Related
I've done numerous tests with and without OC and on different kernels (both normal and custom). So far SetCPU have only given me headaches. Random crashes, unable to wakeup, kernel panics... The list goes on.
According to these tests I've come to the conclusion that SetCPU is very unstable on Galaxy S and should not be considered supported at all.
What's your experience with SetCPU on Galaxy S? Is it conclusive with mine? Do you experience the same problems with Overclock Widget, or is it more stable?
after using setcpu for a few weeks and getting similar problems, i removed it.
There is pretty much no use in it for this phone, they allready scale from 100mhz - max speed (even oc if you have one) automatically
mekwall said:
I've done numerous tests with and without OC and on different kernels (both normal and custom). So far SetCPU have only given me headaches. Random crashes, unable to wakeup, kernel panics... The list goes on.
According to these tests I've come to the conclusion that SetCPU is very unstable on Galaxy S and should not be considered supported at all.
What's your experience with SetCPU on Galaxy S? Is it conclusive with mine? Do you experience the same problems with Overclock Widget, or is it more stable?
Click to expand...
Click to collapse
Are you using a task manager? I had the same problem as you until I removed the auto task killer I had previously installed. Now I can set setcpu to 100000 min 1000000 max ondemand for main with a profile of screen off 100000 min 1000000 max conservative.
mekwall said:
I've done numerous tests with and without OC and on different kernels (both normal and custom). So far SetCPU have only given me headaches. Random crashes, unable to wakeup, kernel panics... The list goes on.
According to these tests I've come to the conclusion that SetCPU is very unstable on Galaxy S and should not be considered supported at all.
What's your experience with SetCPU on Galaxy S? Is it conclusive with mine? Do you experience the same problems with Overclock Widget, or is it more stable?
Click to expand...
Click to collapse
Yes. I have the same problem on my stock SGS.
Setcpu is not working good on sgs. Try to remove it, and you'll get better battery life.
Sent from my GT-I9000 using Tapatalk
I don't know of anyone who has tested it on the I9000 (and there may even be alternatives or the same thing floating around the development subforum), but they claim to have a 1.2ghz overclocked kernel here for the I9000. Maybe best to be avoided, but a 100% use at your own risk alternative, especially if you're a bit unsure on any of the translated steps, is available here.
http://translate.google.com/transla.../matpclub.com/board/index.html?id=tipn&no=144
been using it on a vibrant for about 10 days and had my first issue last nite - it froze the phone - had to remove battery to reboot, tried again, froze it again
i thought it was the last two apps i'd installed so i removed them, then there was an update, downloaded and installed, so i did a number of settngs changes, closing setcpu, running benchmark quad, reopening & changing settings etc
about half dozen times
and it's been fine so far -
I had the same issue. I would pull my phone out of my pocket and it wouldn't wake up. I could call my phone and it would ring but my phone would do nothing until I removed the battery and re-booted. This happened numerous times until I uninstalled SETCPU and it stopped doing that so I knew I found the culprit and haven't tried it again since (I need to receive calls as priority 1)
doesnt anyone search anymore i posted this issue over a month ago. setting cpu mode to anything on than conservative can cause freezes as the system doesnt get enough cpu to wake up.. it is even mentioned in set cpu to use conservative.
lgkahn said:
doesnt anyone search anymore i posted this issue over a month ago. setting cpu mode to anything on than conservative can cause freezes as the system doesnt get enough cpu to wake up.. it is even mentioned in set cpu to use conservative.
Click to expand...
Click to collapse
I have it set to Conservative on my Vibrant, and I'm having the same issues as the OP.
Edit: When I go into SetCPU after pulling the battery and resetting, it shows it's limiting my CPU to 19.2 MHz. Yes, 19.2. Insane.
tt4079 said:
Are you using a task manager? I had the same problem as you until I removed the auto task killer I had previously installed. Now I can set setcpu to 100000 min 1000000 max ondemand for main with a profile of screen off 100000 min 1000000 max conservative.
Click to expand...
Click to collapse
I killed my task manager service, and it didn't freeze up! I'm going to uninstall the task manager and see if I have any more freezing issues in the future.
Edit: Nope, didn't work. Removing SetCPU.
mekwall said:
What's your experience with SetCPU on Galaxy S?
Click to expand...
Click to collapse
I have the Vibrant version with T-Mo. After rooting, Titanium, Barnacle and Root Explorer all seem to be working fine.
After a few minutes of installing SetCPU, my phone wouldn't wake up. Rebooted twice, same results.
Removed SetCPU and the phone's working fine again. Maybe the minimum speed should be set at 250 or so?
George
To begin with, you don't even need SetCPU
SGS phone automatically regulates the speed of the CPU depending on your usage.
most of the time in Stock ROM it hover around 200 Mhz, very few times it uses the max 1000 Mhz
only when there are lots of background apps running
when you runt he CPU at 1000 Mhz it gets hot fast, and it consumes more battery
good thing is it drops back down to 100~200 Mhz as soon as it's done with the heavy tasks
so having the SetCPU installed will defeat the whole purpose of power saving and CPU cycle on demand
mekwall said:
I've done numerous tests with and without OC and on different kernels (both normal and custom). So far SetCPU have only given me headaches. Random crashes, unable to wakeup, kernel panics... The list goes on.
According to these tests I've come to the conclusion that SetCPU is very unstable on Galaxy S and should not be considered supported at all.
What's your experience with SetCPU on Galaxy S? Is it conclusive with mine? Do you experience the same problems with Overclock Widget, or is it more stable?
Click to expand...
Click to collapse
I don't even remember the last time where we had a fixed clock on a cpu Allgamer. The point of setcpu is to either downclock or overclock your cpu. You might want to limit your clock to 0.6GHz just to save some juice.
Anyway, after a few hours of head scratching and 2 reboots, I uninstalled and refunded it.
I agree. Setcpu is rather unstable. I have feeling lots of the kernel overclockers have problems but blame the kernel while they should blame setcpu
i have just installed setcpu. it's probably been about a day or so now and i have not experienced any problems?
i have an Australian model (I9000) just flashed to the canadian firmware for the 3br fix.
Is it just generally not recommended to use setcpu on my SGS?
please advise. Thanks
toansta said:
i have just installed setcpu. it's probably been about a day or so now and i have not experienced any problems?
i have an Australian model (I9000) just flashed to the canadian firmware for the 3br fix.
Is it just generally not recommended to use setcpu on my SGS?
please advise. Thanks
Click to expand...
Click to collapse
It depends on what you're using it for. If you know what you're doing, it can be a useful tool. If you want to overclock, you'll want setcpu. If you want to underclock (lets say you're out and about, with 40% battery), then you can set the maximum cpu speed lower so that the battery will last longer.
You don't need setcpu if you're simply going to tell it that you want your max speed to be 1ghz, with a min of 100mhz and you want it to be "conservative". Your phone does that by default, and frankly the SGS is pretty good on its own when it comes to battery management.
thanks reuthermonkey, i will keep that in mind.
What are the perfect setcpu 2.0.3 settings for froyo JP6 I9000 rooted? Thanx in advance
Telanis said:
I have it set to Conservative on my Vibrant, and I'm having the same issues as the OP.
Edit: When I go into SetCPU after pulling the battery and resetting, it shows it's limiting my CPU to 19.2 MHz. Yes, 19.2. Insane.
Click to expand...
Click to collapse
Menu and rescan the device should restore the normal values. those values are not the ones actually used!
Ok. i have refraind from asking questions for quite some time. I have lurked, and read and read and read. i learned how to flash roms and kernels and most otherthings i want to do without bothering anyone with noob questions...up until now.
I flashed the suckerpuch #50 kernel in attempts to try out 1.4ghz, but unless im completely missing something my phone dosnt ever seem to go above 1 ghz on this kernel. im not positive this is the case, but heres what leads me to beleive it:
cpu usage widget on my screen never shows above 1ghz, while a different 1.2ghz OC i tired shows it peaking up to 1.2 right after boot, and if i exit quadrant it shows 1.2 for a sec before dropping back down, same with any 3d game, when i quickly close it and look at the widget, it will show running at 1.2. Yet suckerpuch never shows above 1.0 UNLESS i run CPUsetting and set it to "performace" which it then appears to be running at 1.4 AT ALL TIMES, and the phone isnt stable so thats a no go.
am i missing some key piece of information here? did i flash it wrong? is there some other type of modification i need to do to get it to actually run at 1.4?
ive read through the suckerpunch #50 opening posts multiple times and i believe it tells me i need to modify some file or other, possibly, but im not understanding.
any info/help would be much appreciated!
P.S. this forum, the devs, and all the things supplied to make my phone awesome are absolutley great.
I believe SuckerPunch is set to 1Ghz as default unless you change the setting with Voltage Control (or SetCPU). Mind you that 1.4Ghz is not run well on a lot of devices and may cause freezing. And I'm glad some people take up reading before asking questions
p.p.s- im aware not to use set CPU.
honestly i cant see the point of it with any kernel. dosnt seem to do anything.
koreancanuck said:
I believe SuckerPunch is set to 1Ghz as default unless you change the setting with Voltage Control (or SetCPU). Mind you that 1.4Ghz is not run well on a lot of devices and may cause freezing. And I'm glad some people take up reading before asking questions
Click to expand...
Click to collapse
yeah im using voltage control- minus 125 on all and the phone seems to work well.
i just find it weird that the kernel peaks at 1ghz, where as another peaks at 1.2ghz (both without use of setCPU, and WITH use of Voltage control)
when i was using setCPU and putting it on performance, i could see it was running at 1.4, pretty much at ALL TIMES. MOST things ont he phone worked well, ultra fast. the problems i noticed is that neocore would freeze (which i understand to be a sign of instability) and speedtest would either give me REALY slow speeds, or force close every time. were as set on conservative it would work perfectly and neocore wouldnt freeze.
oFUNGUSo said:
yeah im using voltage control- minus 125 on all and the phone seems to work well.
i just find it weird that the kernel peaks at 1ghz, where as another peaks at 1.2ghz (both without use of setCPU, and WITH use of Voltage control)
when i was using setCPU and putting it on performance, i could see it was running at 1.4, pretty much at ALL TIMES. MOST things ont he phone worked well, ultra fast. the problems i noticed is that neocore would freeze (which i understand to be a sign of instability) and speedtest would either give me REALY slow speeds, or force close every time. were as set on conservative it would work perfectly and neocore wouldnt freeze.
Click to expand...
Click to collapse
I guarantee you you cannot run -125 at 1.4GHz, unless you have some kind of wonder-phone.
In Voltage Control, you're set to 1.4GHz, correct?
If you are, have you made sure to enable the 1.4GHz state in Voltage Control via the bottom menu as well? If you don't, even though you're set at 1.4GHz it will not run at that.
Auridran said:
I guarantee you you cannot run -125 at 1.4GHz, unless you have some kind of wonder-phone.
In Voltage Control, you're set to 1.4GHz, correct?
If you are, have you made sure to enable the 1.4GHz state in Voltage Control via the bottom menu as well? If you don't, even though you're set at 1.4GHz it will not run at that.
Click to expand...
Click to collapse
my bad, im running -125 on i think its the gb 1.2ghx kernel. thats working fine.
im going to reflash #50 and try what you said.....though im pretty sure i already had it set. i know i was setting the mv from 100 up to 1400
ok i flahsed back to #50, and went into voltage control. it DID cap my freq at 1000, i changed it and set 100 through 1400 to -75mv.
i ran stress test for about 30 seconds and exited, and my CPU widget shows 99% usegage at 1ghz
hmm
i flashed back to gb reorient 1.2ghz and during boot up as soon as the CPU widget loaded it shows 1.2ghz. ran stress test again and exited in the middle of it, widget showed 99% usage at 1.2ghz.
i realyl dont know what im missing here, but it does seem as thought the suckerpunch #50 is not running above 1ghz for me.
comparing all forms of benchmark type tests between the two kernels, everything EXPCEPT necore is scoreing better on the 1.2 kernel. neocore onthe 1.3 kernel is giving me 55.4 fps, whereas on the 1.4 kernel its giving me 65 and some change fps. i know benchmark scores arent that meaningfull in some cases, and on the #50 thread it even states that some may be lower on that kernel.......but i swear ive done something wrong because i dont see how it can be running at 1.4
oFUNGUSo said:
ok i flahsed back to #50, and went into voltage control. it DID cap my freq at 1000, i changed it and set 100 through 1400 to -75mv.
i ran stress test for about 30 seconds and exited, and my CPU widget shows 99% usegage at 1ghz
hmm
i flashed back to gb reorient 1.2ghz and during boot up as soon as the CPU widget loaded it shows 1.2ghz. ran stress test again and exited in the middle of it, widget showed 99% usage at 1.2ghz.
i realyl dont know what im missing here, but it does seem as thought the suckerpunch #50 is not running above 1ghz for me.
comparing all forms of benchmark type tests between the two kernels, everything EXPCEPT necore is scoreing better on the 1.2 kernel. neocore onthe 1.3 kernel is giving me 55.4 fps, whereas on the 1.4 kernel its giving me 65 and some change fps. i know benchmark scores arent that meaningfull in some cases, and on the #50 thread it even states that some may be lower on that kernel.......but i swear ive done something wrong because i dont see how it can be running at 1.4
Click to expand...
Click to collapse
Like I said in my previous post, are you actually enabling the frequencies or just selecting 1.4GHz as the max in Voltage Control?
You have to enable 1.4GHz in the "states" menu of Voltage Control for it to work, as well as set 1.4GHz as the max in Voltage Control. SuckerPunch does not overclock by default because a lot of phones do not play nicely with 1.4GHz.
Auridran said:
Like I said in my previous post, are you actually enabling the frequencies or just selecting 1.4GHz as the max in Voltage Control?
You have to enable 1.4GHz in the "states" menu of Voltage Control for it to work, as well as set 1.4GHz as the max in Voltage Control. SuckerPunch does not overclock by default because a lot of phones do not play nicely with 1.4GHz.
Click to expand...
Click to collapse
ha. this had me confused......there was no "states" menu. that got me pokeing around and then i realized i didnt have the newest version of voltage control(had 2.0 instead of 3.0). i do NOW. and i see the option to select the states. so, i selected 7 states up to 1400. and selected the max frequency of 1400. set a voltage for all, and applied.
still the same result. lol. its not running over 1ghz according to cpu stats. it SAYS it frequency range is from 100 to 1400. but it aint going over 1000.
*sigh* there has to be something else im missing
For suckerpunch's new builds, it's set to startup at 1ghz. After like 15mins. you can run some tests and it should be at max clock.
YESSSSSSSSSSSSS!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
ITS WORKING! and dayum i see and improvement in stats! benchmarks look fantasitc, neocore isnt freezeing, phone isnt freezing on lockscreen.....will have to run stress test for a while to see for sure but it appears stable!
1400 -50
1300 -50
1200 -75
1120 -75
1000-75
800- 100
600-100
400-100
i really cant figure out the right way to do that, but it seems to be working so far
for one, i needed the newest version of voltage control.....for two, im thinking that i was supposed to wipe the davik cache before install, because it did a different song and dance when i did that. for three.....for whatever reson the 1000hz version didnt seem to want to work for me....and finally....waiting for a while after the phone booted up i did see the CPU start to jump up to 1.4.
thanks guys!
Welcome to Myrt Torture Tester.
As always, this app is BETA, expect bugs. Tested on Stock GB/HP Extreme & guestekrnL, CM7/Etana, CM9/HP ICS & Harsh.
Will only work on OC/UV-kernels.
It's primarily intended as a tester to find your stable frequencies and voltages, but can also be used as a battery-life tester and a rough benchmark.
This app comes with no more support than the 2nd post. If you don't know how to use it after reading that you don't need it.
Using this app, you will be capable of causing actual damage to your device. I take no responsibility for any consequences.
This app shall under NO CIRCUMSTANCES be included in any ROM, or uploaded any other place than this post.
Changelog
0.5.6 - Fixes crash when system is unresponsive for long periods.
0.5.5 - Keeps reading the current cpu-frequency throughout the test. (Some beta kernels unexpectedly change frequencies during the test, this allows you to see it.)
0.5.4 - CM7-compatability-fix.
0.5.3 - Fixes frequency not being set on guestekrnL.
0.5.2 - Fixed links on about page.
0.5.1 - Plays nice on non-oc kernels.
0.5.0 - First BETA-release
HOW TO USE:
In its simplest form, the app is a benchmark. You start a CPU-test at a specific frequency for a specific time-period and get the average Mflops-result. If you are going to compare different kernels, it will give you a normalized to 1Ghz result as well. Different kernels usually have different frequency-tables, so test them at the closest steps you can find, then compare the normalized result.
The more useful aspect of it is to test if a specific frequency is stable at a specific voltage. The app only allows to test the frequencies in the voltage table - because if it is stable at those frequencies it will be stable at all other frequencies which fall within those voltage-steps. If you have undervolted too much, the phone will usually reboot pretty quickly. If you have undervolted "on-the-edge", the phone will likely freeze. If you have undervolted so that it is basically stable, but sometimes fails, you'll either get a crash to the desktop or MTT will inform you of a calculation error. If you are testing for stability you need to test for at least 60 minutes to have any confidence in the result, I've had several tests fail after 45-50 minutes.
It can also be used to see what impact, if any, undervolting has on the processors' power-consumption. After you have made sure a frequency/voltage-pair is stable, you can run a battery-test and compare it to an identical test at stock voltage. This will simply run from a full(ish) battery to a certain battery percentage, and give you how long it was able to run. Since the battery-percentage is pretty loosely coupled to the actual battery-charge, it will also give figures for consumption per minute or second. This kind of test should also be run over a long a period as possible to get accurate results. Measuring from 100% to 90% will only give you an indication, prone to error. You can not compare a test between, say 100->50 and 100->20, because the discharge rate varies with the charge-level.
For most accurate testing and benchmarking: enable flightmode, unplug the device, freeze any apps which may run in the background, uninstall everything you don't need, wait 3 minutes after booting before testing, do not touch the screen or move the device while a test is running. Even then there are Android quirks which will cause some variation in the results. Therefore the same test should be repeated as many times as you can afford.
The major enemy of stability, assuming you have enough voltage, is heat. Make sure to test the device under the same conditions as it will be used. If you're going to overclock on a hot summer's day, test it on a hot summer's day. MTT dims the screen to minimize the impact it has on the battery and heat-generation, be aware that your device will be hotter when the screen is at normal brightness.
Stability tests should also be performed at different battery-levels. If your device is stable when the battery is fully charged, it does not automatically mean it will be stable when it is almost discharged.
MTT logs all succesfull tests (max 200 lines.) If you enable "Store log on sdcard" in preferences the log will be saved to /sdcard/MTT_Log.txt.
Known issues:
o Sometimes the device will give you half the score you should get. I do not know if this is a kernel or android-bug, but it seems that both test threads get scheduled to run on the same core, and the second core goes unused, even when it is active. Exiting the app and starting it again does not help usually, but killing it sometimes does. Rebooting is always an option.
o Not a "known issue", but this app lets you under- and overvolt in 5mV steps. Different kernels may handle this differently, either not undervolting at all, or adjusting it to the nearest 25mV step. It has worked on the kernels I have tried, but there are too many kernels out there to be sure.
TrymHansen said:
HOW TO USE:
In its simplest form, the app is a benchmark. You start a CPU-test at a specific frequency for a specific time-period and get the average Mflops-result. If you are going to compare different kernels, it will give you a normalized to 1Ghz result as well. Different kernels usually have different frequency-tables, so test them at the closest steps you can find, then compare the normalized result.
The more useful aspect of it is to test if a specific frequency is stable at a specific voltage. The app only allows to test the frequencies in the voltage table - because if it is stable at those frequencies it will be stable at all other frequencies which fall within those voltage-steps. If you have undervolted too much, the phone will usually reboot pretty quickly. If you have undervolted "on-the-edge", the phone will likely freeze. If you have undervolted so that it is basically stable, but sometimes fails, you'll either get a crash to the desktop or MTT will inform you of a calculation error. If you are testing for stability you need to test for at least 60 minutes to have any confidence in the result, I've had several tests fail after 45-50 minutes.
It can also be used to see what impact, if any, undervolting has on the processors' power-consumption. After you have made sure a frequency/voltage-pair is stable, you can run a battery-test. This will simply run from a full(ish) battery to a certain battery percentage, and give you how long it was able to run. Since the battery-percentage is pretty loosely coupled to the actual battery-charge, it will also give figures for consumption per minute or second. This kind of test should also be run over a long a period as possible to get accurate results. Measuring from 100% to 90% will only give you an indication, prone to error. You can not compare a test between, say 100->50 and 100->20, because the discharge rate varies with the charge-level.
For most accurate testing and benchmarking: enable flightmode, freeze any apps which may run in the background, uninstall everything you don't need. Even then there are Android quirks which will cause some variation in the results. Therefore the same test should be repeated as many times as you can afford.
The major enemy of stability, assuming you have enough voltage, is heat. Make sure to test the device under the same conditions as it will be used. If you're going to overclock on a hot summer's day, test it on a hot summer's day. MTT dims the screen to minimize the impact it has on the battery and heat-generation, be aware that your device will be hotter when the screen is at normal brightness.
Stability tests should also be performed at different battery-levels. If your device is stable when the battery is fully charged, it does not automatically mean it will be stable when it is almost discharged.
MTT logs all succesfull tests. If you enable "Store log on sdcard" in preferences the log will be saved to /sdcard/MTT_Log.txt.
Click to expand...
Click to collapse
Really interesting! Big thanks
Sent by LG Optimus 2x
Doesn't seem to work for me, always goes to my max set freq of 1.1ghz, no matter if i set it higher or lower in the program. Same mflops result too. Changing the max freq in guesteoc results in that new max freq being used all the time in the program.
kfallz said:
Doesn't seem to work for me, always goes to my max set freq of 1.1ghz, no matter if i set it higher or lower in the program. Same mflops result too. Changing the max freq in guesteoc also results in that new max freq being used all the time in the program.
Click to expand...
Click to collapse
Ok, thanks for the feedback. Which version of guestekrnL? I'm mainly using that myself, works here on 1.7.
EDIT: Ok, I've found the reason. I had disabled the 00cpufreqenabler script (to simulate other kernels), but it didn't work after I enabled it. Will release a fix as soon as I can override the permission properly.
kfallz said:
Doesn't seem to work for me, always goes to my max set freq of 1.1ghz, no matter if i set it higher or lower in the program. Same mflops result too. Changing the max freq in guesteoc results in that new max freq being used all the time in the program.
Click to expand...
Click to collapse
App updated to 0.5.3 - fixes the issue where frequency is not set on guestekrnL.
TrymHansen said:
App updated to 0.5.3 - fixes the issue where frequency is not set on guestekrnL.
Click to expand...
Click to collapse
Giving the new version a try, and not that it matters now but I'm using 1.7.0t2
Yea working fine now
Updated to 0.5.4 - Now the sliders behave properly on CM7.
Myrt, do you think that we have more than one different hardware inside our phones? Because the results are way to different from one ppl to another, for example, Temasek can't OC even to 1.1GHz, and Vadonka puts at 1.4GHz without the burn, with the tests, I get to 70ºC in 2min with 1.1GHz, but the strange thing, is that the phone doesn't lags or anything to show that it haves an high-temp. Did you have some screen of the test in your own phone? I have exchanged my first phone because it always get too hot for me, my second gets hot always when I try to play any game.
chaozbr said:
Myrt, do you think that we have more than one different hardware inside our phones? Because the results are way to different from one ppl to another, for example, Temasek can't OC even to 1.1GHz, and Vadonka puts at 1.4GHz without the burn
Click to expand...
Click to collapse
No, we have pretty much the same hardware, specification-wise. It is very common to have different tolerances for speeds and voltages. I'm pretty sure that the moment Vadonka tries my app at 1.4Ghz, he too will get a hot CPU.
, with the tests, I get to 70ºC in 2min with 1.1GHz, but the strange thing, is that the phone doesn't lags or anything to show that it haves an high-temp.
Did you have some screen of the test in your own phone? I have exchanged my first phone because it always get too hot for me, my second gets hot always when I try to play any game.
Click to expand...
Click to collapse
I didn't take a screenshot, if that is what you ask, I have the app abort the test at 72C, and the temp falls quickly back to normal. I don't even have vadonkas kernel installed anymore, I'm a stock man, installed CM7 just to make the app compatible.
However, I guess we can encourage people to post their temps and frequencies here, in this thread. If anyone manages to run 1.4Ghz for more than 5 minutes and not reach 70C I'll be impressed. (I do all my temp-testing when plugged in though, probably easier when unplugged.)
TrymHansen said:
No, we have pretty much the same hardware, specification-wise. It is very common to have different tolerances for speeds and voltages. I'm pretty sure that the moment Vadonka tries my app at 1.4Ghz, he too will get a hot CPU.
I didn't take a screenshot, if that is what you ask, I have the app abort the test at 72C, and the temp falls quickly back to normal. I don't even have vadonkas kernel installed anymore, I'm a stock man, installed CM7 just to make the app compatible.
I had these temps in Spica HP kernel, so stock rom, CM7 with Vadonka, I saw an 84ºC (leave my brother playing in the phone, and he said to me that the phone was hot haha)
Click to expand...
Click to collapse
TrymHansen said:
However, I guess we can encourage people to post their temps and frequencies here, in this thread. If anyone manages to run 1.4Ghz for more than 5 minutes and not reach 70C I'll be impressed. (I do all my temp-testing when plugged in though, probably easier when unplugged.)
Click to expand...
Click to collapse
If we can't run 1.4GHz for me than 5min without reaching 70ºC, and 70ºC is the maximum temp, we can't play or let the phone in the frequency for daily use?
chaozbr said:
If we can't run 1.4GHz for me than 5min without reaching 70ºC, and 70ºC is the maximum temp, we can't play or let the phone in the frequency for daily use?
Click to expand...
Click to collapse
Well, no, I'm not saying that, there are too many unknowns.
1) We don't know if the reported temp is correct.
2) We dont' really know the max temp for tegra2. (The 70C figure is not from an official source.)
3) This app is meant to test stability, so temps will get as hot as possible. A game will most likely not stress the CPUs quite as much.
But, all that taken into consideration, 1.4Ghz is probably too high for sustained operation. It will probably be fine for normal use, where the CPUs get to rest once in a while, but not for prolonged CPU-heavy tasks (which this app demonstrates.) That being said, this app is designed to produce as much heat as possible to test stability.
It took me 41sec till the CPU reached 70°C from 40°C on 1408MHz using latest beta 3.0.y etana.
Tapatalk 2-vel küldve az én Optimus 2X-ről
Thanks for this, this app looks great!
I usually don't UV as it introduces some instability to my device even at low -UV values which isn't worth the rather small gain in battery life - but it's still good to know that this app existis if I'll change my mind someday
tonyp said:
Thanks for this, this app looks great!
I usually don't UV as it introduces some instability to my device even at low -UV values which isn't worth the rather small gain in battery life - but it's still good to know that this app existis if I'll change my mind someday
Click to expand...
Click to collapse
Yeah, I don't usually undervolt either, I carry spare batteries, and the fear of unstability was always lurking when I did. (I did make an undervolt app afterall, had to test it.)
So with this app my hope is that many myths will be dispelled - using this people should be able to find out what they need.
(A few weeks ago I had forgotten the batteries in a bag in a hotellroom, far away. I really, really, really needed 20 extra minutes of battery-life then. So now I will probably undervolt to voltages I know are safe, just in case something similar should happen. Sometimes 5 minutes make all the difference in the world.)
I cannot set any other V (no UV or OV) using Temasek 99 with latest Vadonka Beta 11.05. Kernel. The program always crash after the 2sec warmup.
Gesendet von meinem Optimus 2X mit Tapatalk 2
kennbo82 said:
I cannot set any other V (no UV or OV) using Temasek 99 with latest Vadonka Beta 11.05. Kernel. The program always crash after the 2sec warmup.
Gesendet von meinem Optimus 2X mit Tapatalk 2
Click to expand...
Click to collapse
Ok, I need some more info I think. First of all, is the app the latest version? If yes, do you get a superuser message the first time you start the app after a reboot? Which values do the bottom sliders show right after starting the app for the first time?
Tried it myself today on temasek 100 with both 3.0.31 OC and the latest beta (OC-only of course), worked on both for me. One installation however failed, and I had to repair the system-partition before the app would work again. (Superuser pretended to give access, but /system wasn't really writable, so it failed.) Check which apps are listed in superuser, if you see the same app listed a lot of times, that's your problem.
TrymHansen said:
Ok, I need some more info I think. First of all, is the app the latest version? If yes, do you get a superuser message the first time you start the app after a reboot? Which values do the bottom sliders show right after starting the app for the first time?
Tried it myself today on temasek 100 with both 3.0.31 OC and the latest beta (OC-only of course), worked on both for me. One installation however failed, and I had to repair the system-partition before the app would work again. (Superuser pretended to give access, but /system wasn't really writable, so it failed.) Check which apps are listed in superuser, if you see the same app listed a lot of times, that's your problem.
Click to expand...
Click to collapse
Sooorry I'm too stupid, tried the 216 Mhz only which is too slow for the total phone use and test. Program works fine!
Again sorry and thanks for the tool
kennbo82 said:
Sooorry I'm too stupid, tried the 216 Mhz only which is too slow for the total phone use and test. Program works fine!
Again sorry and thanks for the tool
Click to expand...
Click to collapse
Ok, good to know, thanks. I will have to look into it anyway, it shouldn't crash even at 216Mhz, but I'll take that as low-priority, in other words tomorrow ;-)
I have just updated to Cyanogenmod 7.2 stable on my P990 Optimus 2x (Fido Canadian version).
When I go to the CPU settings I cannot change the min/max frequencies or select a CPU governor.
I have done some research and found kernels like Vork and FauxDS, but they are for 7.1 nightlies so I am not sure if it is a good idea to flash them.
What I am most afraid about is the possibility of getting the black screen of death while my phone sleeps (it happened once already but it stopped after I changed to a different charger and reflashed CM7.2).
The reason why I want to change the settings is that there tends to be scrolling lag (my old HTC Desire Z was faster!) and there is also a bit of lag waking up the phone (lockscreen does not appear instantly when pressing power button).
Does anyone know what is the best kernel I can flash to enable frequency/governor changes in CM7.2 stable? I would prefer not to change the ROM if possible as I am kinda nooby in comparison to the XDA community in general. Thanks for any help.
I suggest you try Etana kernel. It has the things you need. Also use CPU master to set profiles and minimum CPU frequencies. I used something like min freq 352 MHz when asleep (don't worry doesn't drain battery) and I had no lag whatsoever.
heavencanwait said:
I suggest you try Etana kernel. It has the things you need. Also use CPU master to set profiles and minimum CPU frequencies. I used something like min freq 352 MHz when asleep (don't worry doesn't drain battery) and I had no lag whatsoever.
Click to expand...
Click to collapse
Are you referring to http://forum.xda-developers.com/showthread.php?t=1427646 ?
If so, do I just flash the zip off the sd card using ClockworkMod? And what are the differences between lite and DS?
Thanks.
All right, after reading, I determined I don't need LOC/HOC/DS as I don't intend to raise the max frequency, I just want to raise min frequency.
Installed the Etana kernel via CWM and it works. I can raise minimum frequency, though I still cannot see a governor option! Any idea how to get my governor setting back?
You can use CPU master and set some profiles. It's not exactly the same as governors but it's as close as it gets and it gets the job done pretty well. As far as I remember there were two governors for O2X - powersave and performance. I've always used performance and I can't tell if there's a HUGE difference in battery usage. I guess performance utilizes higher CPU frequencies when the device is asleep and somehow scales faster to even higher frequencies when needed to. But this might be complete crap as I am not too much into how CPUs work...
And please use the THANKS button (right below the user name), it kind of forum etiquette and it keeps people motivated to help others. Thank the Etana developer too.
heavencanwait said:
You can use CPU master and set some profiles. It's not exactly the same as governors but it's as close as it gets and it gets the job done pretty well. As far as I remember there were two governors for O2X - powersave and performance. I've always used performance and I can't tell if there's a HUGE difference in battery usage. I guess performance utilizes higher CPU frequencies when the device is asleep and somehow scales faster to even higher frequencies when needed to. But this might be complete crap as I am not too much into how CPUs work...
And please use the THANKS button (right below the user name), it kind of forum etiquette and it keeps people motivated to help others. Thank the Etana developer too.
Click to expand...
Click to collapse
Thanks for letting me know, I didn't realize there was a thanks button at first haha.
Fixed!
As for now CPU Master in his profiles let's you choose governor. We have only 2. Power save nd Performance.
Sent from LG 2X using Tapatalk 2 Elite
Q&A for [KERNEL][D5803&D5833] AndroPlusKernel
Some developers prefer that questions remain separate from their main development thread to help keep things organized. Placing your question within this thread will increase its chances of being answered by a member of the community or by the developer.
Before posting, please use the forum search and read through the discussion thread for [KERNEL][D5803&D5833] AndroPlusKernel. If you can't find an answer, post it here, being sure to give as much information as possible (firmware version, steps to reproduce, logcat if available) so that you can get help.
Thanks for understanding and for helping to keep XDA neat and tidy!
Thanks
Very nice kernel, fast and smooth - great work and port.
Anyone knows if it's possible to use any third party kernel app to boost the headset volume?
Working
Hi this is my first post, infact I joined so I could report my findings.
I would just like to confirm so far that V5 working on my Z3C using the .93 firmware. I didn't need to wipe cache/dalvik/data partitions either.
For purposes of settings configuration, I'll be using TricksterMod
For stress testing purposes I'll be using Stability Test 2.7
Overclocking
Overclocking seems to work, I'll stress test and report back with the results.
I was wrong, it is unstable when overclocking and stress testing, with the phone force rebooting 1-5 seconds in to the stability test which loaded all 4 cores. Tried 2899Mhz and 2880Mhz (were both unstable and didn't try 2.72, 2.75, 2.57 either - I might try stability testing them.) Was completely stable at 2.47GHz, and it actually stuck there, no throttling in place! I stopped the stress test when the CPU temp was around 77-80C as my hand was getting burnt! The battery temp reached about 46C. It took 10 mins of stress testing for this to be reached. When the stock kernel was in place it would begin throttling after about 1-20 seconds under full load. First 2.2GHz, then 2 then 1.7 then 1.5 and eventually 1.25 after about 2-4mins.
I couldn't do a single core test though. I suspect as most games use single core or dual cores it wouldn't get overly hot.
My opinion is that fast clocked phones should be marketed with 2 speeds. The highest it'll reach under a boost mode (turbo for Intel's i5 and i7 series), and a slower speed that it'll average out at under thermal pressure. The Z3Cs would be turbo: 2.5GHz and normal: 1.5GHz.
Governors
These are quick tests I'm performing here to provide a quick look at responsiveness and potential unnecessary CPU jump ups.
Wheatley - most of the time it's hovering at top speed when approx CPU usage is 9%, it'll occasionally hunt down to 300Mhz but then right back up to 2899MHz. This one is speedy though. Governor tunables available.
Lagfree - idles at 300MHz - 960MHz then hunts up to 2.2-2.9Ghz when needed and turns on the second core. It seems to slow back down to idle. This one seems to have fast scrolling, sliding notifications pane quickly). No governor tunables.
SmartassV2 - idles at 300Mhz, speeds up to 422-960 on core 0, and turns on core 1 at 1.2-1.5GHz for a few seconds under fast scrolling and notification pane opening. Seems quite responsive. Probably good for battery life hopefully with the responsiveness of Interactive. No governor tunables.
Lionheart - Idles at 960 MHz for a few seconds then slows down to 300MHz with the odd increase to 729MHz. Core 0 and Core 1 reaches 1.26GHz under scrolling, notification pane opening. As fast as the others. Governor tunables available.
Hotplugging
Intelliplug appears to work better than MP-Decision - now only one core is on during idle, instead of 2.
MP-Decision was disabled to avoid conflicts.
Undervolting
I've only undervolted 300MHz to 675MHz from 775MHz as of writing this post.
Tried quickly undervolting in trickstermod by setting them all about -75mV, stability tested it, appears stable. I'll tweak the voltages a bit better when I do a scaling stability test.
Issues I've experienced
Sound Control is the only thing so far that causes a reboot. However music still plays over the speakers and headphones. Equaliser works too.
Upon rebooting, the CPU top speed will set itself to 2.2GHz, despite being set higher and saved at a higher speed in Trickstermod. Certainly trivial though.
Strange thing I've noticed: On the undervolt part I've noticed that there is a freq called 3033MHz, but no 2899MHz freq. Just an observation.
Misc
Force Fast Charge confirmed working! Before flashing new kernel charge went up 5% in about half an hour. It's now up another 5% in a matter of 5-10mins. This is when plugged to my PC.
Vibrator strength - set to 20 from 31, much quieter on table and can still feel it.
I'd like to say a huge thank you to DooMLoRD and AndroPlus for developing this stable kernel. Now my Z3C is worth the £28 a month I'm paying for again. Since this silly lad decided to bork the camera, Bravia functionality etc by rooting it on the first day. I'll report back and edit the post with my findings.
Max won't go beyond 2.266 GHz, Min won't change.
Hi All,
Firstly, great job with the kernel. Our Z3C is such a capable piece of kit and this just makes it that much better.
I'm running stock, 23.0.A.2.93, unlocked, rooted.
Problem:
I've tried using both SetCPU and No-Frills and while both show frequencies that are supposedly selectable above 2.266 GHz, neither app would actually respond. Meaning the max frequency will still only be 2.266 GHz even if I tried selecting something like 2.458 GHz (which should be selectable given that stock is 2.458 GHz.) See caps.
In addition, the Min value just won't change at all from 300 MHz. The frequencies scale up and down as the load changes but I can't raise the Min (again using both SetCPU and No-Frills) from 300 MHz.
Tried: I've tried turning off stamina mode and frozen apps that could control CPU activity (e.g. I use DS Battery Saver). I changed up Governors and Schedulers. I've tried re-flashing the kernel and it still doesn't change these behaviors.
Question/Need help: Just wanted to ask if anybody else have any problems setting the max frequency beyond 2.266 GHz and/or changing the minimum frequency from 300 MHz? Would appreciate any help resolving this behavior.
Thanks in advance!
pjmanalo said:
Hi All,
Firstly, great job with the kernel. Our Z3C is such a capable piece of kit and this just makes it that much better.
I'm running stock, 23.0.A.2.93, unlocked, rooted.
Problem:
I've tried using both SetCPU and No-Frills and while both show frequencies that are supposedly selectable above 2.266 GHz, neither app would actually respond. Meaning the max frequency will still only be 2.266 GHz even if I tried selecting something like 2.458 GHz (which should be selectable given that stock is 2.458 GHz.) See caps.
In addition, the Min value just won't change at all from 300 MHz. The frequencies scale up and down as the load changes but I can't raise the Min (again using both SetCPU and No-Frills) from 300 MHz.
Tried: I've tried turning off stamina mode and frozen apps that could control CPU activity (e.g. I use DS Battery Saver). I changed up Governors and Schedulers. I've tried re-flashing the kernel and it still doesn't change these behaviors.
Question/Need help: Just wanted to ask if anybody else have any problems setting the max frequency beyond 2.266 GHz and/or changing the minimum frequency from 300 MHz? Would appreciate any help resolving this behavior.
Thanks in advance!
Click to expand...
Click to collapse
Try installing TricksterMod (from Google Play store, trust me, you'll love it!) Then go to General and set the max speed to 2.46GHz or higher, and try using the Ondemand Governor too. I noticed that it wouldn't stick properly sometimes when using Interactive governor. If it doesn't stick for you then turn Frequency Lock on. Then check in the info tab that it's hitting the higher speed. Personally I'd recommend leaving the min speed on 300MHz. If you need constant high speeds, select the performance governor.
DBCJoey said:
Try installing TricksterMod (from Google Play store, trust me, you'll love it!) Then go to General and set the max speed to 2.46GHz or higher, and try using the Ondemand Governor too. I noticed that it wouldn't stick properly sometimes when using Interactive governor. If it doesn't stick for you then turn Frequency Lock on. Then check in the info tab that it's hitting the higher speed. Personally I'd recommend leaving the min speed on 300MHz. If you need constant high speeds, select the performance governor.
Click to expand...
Click to collapse
Thanks! That did the trick!
Odd that my usual app for the job across 4 other phones - SetCPU - doesn't work on what should essentially be the same job. [emoji55]
Please make sound_control drivers work so its possible to boost headphone volume on the Xperia Z3 Compact... Thanks
Nice work, a lot of updates I like it!
Is it possible to add a change log?
Thanks!
kernel for d5803 with the .93 but not .105
i search a kernel for the d5803 with the last .93 french version of phone
.5.77
Works great! Thanks a lot
Link for Z3C_D5803_AndroPlusKernel_v10.zip is dead
Pls upload in another location.
Yay sound control is working, thank you so much you're the best!
How to make this?
Hi AndroPlus,
I'm trying to figure out how one would go about building this boot.img that you've created.
What platform and compiler are you using?
Where are you getting sources the for the kernel? This file?
c9af6fc647060fb85dd646798453ec8f 23.0.A.2.105.tar.bz2
How do you construct boot.img from zImage + recovery?
Sorry if these are dumb questions.
Edit: never mind, I figured this out.
http://developer.sonymobile.com/kno...evices/how-to-build-and-flash-a-linux-kernel/ contains most of the information I needed.
The arm version of gcc that ships with Ubuntu 14.04 worked fine - arm-none-eabi-gcc (4.8.2-14ubuntu1+6) - no need to track down any mystery binaries. I did have to make several modifications to the kernel source to get it to build. Interestingly, some of the cpufreq stuff contained code that was incorrect. Someone at Samsung needs to go look up what "sequence point" means.
The hardest part was figuring out how to turn the zImage + ramdisk into something I could boot.
This: https://github.com/sonyxperiadev/mkqcdtbootimg was the correct tool to use - again, no need to track down any mystery mkbootimg or dtbTool binaries.
Hope this is helpful to someone. As someone new to Xperia dev, I found most of the information out there worse than useless.