ATP Tweak is an app, which provides and loads kernel modules and sets I/O schedulers (also on device startup), which are not provided by the shipped kernel from ASUS.
The app can also set the overclocked CPU modes via the /system/etc/cpu*.sh scripts.
The modules, which you see on the right are the ones from this thread, where I try to explain their purpose: http://forum.xda-developers.com/showthread.php?t=1455382
This version only supports loading and setting of the I/O scheduler on startup.
This app requires ICS and root!
Overclocking can damage your device! Use at your own risk!
Changelog:
V4:
- added I/O scheduler selection spinner
V3:
- added CPU overclocking modes
- added module list, to activate additional modules
V2:
- improved SuperUser handling a bit
V1:
- load cfq-iosched.ko kernel module and set cfq scheduler
Wicked, thanks mate, Seems to be working properly now that an app is loading it!
Just doing a couple of tests. So in theory now if we reboot the scheduler will run automatically?
geinome said:
Wicked, thanks mate, Seems to be working properly now that an app is loading it!
Just doing a couple of tests. So in theory now if we reboot the scheduler will run automatically?
Click to expand...
Click to collapse
yep, should load automatically on reboot.
I'm having some difficulty getting this app to automatically insert the module and set the IO scheduler. Does the app look for the module in a specific folder? I have it in both the usual place (system/libs/modules) and the one suggested in the other thread (data/local/modules).
If I manually insert the module, CFQ shows up as available, but does not activate. If I manually set CFQ as the IO scheduler, it shows as the active scheduler. The app appears not to be doing anything.
Also, SuperUser is not throwing up a request for su access when running the app.
ilys said:
Also, SuperUser is not throwing up a request for su access when running the app.
Click to expand...
Click to collapse
That might be the problem. Force close "ATP Tweak" in settings/apps and retry. The app includes the modules, so no need to install them anywhere.
You should definetly see the SuperUser app requesting permissions.
saturn_de said:
That might be the problem. Force close "ATP Tweak" in settings/apps and retry. The app includes the modules, so no need to install them anywhere.
You should definetly see the SuperUser app requesting permissions.
Click to expand...
Click to collapse
Force-closed, even tried rebooting, and SuperUser is still not requesting permissions. I can verify SuperUser is working with other apps that require root.
Will try re-installing ATP Tweak.
EDIT: Re-installed, rebooted, still no permissions request.
ilys said:
Force-closed, even tried rebooting, and SuperUser is still not requesting permissions. I can verify SuperUser is working with other apps that require root.
Will try re-installing ATP Tweak.
EDIT: Re-installed, rebooted, still no permissions request.
Click to expand...
Click to collapse
Uploaded version 2... please retry with that one.
saturn_de said:
Uploaded version 2... please retry with that one.
Click to expand...
Click to collapse
It is showing the permissions request with v2 and CFQ has become the active IO scheduler. Thank you.
So just install this, launch app, n we good to go n never have to mess with it again? Is it best to install version you have now or just wait for your newer versions with more options? Will this affect OTA updates since adding or changing something in system files?
demandarin said:
So just install this, launch app, n we good to go n never have to mess with it again?
Click to expand...
Click to collapse
If you target is the better I/O scheduler, just install, launch app...
demandarin said:
Is it best to install version you have now or just wait for your newer versions with more options?
Click to expand...
Click to collapse
I will add more kernel modules and I/O scheduler options.
You rock man, thanks!
demandarin said:
Will this affect OTA updates since adding or changing something in system files?
Click to expand...
Click to collapse
Does not mess with system files at all..
I wonder if this will help with random lockup reboot issues? Just a thought.
What does this tweak in general improve? Download speeds from browser into prime itself? Bootup times? Transferring files within prime itself? Just want to be clear what's exactly being improved. As I'm sure others would love to know also.
Or is this more for people transferring files from prime to PC or vice versa?
I was assuming, possibly/probably incorrectly, that this was to do with how the CPU handled tasks in general.
Here is a PDF write up on the difference between CFS (Completely Fair Scheduler) and BFS (Brain **** Scheduler).
http://db.tt/ZYCGCt6V
Both have positives and negatives, and you'll find people on both sides of the fence, kind of a Ford vs. Chevy deal. Although, in general, I'd say CFS is probably more stable, and shows better results for a majority of people.
I spose I should add that this isn't necessarily specific to this thread. But it does give some good info on CFQ, what it is, and how it handles system requests.
Sent from my Transformer Prime TF201 using Tapatalk
Cadillac Johnson said:
Here is a PDF write up on the difference between CFS (Completely Fair Scheduler) and BFS (Brain **** Scheduler).
http://db.tt/ZYCGCt6V
Both have positives and negatives, and you'll find people on both sides of the fence, kind of a Ford vs. Chevy deal. Although, in general, I'd say CFS is probably more stable, and shows better results for a majority of people.
I spose I should add that this isn't necessarily specific to this thread. But it does give some good info on CFQ, what it is, and how it handles system requests.
Sent from my Transformer Prime TF201 using Tapatalk
Click to expand...
Click to collapse
Very interesting read. Just read it all.
Trying it now. Noop io sucks. Cfq ftw!!
Cadillac Johnson said:
Here is a PDF write up on the difference between CFS (Completely Fair Scheduler) and BFS (Brain **** Scheduler).
http://db.tt/ZYCGCt6V
Both have positives and negatives, and you'll find people on both sides of the fence, kind of a Ford vs. Chevy deal. Although, in general, I'd say CFS is probably more stable, and shows better results for a majority of people.
I spose I should add that this isn't necessarily specific to this thread. But it does give some good info on CFQ, what it is, and how it handles system requests.
Sent from my Transformer Prime TF201 using Tapatalk
Click to expand...
Click to collapse
Except CFQ is not CFS. CFS and BFS are process schedulers - they control how running applications get allocated to CPU cores. CFQ is an I/O scheduler - it controls how and when the system reads/writes from storage.
http://en.wikipedia.org/wiki/CFQ
noop should be the best unless the flash memory in the Prime is really, really crappy - but maybe it really is that bad. Has anyone run before/after benchmarks with CFQ?
Noxious Ninja said:
Except CFQ is not CFS. CFS and BFS are process schedulers - they control how running applications get allocated to CPU cores. CFQ is an I/O scheduler - it controls how and when the system reads/writes from storage.
http://en.wikipedia.org/wiki/CFQ
noop should be the best unless the flash memory in the Prime is really, really crappy - but maybe it really is that bad. Has anyone run before/after benchmarks with CFQ?
Click to expand...
Click to collapse
Op said Androbench not reliable. Use either CF-Bench or Bonnie++. Benchmarks seem higher in some area bit lower in others. Others said scores overall seem the same but seems snappier transferring files. I dunno. Lol
That's why I was trying to ask OP or someone who knew what exactly this would benefit. Aside from read/write speeds. What things are governed by that?
But yeah, from what I've read noop seems to be best I/O scheduler as its for SSD devices or devices that uses flash memory or memory with no mechanical parts. Since non of the latter, extra processes or something wouldn't be needed therefore noop I/O is best suited for our device. I would think..lol
Related
Using this kernel is at your own risk. Before flashing make backups!!!
Hey guys, remember this thread?
http://forum.xda-developers.com/showthread.php?t=952365
well, we can not do it for Ecliar, but OK for Froyo, any body want to taste this?http://sugwacee.dyndns.org/
there are two files for this:
NC_stock_1.2_update_sqlite.so.zip is the updated sqlite.so(no sync version), performance boost, cos sync write in high frequency is not needed. with this version sqlite will only sync write when needed. if no system crash, no data lose,just take it easy.
NC_stock_1.2_restore_sqlite.so.zip is the stock sqlite.so(for recovery, just in case).
Use CWM to flash. enjoy it.
reserve for feedbacks
Have you noticed a performance increase?
likearaptor said:
Have you noticed a performance increase?
Click to expand...
Click to collapse
unlike OC kernel, the perf increase is showed at sqlite sync period, not every moment.
1. performance increase when app install/remove, and when large num of data need to sync into database.
2. save disk(TF) i/o irq, good for system i/o wait delay, and good for TF card too(less write)--only when it is sqlite related.
3. quadrant scores, if you like it.
Flashed the zip, ran quadrant and I got 1264. I wasn't sure what it was before, but it seems fast with apps using the database. Thanks for this find
installed.
Linpack score has dropped from 12's to 10's
Quadrant has bumped up a little from 1630'ish - 1680'ish
Hmm, interesting...
I'm not totally convinced that this device is I/O bound.... But for argument's sake, how does this differ from say...changing the caching strategy in the kernel (for overall I/O performance, not just sqlite?)
Also, this only affects writes correct?
HamNCheese said:
Hmm, interesting...
I'm not totally convinced that this device is I/O bound.... But for argument's sake, how does this differ from say...changing the caching strategy in the kernel (for overall I/O performance, not just sqlite?)
Also, this only affects writes correct?
Click to expand...
Click to collapse
emm... maybe you are right, cache does turn high freq i/o cmd down(sync to ram immediately,and sync to disk when idle).
as i know, only effects write action.
actually i got this .so from moto milestone bbs(same kernel version and system env), then tested and modify it into CWM zip to fit NC.
I'm on the Community Rom V1.3 (flashed kernel in the v1.2 patch) and am having problems with my music quality. Any time I run anything or touch the screen while listening to music (via Pandora or my music app) the quality is poor. The music has a "popping", "skipping" or "staticy" sound.
I've attempted to flash older kernels as some have said it's an issue with Voodoo Kernel Version 8. Older kernels did not help.
I also, via Voltage Controll App, attempted to switch to CFQ and BFQ scheduler. That did not help either.
Has anyone experienced this issue? Perhaps there are other resolutions?
skindler24 said:
I'm on the Community Rom V1.3 (flashed kernel in the v1.2 patch) and am having problems with my music quality. Any time I run anything or touch the screen while listening to music (via Pandora or my music app) the quality is poor. The music has a "popping", "skipping" or "staticy" sound.
I've attempted to flash older kernels as some have said it's an issue with Voodoo Kernel Version 8. Older kernels did not help.
I also, via Voltage Controll App, attempted to switch to CFQ and BFQ scheduler. That did not help either.
Has anyone experienced this issue? Perhaps there are other resolutions?
Click to expand...
Click to collapse
The typical hack, work around for latency problems is basically not slowing down whenever possible. eg, disable the minimum 100MHz state, if that is not enough then disable 200MHz also.
beray5 said:
The typical hack, work around for latency problems is basically not slowing down whenever possible. eg, disable the minimum 100MHz state, if that is not enough then disable 200MHz also.
Click to expand...
Click to collapse
Should I slow it down even more if at 200MHz I still experience the same problems?
skindler24 said:
Should I slow it down even more if at 200MHz I still experience the same problems?
Click to expand...
Click to collapse
It is up to you to decide how much is enough.
beray5 said:
It is up to you to decide how much is enough.
Click to expand...
Click to collapse
I'm sorry to say that didn't work. And I slowed it down to 1000MHz.
skindler24 said:
I'm sorry to say that didn't work. And I slowed it down to 1000MHz.
Click to expand...
Click to collapse
Are you disabling your overclock back to default 1000MHz? Or are you speeding up the phone by removing 100MHz and/or 200MHz states?
beray5 said:
Are you disabling your overclock back to default 1000MHz? Or are you speeding up the phone by removing 100MHz and/or 200MHz states?
Click to expand...
Click to collapse
The latter. I am pulling up the "States" tab at the bottom of the home screen and unchecking (or removing) the 100MHz, 200MHz, etc. boxes.
skindler24 said:
The latter. I am pulling up the "States" tab at the bottom of the home screen and unchecking (or removing) the 100MHz, 200MHz, etc. boxes.
Click to expand...
Click to collapse
Ok there's one more thing you can try improving latency response, but make sure you are already had other latency dependent parameters corrected (CFQ only, no more or less than 5 freq states being used, removing bottom 100MHz).
Install this -> https://market.android.com/details?id=com.sdincrease.it
Use it to change your cache setting to find the setting most optimal in reducing crackles and pops. Note - too little does nothing, too BIG or too HUGE can exaggerate latency problems.
beray5 said:
Ok there's one more thing you can try improving latency response, but make sure you are already had other latency dependent parameters corrected (CFQ only, no more or less than 5 freq states being used, removing bottom 100MHz).
Install this -> https://market.android.com/details?id=com.sdincrease.it
Use it to change your cache setting to find the setting most optimal in reducing crackles and pops. Note - too little does nothing, too BIG or too HUGE can exaggerate latency problems.
Click to expand...
Click to collapse
Still popping, sigh.
Sent from my SCH-I500 using XDA Premium App
skindler24 said:
Still popping, sigh.
Sent from my SCH-I500 using XDA Premium App
Click to expand...
Click to collapse
I was on comm rom playing with its latency until version 1.2 then I switched to Evil Fascinate 3.51 because it was easier on my eyes. Why don't you try it? There's zero skips, crackles, pops in anything I play.
What scheduler are you running? Have you tried changing it?
Have you undervolted at all? Try going back to stock voltages.
As a last resort, try flashing a different kernel into your setup (such as PB&J or JTs) and see if the problem persists.
beray5 said:
I was on comm rom playing with its latency until version 1.2 then I switched to Evil Fascinate 3.51 because it was easier on my eyes. Why don't you try it? There's zero skips, crackles, pops in anything I play.
Click to expand...
Click to collapse
I'm a fan of evil fascinate. Thanks, beray5.
Sent from my SCH-I500 using XDA Premium App
I was having the same issue, but now it's gone. Make sure you have the latest voodoo control app from the market.
There's another thread on this popping issue where the op had been in touch with supercurio and shared a partial fix, I suspect it's been updated since I last looked.
Sent from my SCH-I500 using XDA Premium App
That was my thread. It's called sound popping in the General thread. I just reverted to DL09 and OTA to EA28 and to ED03 to try to fix signal issues by esn swapping. On this stock rom, I still here popping but its not nearly as prevalent and doesn't occur when I touch the screen. Supercurio mentioned that the sound drivers were poorly implemented on Froyo kernels so it might take awhile for this to be addressed.
Sent from my SCH-I500 using XDA Premium App
I had this problem, and after tons of testing i have found that the ONLY way to fix it is by doing the full eb01 odin. Odining to eb01 gets rid of this completely. DL09 may work (i havent checked), but after doing the update.zip to ed01 it always comes back.
bmeek said:
I was having the same issue, but now it's gone. Make sure you have the latest voodoo control app from the market.
There's another thread on this popping issue where the op had been in touch with supercurio and shared a partial fix, I suspect it's been updated since I last looked.
Sent from my SCH-I500 using XDA Premium App
Click to expand...
Click to collapse
I have what I believe to be the latest voodoo control app but don't know how to use it to help my issue. I still hear the popping on Evil Fasication v3.5.1.
Any help here is greatly appreciated.
weird popping sickness, once contracted, spans kernels and roms
skindler24 said:
I have what I believe to be the latest voodoo control app but don't know how to use it to help my issue. I still hear the popping on Evil Fasication v3.5.1.
Any help here is greatly appreciated.
Click to expand...
Click to collapse
skindler24, did you recently experiment with CM7 from kaos or jt? I'm not sure what's going on here, but thefunkbot's post that a "full odin" fixed it for him is similar to my experience (although I reverted to di01 to initiate an OTA update, not a full EB01 as shared by thefunkbot). Check out zacisblack's thread in Fascinate General for more details.
Perhaps some developer here with more knowledge than I could consider why an issue like this that seems to affect multiple kernels and roms might be (inadvertently) eradicated by a repartitioning / full odin reflash.
good luck with it...
bmeek said:
skindler24, did you recently experiment with CM7 from kaos or jt? I'm not sure what's going on here, but thefunkbot's post that a "full odin" fixed it for him is similar to my experience (although I reverted to di01 to initiate an OTA update, not a full EB01 as shared by thefunkbot). Check out zacisblack's thread in Fascinate General for more details.
Perhaps some developer here with more knowledge than I could consider why an issue like this that seems to affect multiple kernels and roms might be (inadvertently) eradicated by a repartitioning / full odin reflash.
good luck with it...
Click to expand...
Click to collapse
Odin back to stock worked. Problem solved!
Sent from my SCH-I500 using XDA Premium App
When using a kernel that has cpu1 enabled, the cpu1 scaling_governor will be reset to userspace at random intervals. I haven't been able to figure out a pattern for it, but has anyone else noticed this behavior, or have an idea of how to prevent it?
On a related note, the permissions for cpu1/cpufreq on startup are really weird too (440 for scaling_max_freq and scaling_governor), but I can manually fix those after a reboot...
hboot: 1.40
root: revolutionary
ROM: stock 2.08.651.2
kernel: freeza 4a or bcblend 0.03
If any other info would help please let me know. Thanks in advance.
penguinicus said:
When using a kernel that has cpu1 enabled, the cpu1 scaling_governor will be reset to userspace at random intervals. I haven't been able to figure out a pattern for it, but has anyone else noticed this behavior, or have an idea of how to prevent it?
On a related note, the permissions for cpu1/cpufreq on startup are really weird too (440 for scaling_max_freq and scaling_governor), but I can manually fix those after a reboot...
hboot: 1.40
root: revolutionary
ROM: stock 2.08.651.2
kernel: freeza 4a or bcblend 0.03
If any other info would help please let me know. Thanks in advance.
Click to expand...
Click to collapse
very interesting issue...
cpu1 scaling governor reset to userspace at random intervals .. the first guess i'd have would be some 3rd party android app following a predefined profile. of course, this app should need root access to echo out to the sysfs file and if you have superuser installed the app would have had to prompt at least once for permission.
if you don't have any 3rd party app installed echo'n values then the only other common tool i've seen is the settings in CM7, but you stated you're on a stock ROM.
if you're on a rooted stock ROM with custom kernel, sometimes the devs setup support for startup scripts and might be making some type of tweak in a startup script which could also be set to loop and apply some predefined setting.
my guess would be the perms on sysfs files are setup in the kernel config files when its compiled. would be something the kernel dev should be able to change easily. 440 could be intentional as they're trying to limit access .. don't want any random app/binary messing around with the settings
hope some of those thoughts are helpful!
joeykrim said:
very interesting issue...
cpu1 scaling governor reset to userspace at random intervals .. the first guess i'd have would be some 3rd party android app following a predefined profile. of course, this app should need root access to echo out to the sysfs file and if you have superuser installed the app would have had to prompt at least once for permission.
Click to expand...
Click to collapse
Here are the programs with superuser permission:
BusyBox installer
ConnectBot
Kernel Manager Lite
Quick Boot
ROM Manager
Superuser
Titanium Backup
All of these are pretty common, and I doubt any of them would be poking around the governor, but that's why I'm asking after all.
joeykrim said:
if you're on a rooted stock ROM with custom kernel, sometimes the devs setup support for startup scripts and might be making some type of tweak in a startup script which could also be set to loop and apply some predefined setting.
Click to expand...
Click to collapse
Do you know how I could check this? I don't think the kernel has init.d support.
joeykrim said:
my guess would be the perms on sysfs files are setup in the kernel config files when its compiled. would be something the kernel dev should be able to change easily. 440 could be intentional as they're trying to limit access .. don't want any random app/binary messing around with the settings
Click to expand...
Click to collapse
Yeah, definitely makes sense. It was just odd to me that the CPUs were mismatched (cpu0 had 640 permissions for all of those).
joeykrim said:
hope some of those thoughts are helpful!
Click to expand...
Click to collapse
Definitely, thank you.
I've worked around it for now by modifying Virtuous OC to set the perms to 640 before doing its thing, then resetting them to 440 when done. I've also written a daemon that will report when it's changed to userspace in hopes that I can track down some sort of pattern.
penguinicus said:
Here are the programs with superuser permission:
BusyBox installer
ConnectBot
Kernel Manager Lite
Quick Boot
ROM Manager
Superuser
Titanium Backup
All of these are pretty common, and I doubt any of them would be poking around the governor, but that's why I'm asking after all.
Do you know how I could check this? I don't think the kernel has init.d support.
Yeah, definitely makes sense. It was just odd to me that the CPUs were mismatched (cpu0 had 640 permissions for all of those).
Definitely, thank you.
I've worked around it for now by modifying Virtuous OC to set the perms to 640 before doing its thing, then resetting them to 440 when done. I've also written a daemon that will report when it's changed to userspace in hopes that I can track down some sort of pattern.
Click to expand...
Click to collapse
Okay, well given some of your responses I'm going to assume you're a bit more of a technical/advanced user.
All the Android applications you posted with superuser access are very common as you stated. An easy way to check might be, rename /system/bin/su to something else so no scripts or apps can gain superuser access and see if your sysfs cpu1 files are still changed?
If I were you, I'd want to get to the bottom of the perms on the cpu1 sysfs files. Especially since you're saying cpu0 and cpu1 have diff perms set on the sysfs files. Since Linux kernel code is GPL, the kernel dev should have a public link to their kernel files. I'd go look thru their source to find where they're implementing cpu1 sysfs files and check the perms they're setting up sysfs files with. I want to say, the cpu1 sysfs files are something added/modified by the kernel devs so there is definitely potential somebody missed a proper setting.
I'd rather take this approach than have to patch over some poor coding by using another app to adjust the perms on the file.
Can you post the log produced by your daemon on the changes to the sysfs cpu1 file?
Hope that helps!
EDIT: noticed I missed a question, how to check for init.d type support in the kernel. Generally, they're going to have to make a change to the init.rc file in order to add in anything to the start up sequence. I'd grab the stock init.rc file and diff against the currently loaded init.rc file in the kernel being used.
As an FYI, init.rc is stored in the ramdisk part of the boot.img.
Whew, ok. After a lot of observation, I noticed that the reset to userspace corresponded with me putting the device to sleep.
I then tried disconnecting my device from USB and the daemon started crashing, being unable to read the sysfs files for cpu1, which never happened while connected over USB. I switched adb to TCP mode and found that the files were indeed gone, because cpu1 was offline. Apparently being connected over USB kept cpu1 online more often than it would be otherwise. This was causing some weird permission resets too, as the permissions would reset when the core was brought offline and back online.
When cpu1 is automatically taken offline then back online, it seems to be randomly restored with the userspace governor instead of whatever governor it had when it was taken offline. It sometimes gets stuck online, too, even when the phone has been asleep for a long period of time. I still can't get either to happen consistently, however, and I don't know if nobody else is experiencing it, or if it's just gone unnoticed by others.
Enabling and disabling SetCPU seemed to have no effect on the consistency or frequency of the weirdness either. I may mess with my daemon to have it duplicate cpu0's governor to cpu1 when cpu1 comes online but doesn't match, but I have to assume this is just weirdness due to the whole "async cores are new" thing.
Thanks very much for your help, and if anyone else observes anything similar please let me know! D:
penguinicus said:
Whew, ok. After a lot of observation, I noticed that the reset to userspace corresponded with me putting the device to sleep.
I then tried disconnecting my device from USB and the daemon started crashing, being unable to read the sysfs files for cpu1, which never happened while connected over USB. I switched adb to TCP mode and found that the files were indeed gone, because cpu1 was offline. Apparently being connected over USB kept cpu1 online more often than it would be otherwise. This was causing some weird permission resets too, as the permissions would reset when the core was brought offline and back online.
When cpu1 is automatically taken offline then back online, it seems to be randomly restored with the userspace governor instead of whatever governor it had when it was taken offline. It sometimes gets stuck online, too, even when the phone has been asleep for a long period of time. I still can't get either to happen consistently, however, and I don't know if nobody else is experiencing it, or if it's just gone unnoticed by others.
Enabling and disabling SetCPU seemed to have no effect on the consistency or frequency of the weirdness either. I may mess with my daemon to have it duplicate cpu0's governor to cpu1 when cpu1 comes online but doesn't match, but I have to assume this is just weirdness due to the whole "async cores are new" thing.
Thanks very much for your help, and if anyone else observes anything similar please let me know! D:
Click to expand...
Click to collapse
I've noticed the exact same thing on my E3D. I'm running a "clean" custom ROM (CleanROM DE) with a famous third-party kernel (Tiamat Sense 1.1.0). I've also noticed that there is a lot of work in the Sensation forum trying to get the "Dual-Core" mode "unlocked" and that some ROMs have managed to pull it off. I think having a talk with any of these developers (or anyone who boasts "Dual Core" or "dual core patch") would help you get to the bottom of this.
It's extremely frustrating. Even using apps like System Tuner Pro, which have a special option to "force all CPUs active", does not seem to correct the sporadic governor switching back to userspace on the second core.
---------- Post added at 04:33 PM ---------- Previous post was at 04:12 PM ----------
curiousGeorge said:
I've noticed the exact same thing on my E3D. I'm running a "clean" custom ROM (CleanROM DE) with a famous third-party kernel (Tiamat Sense 1.1.0). I've also noticed that there is a lot of work in the Sensation forum trying to get the "Dual-Core" mode "unlocked" and that some ROMs have managed to pull it off. I think having a talk with any of these developers (or anyone who boasts "Dual Core" or "dual core patch") would help you get to the bottom of this.
It's extremely frustrating. Even using apps like System Tuner Pro, which have a special option to "force all CPUs active", does not seem to correct the sporadic governor switching back to userspace on the second core.
Click to expand...
Click to collapse
Oh. HTC's doing. My bad.
http://forum.xda-developers.com/showpost.php?p=17456447&postcount=7630
Of course, flashing this sucker could cause you to wind up with a bricked phone. Beggers can't be choosers. Now I'm off to flash Synergy. *sigh*
WOW, thanks for this info. That makes a lot of sense actually, I notice blazing performance for the most part on tiamat 1.1.0, and then out of nowhere it's slow as balls. Reboot *seems* to fix. I guess it's stock for me with a modified ramdisk, no OC, but that's cool. The shooter is pretty damn quick.
If any of you folks come up with something that corrects the userspace gov issue and perms, please post, I would like to use it. Thanks again.
Was having the same issue, looked all over and found a solution
http://forum.xda-developers.com/showpost.php?p=17456447&postcount=7630
Keep in mind these are flashable solutions as long as you stick with the same scaling_governor. Its been said to have both cpu's running same one.
Trickster MOD app is a tweaking tool to change various settings included in your kernel and other tweaks to your phone. It was made by me and my friend.
Screenshots and more information at links below.
Links
Google Play Store
AndroidPIT
Donate
WebsiteIf you like the app please rate it 5 stars and if you can, please donate.
Feature requests
Please post & vote features on our feedback site. So we can keep track of them
Bug Report
Please read the FAQ at our website before reporting any problem
If the FAQ does not help, use the Bug Report feature in our app & reference your ID here (the number in the mail subject).
Credits:
- Google
- Our great kernel devs
- grilledbacon for your beautiful icon
- Beta testers and translators for your contributions
- Anyone that helped us in some way
Disclaimer: This is provided with NO GUARANTY whatsoever. Use your common sense.
Initial version has been pushed to Google Play https://plus.google.com/u/0/116709542966452959662/posts/NZBEEJcuY5d, still pretty basic ATM moment. More stuffs will be added in later versions.
Thanks for adding support for the Nexus 10!
New version uploaded to market, should be available to you shortly
https://plus.google.com/u/0/116709542966452959662/posts/796By1dhjso
New version has been uploaded to Play Store, more info https://plus.google.com/116709542966452959662/posts/KDqLwP8gMH7
I had backed up my stock kernel using trickster mod, then flashed franco.kernel r-4. I confirmed everything on my phone was working properly, then paid for the donate version of the app so I could restore my stock kernel back to my nexus 10 4.2.1. All was well until I started using my device,
- OTG support was intermittent and my applications were running abnormally slow.
- Most of the folders that had write privileges could not be written to anymore when I tried making a new TWRP backup and ended up corrupting some of my existing application backups (combination of program not being able to write properly and OTG fading in and out).
I had to go into TWRP and restore from a nandroid after the fact and run Titanium to restore all my apps - OTG and worked fine after that. Any ideas, or a possible fix? I'd like to get the most out of the restore kernel function of the app.
bigdog2k7 said:
I had backed up my stock kernel using trickster mod, then flashed franco.kernel r-4. I confirmed everything on my phone was working properly, then paid for the donate version of the app so I could restore my stock kernel back to my nexus 10 4.2.1. All was well until I started using my device,
- OTG support was intermittent and my applications were running abnormally slow.
- Most of the folders that had write privileges could not be written to anymore when I tried making a new TWRP backup and ended up corrupting some of my existing application backups (combination of program not being able to write properly and OTG fading in and out).
I had to go into TWRP and restore from a nandroid after the fact and run Titanium to restore all my apps - OTG and worked fine after that. Any ideas, or a possible fix? I'd like to get the most out of the restore kernel function of the app.
Click to expand...
Click to collapse
Hi there,
The code to do the backup/restore couldn't possibility do that (e.g. it does not contain anything to change permissions). As I don't have the device, previous tests were done by my testers and I had positive results, so I will ask them to check again. For now, can you provide more information:
- ROM: I guess stocks?
- busybox: which app did you use to install it? busybox version?
- If possible do a new nandroid backup and do two bug report (Trickster MOD menu): once before you do a new backup and another when you restore kernel so I can check if there's anything goes wrong. Also send me the list of contents in /sdcard/kernel_backups
bigeyes0x0 said:
Hi there,
The code to do the backup/restore couldn't possibility do that (e.g. it does not contain anything to change permissions). As I don't have the device, previous tests were done by my testers and I had positive results, so I will ask them to check again. For now, can you provide more information:
- ROM: I guess stocks?
- busybox: which app did you use to install it? busybox version?
- If possible do a new nandroid backup and do two bug report (Trickster MOD menu): once before you do a new backup and another when you restore kernel so I can check if there's anything goes wrong. Also send me the list of contents in /sdcard/kernel_backups
Click to expand...
Click to collapse
Thanks for the quick response! I can do that later today. Yes, Rom is stock Jellybean rooted 4.2.1, busybox was installed using the stericson tool from the play store. Ill pm you later with extra info.
heya, loving the app. got one question if anyone can answer. does anyone have a list of kernels that support s2w? I'm using kmanta but it doesn't show CPU temp.
also on a side not, where would the options show up if its able?
edit: stock kernel from paranoid android for n10 and franco kernel doesnt show temperature either
Big eyes, I'm running trickster for the first time on the gnex and have a question. If I change the max freq from 1700 to 1600 and hit save. Then exit and go back in, its back to 1700. I then tried the freq lock option and that seems to work and it stays at 1600.
What's odd, is on the main screen (device status, core 0/1) it only goes to 1000mhz max. In realtime I mean....if I put my finger on the screen and move around it gets to 1000 but nothing higher. I'm wondering maybe if locking it down by checking that option basically locked it down to 1000 rather than 1600?
Confused..... Any thoughts? Thanks, great app
chugger93 said:
Big eyes, I'm running trickster for the first time on the gnex and have a question. If I change the max freq from 1700 to 1600 and hit save. Then exit and go back in, its back to 1700. I then tried the freq lock option and that seems to work and it stays at 1600.
What's odd, is on the main screen (device status, core 0/1) it only goes to 1000mhz max. In realtime I mean....if I put my finger on the screen and move around it gets to 1000 but nothing higher. I'm wondering maybe if locking it down by checking that option basically locked it down to 1000 rather than 1600?
Confused..... Any thoughts? Thanks, great app
Click to expand...
Click to collapse
Kernel/ROM issue. The app always reads frequency settings from the kernel itself so if it's at something then it's there, if a setting doesn't stick then something else changes it, and if the frequency does not go to the max that is because the set governor and its correspondent tunables, as in load is not high enough for the governor to set it to max.
bigeyes0x0 said:
Kernel/ROM issue. The app always reads frequency settings from the kernel itself so if it's at something then it's there, if a setting doesn't stick then something else changes it, and if the frequency does not go to the max that is because the set governor and its correspondent tunables, as in load is not high enough for the governor to set it to max.
Click to expand...
Click to collapse
Fair enough...I appreciate the quick response. It's Franco's latest R8, so I'm guessing its his kernel. Dam.... good to know. Maybe time to go back to stock kernel on Sentinel ROM...wonder if I can change the freq's in that kernel
chugger93 said:
Fair enough...I appreciate the quick response. It's Franco's latest R8, so I'm guessing its his kernel. Dam.... good to know. Maybe time to go back to stock kernel on Sentinel ROM...wonder if I can change the freq's in that kernel
Click to expand...
Click to collapse
You might wanna read our FAQ as it contains some information of what inside a JB device that might change the set frequencies. This should be patched in the kernel or worked around in the ramdisk. I coded for the N10 without a device so it's hard for me to say more. As for the frequency doesn't go to max, it's likely because of the load as I told you. Touching the screen is a relatively light load so it might not go to max frequency at all. That's normal.
Sounds good...last question, how do u reset the CPU spy like stats? Sorry if this is already covered in the faq
chugger93 said:
Big eyes, I'm running trickster for the first time on the gnex and have a question. If I change the max freq from 1700 to 1600 and hit save. Then exit and go back in, its back to 1700. I then tried the freq lock option and that seems to work and it stays at 1600.
What's odd, is on the main screen (device status, core 0/1) it only goes to 1000mhz max. In realtime I mean....if I put my finger on the screen and move around it gets to 1000 but nothing higher. I'm wondering maybe if locking it down by checking that option basically locked it down to 1000 rather than 1600?
Confused..... Any thoughts? Thanks, great app
Click to expand...
Click to collapse
You can't change the max clock freg because you need a kernel with a patched PowerHAL, which Franco doesn't have.
It only goes to 1000mhz because this is how the interactive governor is tuned. If you put it under increased load it will eventually increase. If you go into the governor parameters on trickster for interactive you will see "hispeed_freq" will most likely be set to 1000000 (1GHZ), which means the CPU will jump to this immediately. You can play around with these parameters for battery/performance but Franco and a few testers tune them significantly, although his N10 kernel is very young compared to his N4 and N7 kernels where the settings have been tuned to a tee, and I think the N10 settings might have just been copied over as a reliable base.
We're going to close individual device thread to consolidate our app thread into one at http://forum.xda-developers.com/showthread.php?p=28659480 , please go there if you need to discuss anything related to our app.
Thread Closed as per OP....
I've tried several apps that could get kernel access, but they either crash or fail to load. I'm rocking android 7.1.2.
Verdinex said:
I've tried several apps that could get kernel access, but they either crash or fail to load. I'm rocking android 7.1.2.
Click to expand...
Click to collapse
Don't have that problem; suggests a possible problem with bootloader unlock. Any issues securing root?
This is what I see after installing su14.zip arm non-64:
https://imgur.com/a/jnnOc03
Strangely enough, I can't see the root option in the developers panel and it should've been under 'demo mode'.
Can you run Kernel Adiutor? Could give me the name of the app you're using so I can check if it works.
Verdinex said:
This is what I see after installing su14.zip arm non-64:
https://imgur.com/a/jnnOc03
Strangely enough, I can't see the root option in the developers panel and it should've been under 'demo mode'.
Can you run Kernel Adiutor? Could give me the name of the app you're using so I can check if it works.
Click to expand...
Click to collapse
- "su14.zip"? Looks like run-of-the-mill "Advanced Root Checker" app which yields unreliable results IME
- "I'm rocking android 7.1.2" - based on your post history assume that means AEX customized for HDX??
- if AEX there is no integrated "root option"; however, it will expose all installed/working tools in "UI &UX" section
- I use multiple kernel apps including EX Kernel Manager and Kernel Adiutor to name a couple of popular ones; no issues
Davey126 said:
- "su14.zip"? Looks like run-of-the-mill "Advanced Root Checker" app which yields unreliable results IME
- "I'm rocking android 7.1.2" - based on your post history assume that means AEX customized for HDX??
- if AEX there is no integrated "root option"; however, it will expose all installed/working tools in "UI &UX" section
- I use multiple kernel apps including EX Kernel Manager and Kernel Adiutor to name a couple of popular ones; no issues
Click to expand...
Click to collapse
So which rom would you recommend instead? It seems this one has some issues regarding the root access. I've rooted my device with the file that can be found on official lineage website:
https://download.lineageos.org/extras
Verdinex said:
So which rom would you recommend instead? It seems this one has some issues regarding the root access. I've rooted my device with the file that can be found on official lineage website:
https://download.lineageos.org/extras
Click to expand...
Click to collapse
None of the popular HDX ROMs have "root access" issues. Your device, more accurately your firmware/software configuration, is the problem. Suggest a full wipe (all partitions) followed by a complete reinstall using the ROM of your choice. If issues persist it is likely the bootloader has not been fully/properly unlocked.
Davey126 said:
None of the popular HDX ROMs have "root access" issues. Your device, more accurately your firmware/software configuration, is the problem. Suggest a full wipe (all partitions) followed by a complete reinstall using the ROM of your choice. If issues persist it is likely the bootloader has not been fully/properly unlocked.
Click to expand...
Click to collapse
I've used magisk to root my device and everything's working fine now. I can run any kernel manager without much problem. Strangely enough, on demand cpu governor keeps my clocks at 300MHz. Interactive works a lot better. Is there any way to overclock my cpu without changing the kernel itself? I'd like to set single core at 2.2-2.3GHz and the rest at 1.9GHz to lower the temperature. 2.15 GHz is the maximum now.
Verdinex said:
I've used magisk to root my device and everything's working fine now. I can run any kernel manager without much problem. Strangely enough, on demand cpu governor keeps my clocks at 300MHz. Interactive works a lot better. Is there any way to overclock my cpu without changing the kernel itself? I'd like to set single core at 2.2-2.3GHz and the rest at 1.9GHz to lower the temperature. 2.15 GHz is the maximum now.
Click to expand...
Click to collapse
No way to overclock on stock kernel; there are no alternatives. Let interactive governor set the clocks and walk away. Twiddling CPU frequencies and other gov parameters on a handheld device of this type/vintage is a fools errand. Heat dissipation is not an problem unless you make it an issue with a thick/unventilated case after messing with CPU tunings. Just don't ...
Davey126 said:
No way to overclock on stock kernel; there are no alternatives. Let interactive governor set the clocks and walk away. Twiddling CPU frequencies and other gov parameters on a handheld device of this type/vintage is a fools errand. Heat dissipation is not an problem unless you make it an issue with a thick/unventilated case after messing with CPU tunings. Just don't ...
Click to expand...
Click to collapse
After playing around with different cpu governors, I discovered that the conservative mode is quite good when it comes demanding apps. Do you know an app that could allow me to set a particular cpu governors for different apps?
Verdinex said:
After playing around with different cpu governors, I discovered that the conservative mode is quite good when it comes demanding apps. Do you know an app that could allow me to set a particular cpu governors for different apps?
Click to expand...
Click to collapse
You'd have to roll your own script using a tool like Tasker or Xposed Edge.
Davey126 said:
You'd have to roll your own script using a tool like Tasker or Xposed Edge.
Click to expand...
Click to collapse
Thx, I'll stick to Kernel Adiutor it works really well and I can switch between governors on all core with a single button.