple app to demonstrate what can be done with more than the mere 2 points we are used to. People have suggested new lock types, security locks, ways to rotate things, etc. [thanks to maejrep]
Some hardware fails to accept 5 points, so a fail-safe has been added to default back to 2 points for some EVOs
AssassinsLament has created an awesome app to provide an easy way for people to change their T2 timing. Check out his app here: http://forum.xda-developers.com/showthread.php?t=716659 [thanks to AssassinsLament]
Changes to display timings are now sticky until rebooted (w/ the above app, or set manually)
What's New
v5.1.1: Using the official kernel. We've lost some fps, but we've gained camera, video, 4G, etc. 46fps is now average.
v5.1.1: Expanded user-modifiable timing settings. Change at your own risk.
v5.1.1: Audio patched [thanks Geniusdog254 and KERKEDAGAIN]
Instructions
Do a nandroid backup, and flash this over pretty much any ROM. If you have problems, you can just flash a stock boot.img again. There should never be any need to wipe anything.
IMPORTANT: This will only work with the stock bcm4329.ko! Using the newer files and flashing this kernel will result in wifi being unable to start, and vice versa. To correct this, you'll need to flash over the correct wifi files.
Changing Timing Settings
You can use the following commands from adb shell to temporarily change timing on the fly. Do it while your screen is on.
Set: echo NUMBER > /sys/devices/platform/mddi_c_b9f6_5582.1/PARAMETER
Check: cat /sys/devices/platform/mddi_c_b9f6_5582.1/PARAMETER
Parameters:
t2 - 245 to 1023 [Default Eclair: 437, Froyo: 432]
t3 - 0 to 7 [Default Eclair: 1, Froyo: 0]
fte - 0 to 879 [Default Eclair: 35, Froyo: 60] (Frame Tearing Effect, change this if you have tearing problems)​
Flashable ZIP - Undervolted kernel v5.1.1 for use with 2.1 (Eclair) ROMs
Flashable ZIP - Normal-volted kernel v5.1.1 for use with 2.1 (Eclair) ROMs
Flashable ZIP - Undervolted kernel v5.1.1 for use with 2.2 (Froyo) ROMs
Flashable ZIP - Normal-volted kernel v5.1.1 for use with 2.2 (Froyo) ROMs
Note: Certain things like the camera still don't work, since our Froyo ROMs aren't official.
Flashable ZIP - WiFi-N + 4G for all custom kernels based on toastcfh's GoDmOdE kernel (compatible with v5.2+)
Flashable ZIP - WiFi-N + 4G for stock/custom kernels (compatible with v5.1 and below)
Flashable ZIP - WiFi-N + 4G for the OTA kernel (NOT compatible with the kernel in this post)
*** REMEMBER! No one here is responsible if you damage your phone! ***​
See the Evo, Droid X and Nexus One benchmarked together here: http://androidspin.com/2010/06/25/droid-x-vs-htc-evo-vs-google-nexus-one-in-benchmark-face-off
Want NenaMark1? It usually doesn't show in the Market, so get it here: http://nena.se/downloads.html

Awesome, just flashed and it is soo smooth, thank you devs.

This works great for me.

Not really related to this but my phone locks up when I change the cpu frequency with setcpu. EDIT: (How do I overvolt the phone?)
Oh and yes I do get 50hz.

Is there 2.2 support or no

ttieder said:
Is there 2.2 support or no
Click to expand...
Click to collapse
Nope I wish there was. Droid X skin + 50hz would be very cool.

ttieder said:
Is there 2.2 support or no
Click to expand...
Click to collapse
Second this question..I'm loving the EvolutionX froyo rom and would be sweet to get this baby on board with froyo.

Does this work for Epson panels too?

This works great. Touchscreen SEEMS a bit more responsive in in multitouch vis test. Grounding issue is nowhere near as bad, but still there. If you pull the same trick that everyone is doing to test grounding problem, you won't notice it at first, and very rarely will it stop sensing you finger all together, but it definitely does cut out for an instant every now and then, and I had it drop visibility where it normally does a few times for a split second.

Epic delete

As this is a development thread, please don't start posting your results like the Godmode thread. It will only clutter the thread.

Can you zip your modified files and post them too? Source would be nice

sweteg said:
As this is a development thread, please don't start posting your results like the Godmode thread. It will only clutter the thread.
Click to expand...
Click to collapse
This thread was made to avoid the clutter from the Godmode thread.

Thanks for the excellent dev work!
You might want to add to the first post that this kernel breaks 4G. it will turn on but not able to find any network.
Other than that, works as described
The device feels so much smoother!

Great work! I get 52 in fps2d, 680 in quadrant, and 51 in gears. I can't wait until this is completely fixed but I'm loving it and the undervolt and oc patches are awesome additions.
automatic brightness still does not work with this kernel though. At least not for me. Is it possible I have a different light sensor as no custom kernel has worked with auto brightness yet and other reports say it works just fine.
Also... fps2d spikes into 99 fps and at one point I watched it flash past 100

Definetly works
Installed the kernel and got over 50 fps on my nova panel i have hardware config 002 it runs much much smoother thanks devs for the hard work.

wow this kernel is insane, I mean FPS hack and overclock hack without force closes...wow is all I can say. Used the OC widget and it detects the oc clocks quickly.
Man if this kernel makes it to Froyo..Evo will be truly evil

2 vertical lines just showed up on the right side of the screen, haven't tried anything yet, let me restart and see if it goes back, then try to duplicate the issue.
Edit: It went away after I got a text, not sure if it was coincidence or not. I'm guessing this is the tearing issue that the FPS fix causes, but nothing about it was written in OP.

Mrtankjump said:
2 vertical lines just showed up on the right side of the screen, haven't tried anything yet, let me restart and see if it goes back, then try to duplicate the issue.
Edit: It went away after I got a text, not sure if it was coincidence or not. I'm guessing this is the tearing issue that the FPS fix causes, but nothing about it was written in OP.
Click to expand...
Click to collapse
My Evo has an Epson panel so I can't use this fix but, my understanding was that this method didn't cause the tearing like we saw from Toast's custom kernel on Epson displays, so your problem might be something else and not necessarily the fps cap being raised.

Thanks for the kernel, but I still have one big problem- my battery still will not charge with this kernel installed. It will only charge when I turn off the EVO and turn it back on. But as soon as I unplug and replug it in, it will look like it charges for 5 seconds, then the orange light shuts off and it stops charging. I've enclosed a dump of my dmesg. Can someone please take a look at this?

[APP] NovaFreq v1.2.1 (For Novatek Panels Only) [ Updated 07/18/10 ]

Evo Switch v1.1.1 NovaFreq v1.2.1
(APP NAME CHANGED TO SOMETHING MORE APPROPRIATE)
This is an Android application developed to change the panel frequency for Novatek Panels for the Sprint HTC EVO 4G, in conjunction with ffolkes' Novatek Kernel and netarchy's Kernel,
and any kernel that has maejrep's T2 Interface built in. Every device is different, and with that, those specified kernel's default settings may or may not be optimized with your device.
With this application, you can change those settings on the fly and find the best setting for your device.
[ Please use this application with extreme caution! I am not responsible for any damage this may cause to your device. ]
Requirements:
-HTC EVO 4G
-Android 2.1 +
-Any custom kernel that is built off of toastcfh's htc source
-Novatek Panel
-Root Access
Uses:
-Use as an fps frequency turbo switch for when you need extra fps.
-Use as a debugging tool to test frequencies.
-Use to keep frequency settings even beyond reboots. (new)
Settings:
-340 [Stock] is how HTC intended the EVO to be. This frequency works out to 61hz, but we are capped at 30hz.
-370 [Kernel Default] sets the panel frequency lower than stock, yet provides most around 55fps.
-250 [Stable High Freq] sets the panel frequency higher than stock, which usually gives around 40fps.
-400 [Stable Low Freq] - V2's setting.
-Favorite Setting - Find the most stable setting, and if it's not one of the listed, then "Set Favorite".
-Sticky Settings - Now you can reboot and not have to re-apply your settings!
Custom Setting:
You can put arbitrary settings here, but be warned this may mess up your screen, but usually you can reboot your phone
to change it back to default. Minimum Recommended: 250 Maximum Recommended: 999.
Please do not go lower then the minimum recommended of 250, nor higher than the maximum recommended 999.
To Do:
-Favorite Setting - So you can save the best setting for your device. 07/04/10 - Done!
Change Log:
-Changed app name from Evo Switch to NovaFreq (Novatek Frequency / Freek)
-Totally redid code, should be more optimized.
-Frequencies stick beyond a reboot
-Changed a few of the defaults
-Changed back sleep time to 500ms so the current displayed setting can update (v1.2.1)
Don't be alarmed if you see NovaFreq running in the background. To be able to sticky settings beyond a reboot, NovaFreq has to be run at boot up.
After that you can kill NovaFreq if you want, android will kill it when needed so don't worry.
Thanks To:
ffolkes, netarchy, maejrep, toastcfh, halfline, Cicitrize, geyes30, kpkimmel, joeykrim, and many others involved in trying to uncap the EVO.

Reserved...

Reserved again.

woah, this is badass. I will check it out soon

works great! Whats the number I set it at to equal the V2 setting of ffolkes kernel? I think its 400, is that correct?

Awesome!! This is the first time I've been able to break 30 fps! (thanks to your app!)
I'm on Froyo and the v4 had never worked, but now v4 + high frequency with your app = 40 fps.
I love you.

this may a silly question... but would this fix camera when i have it disabled as to if i switch it on the camera is now broke? or just the kernel itself breaks the camera??

Great little app, I would recommend users experiment with the values just ever so slightly for optimum performance. Using the default value of 361 netted me about 50-51 fps, whereas using 365 nets me about 54-55. Not a huge bump, but even a little bit is nice

Great application. May I suggest adding maybe a quick cycle to move between frequencies so users can go through a number of settings at once? Kinda like the Scan feature in most radios to pick the best frequencies.

timgt said:
works great! Whats the number I set it at to equal the V2 setting of ffolkes kernel? I think its 400, is that correct?
Click to expand...
Click to collapse
Yes, I think V2 was 400.
madunix said:
this may a silly question... but would this fix camera when i have it disabled as to if i switch it on the camera is now broke? or just the kernel itself breaks the camera??
Click to expand...
Click to collapse
Every custom kernel does not have a working camera, because we do not have the source code for it. When HTC finally releases the source code, then we will be able to fix the camera.

antiphon said:
Great application. May I suggest adding maybe a quick cycle to move between frequencies so users can go through a number of settings at once? Kinda like the Scan feature in most radios to pick the best frequencies.
Click to expand...
Click to collapse
This sounds like something to think about. Only problem is, while scanning, no one will know what the fps is unless they have their favorite fps app running, but then we wouldn't know what the frequency setting is since the fps app is running in focus. Might be a bit difficult to pull off without building an fps test into the app.

Well done!! Thanks

AssassinsLament said:
This sounds like something to think about. Only problem is, while scanning, no one will know what the fps is unless they have their favorite fps app running, but then we wouldn't know what the frequency setting is since the fps app is running in focus. Might be a bit difficult to pull off without building an fps test into the app.
Click to expand...
Click to collapse
Yeah I agree it would be difficult to implement without a built in fps counter. I was thinking more along the lines of something that would flag or eliminate the frequencies that are really bad (ie the ones which make your screen go haywire). So maybe the user could, for example, pick frequencies 300-310 and then the app would cycle through them 1 second at a time. From there the user would be able to run FPS2D or whatever to test out those frequencies that didn't make their screen all blotchy.
Anyways, just something to think about. Great work nonetheless!

antiphon said:
Yeah I agree it would be difficult to implement without a built in fps counter. I was thinking more along the lines of something that would flag or eliminate the frequencies that are really bad (ie the ones which make your screen go haywire). So maybe the user could, for example, pick frequencies 300-310 and then the app would cycle through them 1 second at a time. From there the user would be able to run FPS2D or whatever to test out those frequencies that didn't make their screen all blotchy.
Anyways, just something to think about. Great work nonetheless!
Click to expand...
Click to collapse
This won't work, because the interface that is used to write these settings to the panel cannot handle both full-speed graphics, and setting panel registers. I tried this when I first created the sysfs attr file, and if fps2d is running, it seems impossible to set or query the current T2 value. the display pretty much has to be idle for this to work.
Using a mutex lock might help, but I didn't look into it any further.

maejrep said:
This won't work, because the interface that is used to write these settings to the panel cannot handle both full-speed graphics, and setting panel registers. I tried this when I first created the sysfs attr file, and if fps2d is running, it seems impossible to set or query the current T2 value. the display pretty much has to be idle for this to work.
Using a mutex lock might help, but I didn't look into it any further.
Click to expand...
Click to collapse
Yep, that's why I had to put in a "sleep" command in my app so that the system could process the change and update the app on if the register changed or not. Thanks for the interface maejrep!

Where is the link to the switch app?
Sent from my PC36100 using XDA App

Yeah? Where is version 1.1.1? 1.1.0 is the only available for download? Or am I a special needs kid?

Force close only.

apollooff320 said:
Where is the link to the switch app?
Sent from my PC36100 using XDA App
Click to expand...
Click to collapse
VoXHTC said:
Yeah? Where is version 1.1.1? 1.1.0 is the only available for download? Or am I a special needs kid?
Click to expand...
Click to collapse
Sorry, maybe I forgot to change the version in the app, 1.1.1 should be the only one linked in the original post.
Sprockethead said:
Force close only.
Click to expand...
Click to collapse
Do you meet the minimal requirements on the original post? Because you have to be running a custom kernel with the T2 interface in it.

wont recognize
im not sure whats wrong, i downloaded this on my phone and it says it dosent recognize the file i downloaded

[MOD] Aroma | UV | 09-Jan | New dualcore configuration of 1.1/1.2-abov bricked kernel

Read new informations in installation guide​
Click to expand...
Click to collapse
Click to expand...
Click to collapse
IF YOU HAVE REBOOTS WHEN YOU ARE USING THE PHONE, IT'S BECAUSE YOU SET A UV TOO AGGRESSIVE, SO, YOUR CPU CAN'T BE STABLE. CHOSE A LOWER UNDERVOLTAGE
IF YOU WANT TO KNOW HOW VOLTAGE CAN REACH YOUR CPU, START STABILITY TEST (DOWNLODABLE FROM MARKET) AND USE IT FOR ABOUT 10 MINUTES, IF PHONE REBOOTS IN THAT TIME, THAT VOLTAGE IS TOO LESS FOR A STABLE CHIP
NOT ALL CHIP ARE THE SAME, IF YOU HAVE A LUCKY CHIP, YOU CAN REACH A LOWER VOLTAGE (AS MINE)
Click to expand...
Click to collapse
Click to expand...
Click to collapse
Ok Guyz, since show creates the badass governor, i'd want to use it for a better one multicore configurtion
What's the advantages of multicore sistem?
Symply, more parallelization means more performance per second, as i can say, 2 core can compile at last the double quantity of threads than 1, so, 1 core at double frequency can't compile the same number of thread, but it do it faster..
ex..
if you have 2 thread (impossibile), 1 audo and 1 video, a single core @ 2ghz compile firstly the audio, than the video.. the dualcore architecture can compile on core0 the audio and on core1 the video!! this is the powerfully of multicore system.
Ok, since i say it, i create a MOD witch TRY to use this logically.. this MOD uses the limitate frequency of the badass governor, that's why, with ondemand, all of 2 core goes at maxFreq too fast, now, with this badass, the system will not grow up the first limitate frequency until the single core uses more than that frequency.. so.. we can limitate the "ondemand" issue in 3 freq phases..
thus, the actual MOD configuration is it:
Click to expand...
Click to collapse
SCREEN ON:
- dualcore with all cores TURNED ON (hope you understand now )
- governor badass for all of 2 cores
- min freq at 384mhz for all of 2 cores
- max freq at 1.566ghz for all of 2 cores
- first phase freq at 810mhz for bla bla bla
- second phase freq at 1.188ghz for bla bla bla
- third phase freq at 1.566ghz (is the maxfreq)
SCREEN OFF:
- only core0 is TURNED ON
- governor conservative
- minfreq 192mhz
- maxfreq 540mhz​
Click to expand...
Click to collapse
Click to expand...
Click to collapse
Badass governor has the same configuration as last show's ondemand governor
So, now, with this MOD we have a pure functional dualcore with the advantages of multicore system, with limitate frequency. The stock situation is 384-810mhz, if the single core (is the same, qualcomm uses asyncronic frequency) needs more than 810mhz, it increases the max freq to 1.188ghz (the sensation original frequency), if 1.188 isn't sufficently, it increeses at 1.566
remember, qualcomm has asyncronic frequency, so, one core can be run at 1.56ghz, the other at minimum mhz..
Click to expand...
Click to collapse
Click to expand...
Click to collapse
How to use it?​
Click to expand...
Click to collapse
INSTALLATION:
-) flash a kernel witch has badass governor (i advise bricked 1.1) with NO GPU OC and what governor you want, the same for the frequency, mod autochanges those, chose what scheduler you prefer.
1) Flash MOD zip file by recovery
2) Select what MOD YOU want
3) Reboot.
#) You have to reinstall at every Rom Upgrade
# If you want only the UV, select Remove in mod choseing (it's the same as no installing it); you can use the UV mod with all kernels, and all governors
# If you come from V2.0 to 3.0, go to recovery, wipe, format system partition, after that, reinstall rom, reinstall bricked kernel (if it hasn't), reinstall other mods (like battery %), install my mod, reboot.. You will not loose your data
# ONDEMAND VERSION IS NOT FOR GOOD BATTERY CONSUPTION, IF YOU HAVE HUGE BATTERYDRAIN (IT'S WHAT YOU'LL HAVE) INSTALL THE BADASS - BOOST+BOOST VERSION.
DISINSTALLATION:
1) Boot in recovery, flash the Mod
2) Select Remove
3) Reboot​
Click to expand...
Click to collapse
Click to expand...
Click to collapse
Code:
bricked 1.2 -above
V 11.9
> Fixed Screenoff chose
> Changed DEFAULT first phase for better reactive with less power usage
> Other minor changes
###
Added Version for AOSP kernel (NoSense)
V 11.8
> Changed some settings
V 11.7
> Changed some settings for a better fluidity
V 11.6
> Added Remove chose
> Fixed an installation issue of the Aroma
V 11.5
> Added possibility to know if MOD is working, you have to give a cycle on/off of the screen, after that, if it works, you will find a new file in home of sdcard!
V 11
> Changed configuration for less battery drain
V 10.5
> Tweaked badass experience for better battery life (only with bricked kernel 1.4)
V 10.1
> Tweaked screen off conservative
V 10.0
>> BIG RELEASE
> Added possibility to chose frequency
> Added possibility to chose screen off governor
V 1.01
> Fixed freq values in Aroma installer
V 1.0
> Initial Release
​
HaVe FuN!​
Click to expand...
Click to collapse
Click to expand...
Click to collapse
FAQ in second post​

VIDEO
Click to expand...
Click to collapse
FAQ
Click to expand...
Click to collapse
Witch version could you chose?
The versions of the mod are two:
The 3.7 is the older, it's for the bricked 1.1 version AND for other kernels (don't know if all rom will boot up with my mod, some has issues)
The version 10 and above are for bricked 1.36 and 1.40
​
What setting could you use?
You can intall what you want, i create those (expecially the 10 versions) witch has the possibilities to be changed as users experience.
I advice to use default settings if you don't know how change, but, you should try to, it will not brick your phone, in fact, you could find a new setting could it be better for your use.
So, try try try
One EX:
If you want a faster phone, grow up the minfreq to 300-400mhz, grow up the first phase at 1.2ghz, put the second to 1.2 or more, and change the maxfreq as you would, OC too.
If you want a better battery experience, decrese minfreq, the first and second phase, decrese the maxfreq too, maybe, with this settings: 192 - 648 - 1020 - 1180 mhz
If you want the lowest battery experience, try with this setting: 192 - 648 - 1020 - 1020 mhz
Be careful, you could have some lags if you chose the lower setting, but, for ex, if you don't play, you can chose a setting as like, 192 - 810 - 1020 - 1180 mhz
So, you have to install, select one setting with you think it's ok, try, if you have lags in some situation, grow up some setting, an re-try til you find your best settages. I don't have your use, so, i don't know if you could have a good esperience with my setting.​
How to see if mod is working?
If you have the version 10 or above, you will find a new file in SD card, it calls IBA21ModWorking. if you don't find that, turn on the screen, wait 5 seconds, turn on, refresh the explorer app, and try to see if there is.
If you have the 3.7v you have to download and install system tuner, and see if all cores are turned on.
​
Why my phone doesen't boot after apply the mod?
Not al roms has the same quality, some doesen't have some little features; my mod uses a whyle cicle witch chages settings from screen-on use to screen-off (standby). If phone doesen't boot up, rom doesen't have the possibility to use the cycles in the script, so, script is unusefull.​
Do you need to delete or wipe the rom at every upgrade?
ABSOLUTELY NO, mod auto removes the old script and changes that with the newer one, so, YOU DON'T NEED TO DO ANYTHING, ONLY FLASH THE UPGRADE OR THE NEW SETTINGS WITCH YOU ARE CHANING!​
Click to expand...
Click to collapse
Click to expand...
Click to collapse
enjoy my work and, press a lot of thanks buttons! :d​

Very very good, the system is smooth and stable, I've only a lag after reboot, and sometimes when start synchronization...
Good work!

Hi nice script. Must the CPU values be truly stock ? Could it take something like 384-1566?
Sent from my Sensation running HyperNonsense v2.3

remus82 said:
Hi nice script. Must the CPU values be truly stock ? Could it take something like 384-1566?
Sent from my Sensation running HyperNonsense v2.3
Click to expand...
Click to collapse
Yep, i could do it.. but.. i tryed it, and.. it only increase the power consuption

framir said:
Very very good, the system is smooth and stable, I've only a lag after reboot, and sometimes when start synchronization...
Good work!
Click to expand...
Click to collapse
Lags depends on badass governor.. it's "normally".. but.. 1 lag per day is not bad, remember, with my configuration you have a better badass configuration with lower power consuption than faux kernel..

Looks awesome
Noticed bug appears after every reboot?
How to change second phase to 810-1134 and delete third phase? Or I don't need it
And soo.. It's fully better than original badass?

http://i.imgur.com/gavOn.png
Standby power consuption with data active!

D-SHEL said:
Looks awesome
Noticed bug appears after every reboot?
How to change second phase to 810-1134 and delete third phase? Or I don't need it
And soo.. It's fully better than original badass?
Click to expand...
Click to collapse
You don't need to delete it since system uses third phase (1.2 to 1.5) only in heavy games and web browsing+flash player.. in web usage cores goes to 1.2 max
Btw.. third phase is the maxfreq governor.. i can change it by script if you want to try

Take.. flash it as you would.. you can come back with stock script
For who doesen't understand.. the phase limis is a real limit.. since system needs more performance, system use the lower limit.. as i can see in my use:
For home scrolling 1 core goes to 1.2ghz, the other is at 192mhz (low power consuption)
For notice bar scrolling 1 or 2 cores goes to 810mhz, it depends on other app.. 1 core at 810mhz can calculate the notify bar
For chrome beta browsing.. in web rendering ( page loading) all of 2 cores goes to 1.2ghz, after that, core goes to 192-810 since heavy scrolls, heavy zoom, or animations in the web page (in google page cores goes to 192mhz )
For browser + flash player, 2 cores are uset at 100% at 1.56ghz since it comes down.. after that.. system uses 192-810.. sometimes only one core goes to 192-1.2
In games core i dont know why i dont play games
So.. the third pase is only a turbo boost, it's used only when system needs more and more power.. the phase is a real limit..
For example.. you have a stock 192-810.. when system needs more than 100% (810mhz) badass grow up the frequency to second phase.. it's the same for third phase.. so.. if system doesen't need more power, badass doesen't grow up the cpu frequency..
In my test i see.. ondemand tweaked + core1 off at my usage has a cpu temperature at about 55degrees C.. with badass, 47
So.. lower tenperature.. lower power consuption!
Sure.. ondemand has a faster one response.. since it grow up frequency every time, for low needs too..
you couldn't have performance plus heavy power..
The "busg" is too.. until system needs more than phase freq, it doesent grow up, so, when there's the necessity of more freq, upgrading is slower than ondemand, and it creates a little lag..
On boot, don't know why but.. only one core is turned on.. it's a system fault of reading script.. if you turn off screen, and turn on, it returns at normally (so.. needs an off-on cicle)

iba21 said:
You don't need to delete it since system uses third phase (1.2 to 1.5) only in heavy games and web browsing+flash player.. in web usage cores goes to 1.2 max
Btw.. third phase is the maxfreq governor.. i can change it by script if you want to try
Click to expand...
Click to collapse
Thanks! Maybe really don't need this
Noticed bug appears after every reboot?
"69force192-1188.zip"
As I can see, this is version without third phase?

D-SHEL said:
Thanks! Maybe really don't need this
Noticed bug appears after every reboot?
"69force192-1188.zip"
As I can see, this is version without third phase?
Click to expand...
Click to collapse
Yep that has only 2 phase.. 192-810 and 810-1.2
I edited last message with informations about your questions

Shouldn't this be in the Development section?

nope.. it's not a developement app.. those are some settages..
if you load the script, and delete it, the settages has been changed..
when you turn on display
screen ON:
- dualcore able
- governor badass for all of 2 cores
- min freq at 192mhz for all of 2 cores
- max freq at 1.566ghz for all of 2 cores
- first phase freq at 810mhz for bla bla bla
- second phase freq at 1.188ghz for bla bla bla
- third phase freq at 1.566ghz (is the maxfreq)
when you turn off
screen OFF:
- only core0 able
- governor conservative
- minfreq 192mhz
- maxfreq 540mhz
it changes that values at every turn on-off of the display.. only 1 time.. so.. if you load the script, and delete it, at display turning off, the script doesen't change the values, and, you will not have the "screen off" setting, so, you'll have the stock badass, so, more power consuption
i repeat, my script is only a numerical setting, it's not a kernel, it changes only the superficially values of the kernel, after that, is the kernel governor the real fault of frequency changing
sure.. with my script you have that 2 configuration.. and.. if you see.. conservative in standby means low power consuption (from 20 to 70ma in standby, depends on the system performance needed)

Is step 6 necessary? I don't have that function in root explorer
Sent from my Sensation running HyperNonsense v2.3

remus82 said:
Is step 6 necessary? I don't have that function in root explorer
Sent from my Sensation running HyperNonsense v2.3
Click to expand...
Click to collapse
it's not obbligatory.. but.. you have it.. long press.. OPEN WITH -> LINUX script (it's the third link from top)
http://i.imgur.com/BbeS0.png
Or you can reboot 2 times.. in first run script changes the name of the file witch turn on the second core.. the mpdecision.. so.. or you do a first manually run.. or you do a first reboot.. script changes the name, but the mpdecision script is already loaded.. second reboot use only my script

Thanks! Will report back after a day of use!
Sent from my Sensation running HyperNonsense v2.3

Edit: solved.
Sent from my Sensation running HyperNonsense v2.3

Iba I notice that media scanner takes extremely long to complete after applying the script. Is it possible to use badass or on demand as screen off governor?
Sent from my Sensation running HyperNonsense v2.3

Sure.. but you loose the advantages of low power consuption..
What is media scanner? How many mhz does it needs?

[Post your benchmark] Simple Antutu video (N7 2013 vs N5 vs Moto G vs Xperia S)

Hello guys,
Here is a simple video to show benchmark scores in Antutu between four devices:
- Nexus 7 2013: CleanROM 3.5 (Kitkat 4.4.4), ElementalX kernel
- Moto G Gpe: Stock rooted 4.4.2 Google play edition, stock kernel
- Xperia S: Stock 4.1.2, stock kernel
- Nexus 5: Stock 4.4.3, stock kernel
Link: http://www.youtube.com/watch?v=e8Q7PORuxi0
Thanks for watching !

Cool video, well made!!
SM-G900F
Alliance ROM
KT Kernel OC to 2.89Ghz
Antutu: 42672 (#2 overall)

n01z said:
Cool video, well made!!
SM-G900F
Alliance ROM
KT Kernel OC to 2.89Ghz
Antutu: 42672 (#2 overall)
Click to expand...
Click to collapse
Well thank you !
Nice score :good:

lulu14150 said:
Well thank you !
Nice score :good:
Click to expand...
Click to collapse
Thanks, right back atcha!! I think I'm the first S5 to break 42K, if I combined all of my best individual catagory scores, I would break 44K.. Got some work to do, got a bunch of tweaks lined up that I haven't tested yet, I think I'm gonna wait for some others to catch up.. or until I get bored again!!
---------- Post added at 07:11 PM ---------- Previous post was at 07:04 PM ----------
The settings that I use on my overclock (OC) are borderline stupid dangerous. And the settings I use to bench are straight up dangerous. For this reason, I will not be sharing them.
HOWEVER, You can find my non-overclocked faux settings here:
https://plus.google.com/app/basic/stream/z13tu3aq0krxchnnv234jdkbrnzrvlpcq
These settings scored 38256, no OC, AEL Kernel controlled by fauxclock.
Note: What we are doing here can result in you frying your snartphone- not worth it just to get a ranking.
For me, I just can't help but to experiment and push my device to it's limits. HOWEVER, my benchmark settings differ from my daily use settings, slightly (for stability) because I love my S5 and would possibly cry if we were to have to part!!
I'm going to share some of my best benching tips (no "tricks" just my observations)
1. Make sure that your Kernel control app has a boot delay of at least 20 seconds. (Mine is at 30 seconds, remember-patience is a vitrue!!)
2. Completely dim your screen, put your phone into airplane mode, disable any motion gestures.
3. Turn phone off for 5 min.
4. Boot your phone, but don't touch it, just let the lockscreen time out, wait 2 minutes, unlock the phone and let it sit for another minute.
5. Do not flip screens or pull the notification bar (we want to NOT trigger any background processes, so turn off any motion gestures- because they are using the CPU while monitoring your sensors and be sure to leave your default homescreen free of widgets).
6. Open Antutu, wait about 30 sec..
7. Click Test Again then uncheck Optimize Memory (your ROM should already have optimized memory and I assure you thay Antutu doesn't know your device better than the OEM or Custom ROM devs) Now, wait about a minute and a half..
8. Click TEST then don't touch/move your phone until it's done..
More Notes:
Make sure that you only ever change one parameter at a time and be sure to run a CPU Monitor when you do, to see the affects of each change..
Reboot inbetween every bench and turn your screen brightness to MIN before you reboot to bench..
Also, I recommend to try running a CPU Monitor once or twice while you run a bench in antutu so you can get an idea of how long it takes for antutu's CPU use to chill out, (same can be done after boot to find out what your device likes, since I've never had any samsung bloat- as I rooted and flashed custom ROMS and Kernels the first day I got my S5- I literally have no experience with Stock)..
Also, running a CPU Monitor when you are getting low benches will help you see if it's being thermal throttled during a bench (throttling has caused me to score as low as 32K) and it will give you the best idea of what may be going wrong and where.
I actually know very little about Kernels, I am just patient, never give up and trial and error sweetspots until I get what I want..
Extra Note: Sweetspots vary from each individual device, so ya just gotta keep at it!!
P.S. These tips should work/help with any benching app.
SM-G900F
Alliance ROM
KT Kernel OC to 2.89Ghz
Antutu: 42672 (#2 overall)

n01z said:
Thanks, right back atcha!! I think I'm the first S5 to break 42K, if I combined all of my best individual catagory scores, I would break 44K.. Got some work to do, got a bunch of tweaks lined up that I haven't tested yet, I think I'm gonna wait for some others to catch up.. or until I get bored again!!
---------- Post added at 07:11 PM ---------- Previous post was at 07:04 PM ----------
The settings that I use on my overclock (OC) are borderline stupid dangerous. And the settings I use to bench are straight up dangerous. For this reason, I will not be sharing them.
HOWEVER, You can find my non-overclocked faux settings here:
https://plus.google.com/app/basic/stream/z13tu3aq0krxchnnv234jdkbrnzrvlpcq
These settings scored 38256, no OC, AEL Kernel controlled by fauxclock.
Note: What we are doing here can result in you frying your snartphone- not worth it just to get a ranking.
For me, I just can't help but to experiment and push my device to it's limits. HOWEVER, my benchmark settings differ from my daily use settings, slightly (for stability) because I love my S5 and would possibly cry if we were to have to part!!
I'm going to share some of my best benching tips (no "tricks" just my observations)
1. Make sure that your Kernel control app has a boot delay of at least 20 seconds. (Mine is at 30 seconds, remember-patience is a vitrue!!)
2. Completely dim your screen, put your phone into airplane mode, disable any motion gestures.
3. Turn phone off for 5 min.
4. Boot your phone, but don't touch it, just let the lockscreen time out, wait 2 minutes, unlock the phone and let it sit for another minute.
5. Do not flip screens or pull the notification bar (we want to NOT trigger any background processes, so turn off any motion gestures- because they are using the CPU while monitoring your sensors and be sure to leave your default homescreen free of widgets).
6. Open Antutu, wait about 30 sec..
7. Click Test Again then uncheck Optimize Memory (your ROM should already have optimized memory and I assure you thay Antutu doesn't know your device better than the OEM or Custom ROM devs) Now, wait about a minute and a half..
8. Click TEST then don't touch/move your phone until it's done..
More Notes:
Make sure that you only ever change one parameter at a time and be sure to run a CPU Monitor when you do, to see the affects of each change..
Reboot inbetween every bench and turn your screen brightness to MIN before you reboot to bench..
Also, I recommend to try running a CPU Monitor once or twice while you run a bench in antutu so you can get an idea of how long it takes for antutu's CPU use to chill out, (same can be done after boot to find out what your device likes, since I've never had any samsung bloat- as I rooted and flashed custom ROMS and Kernels the first day I got my S5- I literally have no experience with Stock)..
Also, running a CPU Monitor when you are getting low benches will help you see if it's being thermal throttled during a bench (throttling has caused me to score as low as 32K) and it will give you the best idea of what may be going wrong and where.
I actually know very little about Kernels, I am just patient, never give up and trial and error sweetspots until I get what I want..
Extra Note: Sweetspots vary from each individual device, so ya just gotta keep at it!!
P.S. These tips should work/help with any benching app.
SM-G900F
Alliance ROM
KT Kernel OC to 2.89Ghz
Antutu: 42672 (#2 overall)
Click to expand...
Click to collapse
Thank you for sharing your configuration !
My Nexus 7 2013 configuration:
- ElementalX Kernel (1,7GHz CPU, 490MHz GPU and a small undervolt --> Perferct for daily use)
- Rom CleanROM (just with Forced GPU 2D rendering in Developer Options)
Score between 26k and 27k

FEB-9-15 [] Post_boot tweaks - An *interactive* boost

*FOR ANY DEVICE USING QUALCOMM CHIP***
Drop this file in your /system/etc folder, replacing the stock one. Cleaned up, and modified to work with a stock kernel or lean kernel.
Please open up the file, and read some of the notes I added, so you know what this does and how to change it if you so desire. It is very basic and straightforward.
Also, please remember one thing, if you want to go to the stock kernel, simply change any words in the scripts from "interactiveX" to "interactive".
Recommended lean kernel 3.16 with these scripts.
Designed to work with ALL Galaxy Note 3's running touch wiz 4.4.2 ROMs and running lean kernel or stock kernel. Other kernels are not guaranteed due to the voltages applied.
This will make a huge difference in performance and battery. Been using this (developed by me) for a long time now.
Summary:
*I/O optimization
*tuned interactive parameters
*fixed sysfs permissions for cpu's (root now has full control of sysfs tuneables)
*GPU sysfs identified for user to customize
*screen off frequency set to 1.2 GHz (can be changed by user)
*custom voltage set for bin 2 device. (Read instructions within file to adjust for others bins accordingly)
*custom mpdecision tuning and other relevant instruction regarding mpdecision binary
*general CPU tweaks for better performance, responsiveness and battery life (yes all 3 can be achieved, and are)
Again, open the file with a file explorer (or notepad+ in windows). Read it, understand it, adjust what you want, or leave it be. These settings are very much dialed in for the interactiveX governor of LK. It won't get much better. I'd say take a look at the voltage portion and mpdecision for now (found at the end).
Once more, drop file into the /system/etc directory... Replace the existing one, and reboot.
FILE DOWNLOAD
Updated today... many useful changes, and fixes (set up for stock kernel for now)
https://www.dropbox.com/s/tjlygujptls3lks/init.qcom.post_boot.sh?dl=0

Are power save settings included in your mod?
I made some test with Antutu on stock CPU speed..
1.) Stock kernel
- original file = > 43911
- your file "interactiveX" => 43545
- your file "interactive" => 41690
2.) Lean kernel
- original file = > 44047
- your file "interactiveX" => 42484
- your file "interactive" => 41804

tkari4 said:
Are power save settings included in your mod?
I made some test with Antutu on stock CPU speed..
1.) Stock kernel
- original file = > 43911
- your file "interactiveX" => 43545
- your file "interactive" => 41690
2.) Lean kernel
- original file = > 44047
- your file "interactiveX" => 42484
- your file "interactive" => 41804
Click to expand...
Click to collapse
I wouldn't be too concerned with benchmark scores.
Performance ≠ a higher score in antutu. But suit yourself.

Just giving you all a heads up.... I think there may have been a few errors in the stock init.qcom.post_boot file (they are human too, remember).
I am testing a few small changes to certain lines in the "factory" script to make sure. A brief explanation of it is that there are certain lines I removed from the one you all are using right now, and I did so because the file paths being called out did not exist!
However, I was going over this again tonight, getting ready for the official LP release and decided to review this odd discrepancy one more time. After doing some digging, I found that the file that some of those values were intended to be written to, were actually somewhere else!
Doh! - Sammy/QC. It is very possible that this location changed from a previous design or fs structure and somebody simply forgot to modify the script. Very interesting indeed.
Anyways, after I run this for a couple of days and do a little bit more investigating, I'll post the updated script for all of you.

Updated file and link is now available... I highly recommend anybody (devs included) use this in your current setup. Lot of fixes to the stock file that were critical to stability, power saving, and performance.

red_can_soda said:
Updated file and link is now available... I highly recommend anybody (devs included) use this in your current setup. Lot of fixes to the stock file that were critical to stability, power saving, and performance.
Click to expand...
Click to collapse
I wonder if this would work on lollipop leak. I'm running ayra mod v2 and could test this out.

fookea said:
I wonder if this would work on lollipop leak. I'm running ayra mod v2 and could test this out.
Click to expand...
Click to collapse
My guess is that it would work on 5.0... I would need to see a 5.0 post_boot file, though, to compare and look for any major changes.

red_can_soda said:
Updated file and link is now available... I highly recommend anybody (devs included) use this in your current setup. Lot of fixes to the stock file that were critical to stability, power saving, and performance.
Click to expand...
Click to collapse
I've installed your script (on stock 4.4.2 kernel) but didn't notice any performance, stabilty nor battery improvements.. It's more placebo effect than a real improvement.

ranwej said:
I've installed your script (on stock 4.4.2 kernel) but didn't notice any performance, stabilty nor battery improvements.. It's more placebo effect than a real improvement.
Click to expand...
Click to collapse
You are probably right, evidenced in the reports by everyone else that has used this saying the exact opposite. Lol
For me, it's noticeable. I have tested it pretty thoroughly and the script isn't just some cute ideas thrown together. From the lpm fixes (you can thank Qualcomm for that mistake), to some of the finer things like the deadline IO tuning (geared specifically for flash memory access), to some of the virtual machine parameters I've adjusted... There is logic and reasoning behind every single line of code I put together for this.
To be frank, I don't mind opinions. Hell I love them. But you need to know what you are talking about before you make a snide comment... Not for my sake, but for somebody who might not use it simply because they saw your remarks.
Many things appear to be placebo when it comes to this kind of stuff... But go start unzipping a large compressed file before and after using this script, or move a massive amount of data from your internal storage while playing music, browsing the web, and downloading an app at the same time.... I'd doubt you come in here talking about placebo.
You have yourself a wonderfully intelligent day.

Here you go, this is the post boot file. I'm running ayra mod v3 on noa5 leak with nl8 bootloader. The file appears to be modified on 11 Feb 2015