Related
I got an EVO myself and managed to overclock it to 1.267GHz. We could probably go even higher, but that requires more extensive changes. Here are the changed source files: http://dl.dropbox.com/u/36553/supersonic_oc-uv.tar.gz
It also includes an undervolt courtesy of pershoot (a whole 0.1V lower when idle, which should help battery drain a lot!). Here is a test update.zip: http://dl.dropbox.com/u/36553/supersonic_oc-uv_signed.zip (working WiFi). Make sure to do a nandroid backup first!
Sensors/camera do not currently work in this kernel because the source is not yet complete.
There is an overvolt on the highest frequencies. It should not cause any issues. Thanks to toastcfh for his source, Herver for creating the first booting overclocked kernel on the EVO, all the kernel devs in the N1 community, especially pershoot and kmobs, richardtrip from the Desire community, koush for anykernel updater, and all who helped to root the EVO. Enjoy!
Update 6/15: Fixed 950mV undervolt. Raised voltages from 1228800 above to improve stability. Source files updated and update.zip uploaded.
old said:
I posted these in an earlier thread, but I think I should bring this to everyone's attention. Here are some simple patches that will overclock the Evo up to 1.267GHz (though by any means it's not guaranteed to get that high!):
Edit: please don't flash the below update.zip, as it doesn't seem to boot. Devs, the patches are still below, and a zimage is in the anykernel update zip.
Update: I compiled it myself. Here is an update.zip: http://dl.dropbox.com/u/36553/update-oc_signed.zip. This uses toastcfh's sources, Koush's excellent anykernel updater and a config.gz pulled from a stock Supersonic. If this doesn't boot, no damage will be done to your phone. Just reboot into recovery and reinstall your ROM. You can test overclocking by grabbing SetCPU from my sig or changing the cpufreq values if you know how. Once we get this working, I will create an undervolted version for battery savings. I'd get an Evo, but it's sold out everywhere T_T
http://dl.dropbox.com/u/36553/supersonic_oc.zip
cpufreq.c and acpuclock-scorpion.c are in arch/arm/mach-msm, and supersonic_defconfig is just the config. I don't have an Evo, so I can't test these out, but these should work with the source files toastcfh posted earlier: http://forum.xda-developers.com/showthread.php?t=686240.
Undervolting can be achieved simply by adjusting the voltages (in mV) in the acpu_freq_tbl table in acpuclock-scorpoion.c. For those not familiar with his work, you can see pershoot's safe (but lower) voltages for the N1 here: http://github.com/pershoot/nexus-kernel/blob/master/acpuclock-scorpion.c. If you want to overvolt or go under 1000mV, you have to make one further adjustment to the voltage regulator in board-supersonic.c.
Note that the N1 and the Evo are completely different devices (with similar processors), so what may work for the N1 may not work for the Evo.
Click to expand...
Click to collapse
Awesome... until I am sure this is stable, I will wait, but I must have overclock
jigglywiggly said:
Awesome... until I am sure this is stable, I will wait, but I must have overclock
Click to expand...
Click to collapse
I don't recommend overclocking, the devices already can run warm at current settings.
But then again, you are entitled to do as you wish.
Options better than overclocking are to remove the Sprint bloatware and strip down parts of Sense, this speeds things up alot by removing excess background applications.
Well it ain't too bad, my Droid at 1 ghz(550 stock) runs great.
So this should be good too.
I like overclocking as much as the next geek, but on the Evo... WHY?
This phone has yet to respond anything slower than instantly to my input and I expect a difference of .26 ghz would not make it so it could read my thoughts.
Undervolting, on the other hand, could help with the abysmal battery life...
phinnaeus said:
I like overclocking as much as the next geek, but on the Evo... WHY?
This phone has yet to respond anything slower than instantly to my input and I expect a difference of .26 ghz would not make it so it could read my thoughts.
Undervolting, on the other hand, could help with the abysmal battery life...
Click to expand...
Click to collapse
Some people love to run benchmarks. Try to improve on them over and over. Others like the feeling of just going faster. Also some apps or some in the future might require a faster running phone. What better way to achieve that (other then buying a new phone) then OC.
Also in the future, are you not gunna buy a 1.5ghz processor phone running 2.1 because a 1ghz processor can already do it? OC has many benefits, just not to everyone, most because they dont understand the consequences or benefits.
My htc hero clocked to 768mhz can run 2.1 and everything near instant. Still doesnt keep be from wanting a 1ghz or higher.
Also one more thing, 2.2 will be out soon. With full flash support, i have a feeling its going to require a lot of CPU. An OC Evo would probably run many things from 2.2 a lot easier.
Well, I agree with the benchmarking comments but on the subject of Froyo... Froyo is designed to be even faster on equivalent hardware than 2.1... and 2.1 on my Evo is smooth as silk. And the Evo handles most Flash I can throw at it, so I'm not sure. I have however noticed a small performance difference with live wallpaper running, so maybe OCing will clear that up a bit.
Jus10o said:
Some people love to run benchmarks. Try to improve on them over and over. Others like the feeling of just going faster. Also some apps or some in the future might require a faster running phone. What better way to achieve that (other then buying a new phone) then OC.
Also in the future, are you not gunna buy a 1.5ghz processor phone running 2.1 because a 1ghz processor can already do it? OC has many benefits, just not to everyone, most because they dont understand the consequences or benefits.
My htc hero clocked to 768mhz can run 2.1 and everything near instant. Still doesnt keep be from wanting a 1ghz or higher.
Also one more thing, 2.2 will be out soon. With full flash support, i have a feeling its going to require a lot of CPU. An OC Evo would probably run many things from 2.2 a lot easier.
Click to expand...
Click to collapse
You know, this is an excellent point and I feel like I failed the developer community in some small way when you had to bring it up. You shouldn't have to provide a reason for wanting progress, it should happen for its own sake.
Proceed.
Jus10o said:
Some people love to run benchmarks. Try to improve on them over and over. Others like the feeling of just going faster. Also some apps or some in the future might require a faster running phone. What better way to achieve that (other then buying a new phone) then OC.
Also in the future, are you not gunna buy a 1.5ghz processor phone running 2.1 because a 1ghz processor can already do it? OC has many benefits, just not to everyone, most because they dont understand the consequences or benefits.
My htc hero clocked to 768mhz can run 2.1 and everything near instant. Still doesnt keep be from wanting a 1ghz or higher.
Also one more thing, 2.2 will be out soon. With full flash support, i have a feeling its going to require a lot of CPU. An OC Evo would probably run many things from 2.2 a lot easier.
Click to expand...
Click to collapse
Agree with everything you just said, and I am less than excited about the benchmarks with the EVO thus far
Roman G said:
Agree with everything you just said, and I am less than excited about the benchmarks with the EVO thus far
Click to expand...
Click to collapse
HAHAHA.. im sorry but i had to laugh at your sig.
Jus10o said:
HAHAHA.. im sorry but i had to laugh at your sig.
Click to expand...
Click to collapse
Ya I sold my hero a couple weeks before the EVO dropped, and had to use my wife's old phone.
phinnaeus said:
I like overclocking as much as the next geek, but on the Evo... WHY?
This phone has yet to respond anything slower than instantly to my input and I expect a difference of .26 ghz would not make it so it could read my thoughts.
Undervolting, on the other hand, could help with the abysmal battery life...
Click to expand...
Click to collapse
I'm not trying to start an argument at all but you have to think broader mathematically. If the processor runs at 1000Mhz or 999mhz (In the case of my N1) a .267 increase is 267mhz, a 26.7% increase in computational performance... That's pretty raw especially in cellphone terms, that have no way to vent or displace that kind of heat.
Has anyone noticed how much hotter the kickstand gets? I bet you that it's attached to something the cpu is connected to. $5 bucks says we will see KICKSTAND HEATSINK + FAN MOD 2GHZ OC! LOL
thebluecoat said:
I'm not trying to start an argument at all but you have to think broader mathematically. If the processor runs at 1000Mhz or 999mhz (In the case of my N1) a .267 increase is 267mhz, a 26.7% increase in computational performance... That's pretty raw especially in cellphone terms, that have no way to vent or displace that kind of heat.
Has anyone noticed how much hotter the kickstand gets? I bet you that it's attached to something the cpu is connected to. $5 bucks says we will see KICKSTAND HEATSINK + FAN MOD 2GHZ OC! LOL
Click to expand...
Click to collapse
This is a patch for developers, not a discussion thread for the merits (or lack thereof) of overclocking.
If you don't want to overclock your phone, nobody is making you.
It has worked out fine in the Desire and Nexus One community.
coolbho3000 said:
This is a patch for developers, not a discussion thread for the merits (or lack thereof) of overclocking.
If you don't want to overclock your phone, nobody is making you.
It has worked out fine in the Desire and Nexus One community.
Click to expand...
Click to collapse
how do i apply this ???????
can u make it flashable?
mccurt29 said:
how do i apply this ???????
can u make it flashable?
Click to expand...
Click to collapse
need to follow the directions in the first post. the .c files need to be loaded into the kernel source (right now just have toast's soruce) and the kernel has to be compiled with them.
if somebody wants to compile up a kernel with these, we could start testing. if i get some free time i'll help out here and compile a kernel but somebody else will prob beat me to it!
please someone let us know how this works out for your evo 4g... benchmark, test results... anything to show better/worse performance, whether your got ALOT warmer, or just a lil, etc.. ne information would be greatly appreciated it, i'm not a 100% noob, i just know how to follow instructions very well, doesn't mean i understand everything that's going on, just the basic concept of what i want. thank you in advance.
madunix said:
please someone let us know how this works out for your evo 4g... benchmark, test results... anything to show better/worse performance, whether your got ALOT warmer, or just a lil, etc.. ne information would be greatly appreciated it, i'm not a 100% noob, i just know how to follow instructions very well, doesn't mean i understand everything that's going on, just the basic concept of what i want. thank you in advance.
Click to expand...
Click to collapse
I'm waiting for someone to compile this first.
If nobody steps forward, I'll make my own attempt, but I'll be working blind with no test hardware.
I'm pretty sure most devs are waiting on the htc source kernal before playing with this. I can't wait but I understand them being hesitant to do too much with this for now. Still awesome of you to post it and have it rdy for when that time comes!
sounds good.
madunix said:
sounds good.
Click to expand...
Click to collapse
Your signature says you have a radio of 1.32.651.6, just to let you know that is not a radio, that is an HTC software OTA update. My radio is 1.39.00.05.31 for example
Has anyone used any kernels for HTC Sensation 4g (TMO US) above 1.566Ghz?
I am currently using a ROM by Mike1986 and kernel witch took my phone to 1.566Ghz and i am loving it.
However, i am a big OC freak and i would like some pointers on any tested kernels above 1.566Ghz, preferably that work well with Mike1986's ROM as i find them to be most stable our of all i have tried (i have tried 3 different types).
Any input is highly appreciated and welcome. Thanks team.
dronepro said:
Has anyone used any kernels for HTC Sensation 4g (TMO US) above 1.566Ghz?
I am currently using a ROM by Mike1986 and kernel witch took my phone to 1.566Ghz and i am loving it.
However, i am a big OC freak and i would like some pointers on any tested kernels above 1.566Ghz, preferably that work well with Mike1986's ROM as i find them to be most stable our of all i have tried (i have tried 3 different types).
Any input is highly appreciated and welcome. Thanks team.
Click to expand...
Click to collapse
http://forum.xda-developers.com/showthread.php?t=1256668
I honestly don't see the benefits except lower battery life and a warmer device, I doubt you'll notice the speed difference past like 1GHz. All of the higher freqs seem pretty much the same and aren't really worth the bother with, use a governor that accesses higher freqs quicker and you won't be needing high freqs.
Ace42 said:
http://forum.xda-developers.com/showthread.php?t=1256668
I honestly don't see the benefits except lower battery life and a warmer device, I doubt you'll notice the speed difference past like 1GHz. All of the higher freqs seem pretty much the same and aren't really worth the bother with, use a governor that accesses higher freqs quicker and you won't be needing high freqs.
Click to expand...
Click to collapse
What governor would you suggest??? also, i know that when using you phone and maybe even playing games i wouldn't see much of a difference between the stock 1.2 Ghz vs 1.5 or 1.9Ghz but it's all about the bragging rights, lol. I like to go to my geeky friends and slappin’ a higher clock frequency than they have on their poor devices or silly I Phones. Also I have to tell you from my tests I do see Mike1986’s latest 3.6.7/4.1.7 Rom using much less battery life than stock Rom. I have noticed about a 10% improvement, maybe even more.
Also i do see a big improvement in Benchmarks. When i was using stock 1.2Ghz setting i was benchmarking much lower than the 1,5Ghz i am at now.
Thanks again for the input.
The benefit of custom ROMS and kernels, is that they are optimized to run well. Unless the chef sucks, most big name ROMS are optimized and more efficient.
I was running 1.2ghz for a while and whenever I'm playing Snes roms it would FC. After bumping it up to 1.72ghz it never did it again. Although 1.72ghz may be overkill for what I use my phone for but everything seems to be running extremely smooth.
dronepro said:
What governor would you suggest??? also, i know that when using you phone and maybe even playing games i wouldn't see much of a difference between the stock 1.2 Ghz vs 1.5 or 1.9Ghz but it's all about the bragging rights, lol. I like to go to my geeky friends and slappin’ a higher clock frequency than they have on their poor devices or silly I Phones. Also I have to tell you from my tests I do see Mike1986’s latest 3.6.7/4.1.7 Rom using much less battery life than stock Rom. I have noticed about a 10% improvement, maybe even more.
Also i do see a big improvement in Benchmarks. When i was using stock 1.2Ghz setting i was benchmarking much lower than the 1,5Ghz i am at now.
Thanks again for the input.
Click to expand...
Click to collapse
I use conservative(810/192), super charger v6, ultra smooth rosie on 4.1.6. I haven't used benches, but they'll be low on this ROM since I'm assuming it's based on the Sense 3.5 leak for sensation.
Rickdaddy said:
I was running 1.2ghz for a while and whenever I'm playing Snes roms it would FC. After bumping it up to 1.72ghz it never did it again. Although 1.72ghz may be overkill for what I use my phone for but everything seems to be running extremely smooth.
Click to expand...
Click to collapse
What Kernel is this? Can you provide a LINK?? thanks.
Thanks again to all for the imput.
This thread is regarding discussion about Doomlord UC kernel for our device. This thread is specially for general chit chat, opinions, Q and A related to his kernel so that main thread of Doomlord kernel in developement section will have less crowd of unnecessary questions & posts.
So I request senior members to help noobs here. Also if necessary Doomlord will also answer ur important querries here.
Important :-Kernel is working universally perfectly with EVERY firmwares without any problems.
Current version :- v14 (21st MAY 2012)
Original Thread for updates
http://forum.xda-developers.com/showthread.php?t=1357747
http://forum.xda-developers.com/showpost.php?p=26353627&postcount=1659
Thank you.
Regards,
Vishal
Well now that kernel development is in full swing, could we look into fixing the notification bug? I don't know much about android development, but I read someone say it could be kernel related... That I feel is the biggest bug facing our beloved phone on gb!
manrock111 said:
@doomlord finally you got our device, this device has seen lots of bad times and always neglected by senior members or devs. I know lots of your work involve sony ericsson devices,qualcomm devices .
So how do you feel about our device? It's cost only 17k and you know now how it is.
What do you think of it compare to others.
Sent from my GT-I9003 using Tapatalk
Click to expand...
Click to collapse
Sent from my GT-I9003 using Tapatalk
I've just loaded UC v2 on my DXKP9 SL with ext4 and it works fine.
bigeyes0x0 said:
I've just loaded UC v2 on my DXKP9 SL and it works fine.
Click to expand...
Click to collapse
can u tell us which governer & scheduler will be best for battery & performance?
I'm not an expert in android governer and i/o scheduler but I've been using smartassv2 and vr with better performance than stock with a slightly higher battery drain than original settings when actively used. When idle I have about the same battery drain as default.
vishal24387 said:
can u tell us which governer & scheduler will be best for battery & performance?
Click to expand...
Click to collapse
I've used smartassv2 (proposed by XDA_Bam) and Scary and interactive (proposed by Doomlord). I'm using my phone mostly as audioplayer (ie almost always audioplayer is working). For me is more better smartassv2. Because with other I have lags while changing pages in the applications menu or changing screens in the Go launcher Ex.
Also I'm using BFQ as IO scheduler, because I know it's the best for not too many concurrent processes reading/writing to disk. Also usually it doesn't have too big sense for mobile devices, because we don't have a lot applications working with hdd/sd card
can anyone tell me which governer and i/o should i choose for best performance while playing hd games (i dont care about the battery drain). actually i downloaded shadowgun non tegra but it gets a bit laggy when too many enemies appear on screen, i tried setting performance governer but its the same.
nail16 said:
can anyone tell me which governer and i/o should i choose for best performance while playing hd games (i dont care about the battery drain). actually i downloaded shadowgun non tegra but it gets a bit laggy when too many enemies appear on screen, i tried setting performance governer but its the same.
Click to expand...
Click to collapse
1. You have to close all other application and services
2. The performance governor just put CPU to the max frequency, that's why it's the best for gaming
3. If it's still laggy you need or overclocked kernel (we don't have it yet) or make lower settings in the game (or in the Chainfire 3D)
@nail16: You need to OC or a new phone, that game is just too heavy for this phone.
After doing some research on available i/o scheduler I say simple i/o (sio) is the best from theory considering we're accessing a random access device. For quantitative conclusion someone needs to bench.
EDIT: @vishal, can you include a list of firmware versions that work with current UC kernel in your OP? For now I think we have XXKPM, XXKPQ and mine DXKP9. Also I think "[UC Kernel] Q & A" is a better title (the shorter it is the better), please change it if you feel appropriate.
Also here is very good benchmark of the IO schedulers
But remember that it was 2009, now it's 2011 and some things can be changed/optimized
what are the benefits of running custom kernel other than governer and i/o?, sorry noob question
falex007 said:
Also here is very good benchmark of the IO schedulers
But remember that it was 2009, now it's 2011 and some things can be changed/optimized
Click to expand...
Click to collapse
I dare say those benchs are not enough to evaluate those i/o schedulers. e.g. In read_test.c, you see that the program will read chunk of 1MB one by one which is a very rare case that our phone would do in day to day usage. We need a better benchmark methodology to reach a conclusive answer.
What is TinyRCU? Link
---------- Post added at 05:40 PM ---------- Previous post was at 05:37 PM ----------
nail16 said:
what are the benefits of running custom kernel other than governer and i/o?, sorry noob question
Click to expand...
Click to collapse
On the kernel page, you saw lots of line saying added in v1, added in v2.
All those are not present in the vanilla kerel by samsung.
U can check each in google to see what it means and what good it brings the mobile.
Also read these links to understand more:
Fugumod Kernel
Steam Kernel
There are lots more.
Thanks ganesh, you are very noob friendly.
bigeyes0x0 said:
I dare say those benchs are not enough to evaluate those i/o schedulers. e.g. In read_test.c, you see that the program will read chunk of 1MB one by one which is a very rare case that our phone would do in day to day usage. We need a better benchmark methodology to reach a conclusive answer.
Click to expand...
Click to collapse
How many applications do you start every day that hard works with hard disk? Do you think IO scheduler can be bottleneck for usual usage of our device? I don't think so. Only games during load of new level, gallery (during making thumbnails) and launcher (during loading icons) work hard with "hard disk"
Updated op.
I was looking into fs benchmarks for debian linux.
Found that JFS was the least CUP intensive off all.
Would it be the same on Android?
falex007 said:
How many applications do you start every day that hard works with hard disk? Do you think IO scheduler can be bottleneck for usual usage of our device? I don't think so. Only games during load of new level, gallery (during making thumbnails) and launcher (during loading icons) work hard with "hard disk"
Click to expand...
Click to collapse
No, I actually don't really care as long as my phone feel smooth, that's why I don't go out and do a benchmark myself to ultimately find an answer. My point simply was that benchmark method in your link is inadequate. Finding faults in other ppl's works is one thing I do at my job after all.
bigeyes0x0 said:
No, I actually don't really care as long as my phone feel smooth, that's why I don't go out and do a benchmark myself to ultimately find an answer. My point simply was that benchmark method in your link is inadequate. Finding faults in other ppl's works is one thing I do at my job after all.
Click to expand...
Click to collapse
What these benchmarks do is to guide us to use the appropriate one for our use.
As Doom says we can switch the IO Scheculer/Governor on the fly.
So where is the problem in doing so.
To keep Andrei from going crazy with non-development related questions, I've opened the following thread. That way we keep the development thread for development topics and this for general discussion.
Good deal.
First feature request: Fading in/out LEDs! I've pinged the dev for LightFlow, and while he can implement it on a software level, it would be supremely inefficient and probably drain the battery like crazy.
cmd512 said:
Good deal.
First feature request: Fading in/out LEDs! I've pinged the dev for LightFlow, and while he can implement it on a software level, it would be supremely inefficient and probably drain the battery like crazy.
Click to expand...
Click to collapse
Actually that's something which belongs in the dev thread. If anything you can just post on the issue tracker on Github, I've grown vary of XDA.
There's a new TI LED controller so I can't port the old Philips controls from the old phones. I'll rewrite it eventually but I got other things to do now.
AndreiLux said:
Actually that's something which belongs in the dev thread. If anything you can just post on the issue tracker on Github, I've grown vary of XDA.
There's a new TI LED controller so I can't port the old Philips controls from the old phones. I'll rewrite it eventually but I got other things to do now.
Click to expand...
Click to collapse
Done, posted on Github!
just got my s4 one day ago. rooted immediately, flashed omega7.0 and the latest perseus. did not touch anything on the u/v in stweak. no o/c as well. but the antutu score is so low 21000+, i thought it at least supposed to be 27000+?
i tried adam kernel, had the same result. is that normal? did i miss anything?
astoncheah said:
just got my s4 one day ago. rooted immediately, flashed omega7.0 and the latest perseus. did not touch anything on the u/v in stweak. no o/c as well. but the antutu score is so low 21000+, i thought it at least supposed to be 27000+?
i tried adam kernel, had the same result. is that normal? did i miss anything?
Click to expand...
Click to collapse
Congrats, you win the award for the first post about useless benchmark scores in this thread.
cmd512 said:
Congrats, you win the award for the first post about useless benchmark scores in this thread.
Click to expand...
Click to collapse
thanks for you response. as i mentioned i got it only one day, without touching anything on u/v & o/c. i guess this is considered lower score than the stock one? i merely care about bench mark, i just want to know is this normal? or i need few days to let it settle down? fyi, i tried more than 10 times and it only score between 20k-21k. perhaps you can be kind enough to give me some idea rather than congrats me.
astoncheah said:
thanks for you response. as i mentioned i got it only one day, without touching anything on u/v & o/c. i guess this is considered lower score than the stock one? i merely care about bench mark, i just want to know is this normal? or i need few days to let it settle down? fyi, i tried more than 10 times and it only score between 20k-21k. perhaps you can be kind enough to give me some idea rather than congrats me.
Click to expand...
Click to collapse
It is normal for benchmark to be lower than on stock kernel all devices differ. Following the development of this kernel it has been stated that the kernel is not intended for high benchmarks but more for overall stability, smoothness and better battery life.
Benchmarks is just a number and a high benchmark does not always mean your phone is smooth and performs very well under normal user conditions.
Hope this answers your question and should I be mistaken about anything anyone welcome to rectify me on those points.
Sent from my GT-I9500 using Tapatalk 2
xtacy1 said:
It is normal for benchmark to be lower than on stock kernel all devices differ. Following the development of this kernel it has been stated that the kernel is not intended for high benchmarks but more for overall stability, smoothness and better battery life.
Benchmarks is just a number and a high benchmark does not always mean your phone is smooth and performs very well under normal user conditions.
Hope this answers your question and should I be mistaken about anything anyone welcome to rectify me on those points.
Sent from my GT-I9500 using Tapatalk 2
Click to expand...
Click to collapse
i never meant that perseus is running low on benchmark, in fact, it's the highest between chainfire's cfroot, wanam's adam. i tried all these 3 kernel, all score only between 19k-21k. i just wonder if i have accidentally limited my cpu speed? i will do a factory reset and re-test it.
again, i dont or merely care about benchmark, i just care if my s4 is running normally.
astoncheah said:
i never meant that perseus is running low on benchmark, in fact, it's the highest between chainfire's cfroot, wanam's adam. i tried all these 3 kernel, all score only between 19k-21k. i just wonder if i have accidentally limited my cpu speed? i will do a factory reset and re-test it.
again, i dont or merely care about benchmark, i just care if my s4 is running normally.
Click to expand...
Click to collapse
I have kanged Andrei's response with respect to benchmarks.
Originally Posted by AndreiLux
Samsung cheats during benchmarks by enabling a thermal boost mode which raises the throttling temps by 10°C (90° -> 100°C). I disabled the permissions to this. Practical effects is things like Antutu will give lower scores. This is more realistic of actual performance and I will leave it as such. Benchmark freaks can enable the boost mode manually: /sys/class/thermal/thermal_zone0/boost_mode .
The first throttling speed is 1400MHz and that's why it spends a lot of time there during benchmarking.
Effectively, overclocking on this device makes very little sense unless you can achieve more thermal headroom with undervolting. I personally couldn't get over 1.8GHz on random loads, and couldn't get over 1.7GHz on benchmarking (Antutu 31300).
Nevertheless, the option is there for people who want to stick their phones into their freezers. For everybody else, stop being concerned about numbers.
This seems to affect only a rare number of people and I still don't know the cause. Just cycle the screen once and be done with it.
jlevy73 said:
I have kanged Andrei's response with respect to benchmarks.
Originally Posted by AndreiLux
Samsung cheats during benchmarks by enabling a thermal boost mode which raises the throttling temps by 10°C (90° -> 100°C). I disabled the permissions to this. Practical effects is things like Antutu will give lower scores. This is more realistic of actual performance and I will leave it as such. Benchmark freaks can enable the boost mode manually: /sys/class/thermal/thermal_zone0/boost_mode .
The first throttling speed is 1400MHz and that's why it spends a lot of time there during benchmarking.
Effectively, overclocking on this device makes very little sense unless you can achieve more thermal headroom with undervolting. I personally couldn't get over 1.8GHz on random loads, and couldn't get over 1.7GHz on benchmarking (Antutu 31300).
Nevertheless, the option is there for people who want to stick their phones into their freezers. For everybody else, stop being concerned about numbers.
This seems to affect only a rare number of people and I still don't know the cause. Just cycle the screen once and be done with it.
Click to expand...
Click to collapse
"This seems to affect only a rare number of people and I still don't know the cause" guess i am one of them:silly:
did you guys note the download link for perseus alpha 4 rc1 on andrei github? , dont flash it unless you know whats in there
edit: just flashed it (couldnt resist). based on the newer sources it seems, camera works fine with latest cam firmware
lets wait for Andrei release......build seems to be a release candidate....so its should not be for long....
bala_gamer said:
did you guys note the download link for perseus alpha 4 rc1 on andrei github? , dont flash it unless you know whats in there
edit: just flashed it (couldnt resist). based on the newer sources it seems, camera works fine with latest cam firmware
Click to expand...
Click to collapse
Thanks for sharing bro but i'm too stupid to find the link would you mind sharing more please?:victory:
update: found it: deleted. sorry
astoncheah said:
just got my s4 one day ago. rooted immediately, flashed omega7.0 and the latest perseus. did not touch anything on the u/v in stweak. no o/c as well. but the antutu score is so low 21000+, i thought it at least supposed to be 27000+?
i tried adam kernel, had the same result. is that normal? did i miss anything?
Click to expand...
Click to collapse
I received mine yesterday, rooted, flashed doc rom to remove bloatware, flashed perseus kernel.
I ran a few antutu and I was getting 27-28k without changing any setting.
Sent from my Nexus 4 using xda premium
@GSeeker
Pls remove the link since we dont know the nature of the version or lets wait unitl andrei unveils it.
bala_gamer said:
did you guys note the download link for perseus alpha 4 rc1 on andrei github? , dont flash it unless you know whats in there
edit: just flashed it (couldnt resist). based on the newer sources it seems, camera works fine with latest cam firmware
Click to expand...
Click to collapse
@Power, What rom are you running? I suspect it's rom related.
I'm using omega 7, one of the user experienced the same as mine. Going to flash new rom.
Sent from my GT-I9500 using xda app-developers app
astoncheah said:
just got my s4 one day ago. rooted immediately, flashed omega7.0 and the latest perseus. did not touch anything on the u/v in stweak. no o/c as well. but the antutu score is so low 21000+, i thought it at least supposed to be 27000+?
i tried adam kernel, had the same result. is that normal? did i miss anything?
Click to expand...
Click to collapse
With Omega / Perseus, is your camera still working????
Sent from my GT-I9500 (Hassan Khalid Malik) using xda premium
HassanM said:
With Omega / Perseus, is your camera still working????
Sent from my GT-I9500 (Hassan Khalid Malik) using xda premium
Click to expand...
Click to collapse
From recovery flash Perseus kernel + Camera Fix https://www.dropbox.com/s/20kyqg154h8w198/camera-ISP-fw.MEAplusFix.zip
Then reboot. Your camera should be working.
What exactly do i get if i will flash custom kernel to my hammerhead ? What do they provide to make it noticeable to an user ?
m4nu4l said:
What exactly do i get if i will flash custom kernel to my hammerhead ? What do they provide to make it noticeable to an user ?
Click to expand...
Click to collapse
Custom kernels provide improvements, tweaks and more power to the user.
Some provide patches to use certain mods... E.g. Multirom patched kernels
Others allow your phone to be overclocked and undervolted. (Both CPU & GPU)
Some just save more battery than others, some have display tweaks included allowing the user to copy Samsung's saturated screens or even lower saturation.
Remember, not all kernels are compatible with all roms.
Read on the bases of the roms and kernel compatibility!
Happy flashing and good luck!
Valdorous said:
Custom kernels provide improvements, tweaks and more power to the user.
Some provide patches to use certain mods... E.g. Multirom patched kernels
Others allow your phone to be overclocked and undervolted. (Both CPU & GPU)
Some just save more battery than others, some have display tweaks included allowing the user to copy Samsung's saturated screens or even lower saturation.
Remember, not all kernels are compatible with all roms.
Read on the bases of the roms and kernel compatibility!
Happy flashing and good luck!
Click to expand...
Click to collapse
Thanks for the explanation, feels like i'm not the guy who would need to overclock my nexus 5, its fast enough for me out of the box. Multi-rom sounds attracting tho.
Stock kernel works just fine
Sent from my Nexus 5 using XDA Premium 4 mobile app
m4nu4l said:
Thanks for the explanation, feels like i'm not the guy who would need to overclock my nexus 5, its fast enough for me out of the box. Multi-rom sounds attracting tho.
Click to expand...
Click to collapse
One thing I forgot to mention are also the sound tweaks!
There are a LOT out there that are really good, although the App Viper4Android will always reign superior!
I do recommend you have a look at some of the kernels.
You can get 150% out of your battery with the right rom/kernel combo and probably more if you sacrifice some performance.
Just saying it's something a lot of people don't experiment with, I personally am a flash-aholic and need to flash new kernels and roms every other day!
Valdorous said:
I do recommend you have a look at some of the kernels.
You can get 150% out of your battery with the right rom/kernel combo and probably more if you sacrifice some performance.
Click to expand...
Click to collapse
not true.
battery life is dependant on how you personally use your phone, how you set it up, what apps you use, and the quality of your phone/data connection. everything else, including kernels, play a very minor role in battery.
simms22 said:
not true.
battery life is dependant on how you personally use your phone, how you set it up, what apps you use, and the quality of your phone/data connection. everything else, including kernels, play a very minor role in battery.
Click to expand...
Click to collapse
Not necessarily, depending on the govenor (that comes with the kernel) your phone can idle better and some kernels have adjustments to the LMK (LowMemoryKiller) which handles background apps differently.
So in my opinion with the same usage and different handling of background apps, including undervolting and a battery saving governor.
You could easily hit 150% of the usual battery life,
Kernels play in that matter a HUGE role.
You could even throw in some display tweaks (in the battery's favour).
Valdorous said:
Not necessarily, depending on the govenor (that comes with the kernel) your phone can idle better and some kernels have adjustments to the LMK (LowMemoryKiller) which handles background apps differently.
So in my opinion with the same usage and different handling of background apps, including undervolting and a battery saving governor.
You could easily hit 150% of the usual battery life,
Kernels play in that matter a HUGE role.
You could even throw in some display tweaks (in the battery's favour).
Click to expand...
Click to collapse
first off, your "battery saving" govornor doesnt exist. it would depend on exactly what you are doing with the phone. the governor doesnt save battery, it tells your cpu how to behave with the incoming need. and they dont control how your phone idle. i use ondemand, and i doubt many phones idle better than mine.
undervolting DOES NOT save you battery, it can even use more than if not undervolted. the only benefit you will get from undervolting is a little less heat.
and again, kernels do not play a role in battery. if they did, every single person would be using the same battery saving kernel. but we are not. with every single kernel someone will have great battery, and someome will have terrible battery. that's because how ypu use it and your quality phone/data connection pre-decides what your battery will be like.
what you are believing are rumors. stuff people dont really know much about, but still keep repeating the non truth over and over. in reality, kernels have almost no influence in battery.
and btw, everything you can control with a custom kernel, you can with the stock kernel. and people get just as great battery on stock than on a custom kernel.
and yes, i do know what im talking about, ive been part of trinity kernel for 5 years now. and ive helped thousands with android and android related issues, including battery.
simms22 said:
first off, your "battery saving" govornor doesnt exist. it would depend on exactly what you are doing with the phone. the governor doesnt save battery, it tells your cpu how to behave with the incoming need. and they dont control how your phone idle. i use ondemand, and i doubt many phones idle better than mine.
undervolting DOES NOT save you battery, it can even use more than if not undervolted. the only benefit you will get from undervolting is a little less heat.
and again, kernels do not play a role in battery. if they did, every single person would be using the same battery saving kernel. but we are not. with every single kernel someone will have great battery, and someome will have terrible battery. that's because how ypu use it and your quality phone/data connection pre-decides what your battery will be like.
what you are believing are rumors. stuff people dont really know much about, but still keep repeating the non truth over and over. in reality, kernels have almost no influence in battery.
and btw, everything you can control with a custom kernel, you can with the stock kernel. and people get just as great battery on stock than on a custom kernel.
and yes, i do know what im talking about, ive been part of trinity kernel for 5 years now. and ive helped thousands with android and android related issues, including battery.
Click to expand...
Click to collapse
I know and have used your kernel.
If you noticed I said "In my opinion". No reason to get aggressive.
I have noticed something else.
And thought it worth sharing.
If I've offended you, I apologize.
Valdorous said:
I know and have used your kernel.
If you noticed I said "In my opinion". No reason to get aggressive.
I have noticed something else.
And thought it worth sharing.
If I've offended you, I apologize.
Click to expand...
Click to collapse
oh gosh, please dont apologize. i also meant no offence to you.
its not you i am aggressive with btw, its that false info that just wont die. i had still many people do, but its not as bad as before. you dont need to wipe anything when flashing kernels. but, many people used to insist that its a must. i spent a lot of time fighting that falsehood, now i dont see it as much. its the same about battery. if there was a "magic" pill, everyone would be using it. but just as many people using any random kernel will get great battery life on a "battery" saving kernel, and just as many people will get bad battery life as well.
please, again, i didnt mean offence. im just trying to pass the truth around, thats all. its just that this kind of thing isnt an opinion, there is a truth and a false here. id never go against any persons opinion if it was just about opinion.
Yeah....at best....just changing from one kernel to another MAY affect your battery about 5 to 10 percent one way or the other. In other words.....the direct affect on battery is minimal.
The biggest myth in the threads are that rom/kernel combos make big differences in battery life.
Simply put, they don't.
And everything @simms22 said. ?
---------- Post added at 07:59 PM ---------- Previous post was at 07:57 PM ----------
And in modern devices anyway....undervolting does nothing but keep heat down a bit.....and under clocking has an extremely minor affect as well.
Don't believe, try it and see. ?
I get the same battery life on any ROM....any kernel.
You can make your own ultra powersaving mode or your own hyper performance mode...
I'm just using custom kernel for gamma control, s2w/s2s (which is very useful for me when reading some articles on Chrome), also for USB OTG mounting. Oh oh, and also for killing time, you know messing up with frequency and stuff lol
Well, personally I never believe about the effect on battery life by using this or that kernel since there are lots of things that can affect your battery life, and for my case data connection is the battery killer :angel:
What are you guys talking about??. No, I'm not a kernel dev BUT I have tried so many ROM's and kernels on different devices for a long time now and by experience I can say that they DO influence battery life. I have seen %~50 increase (compared to stock) and aso have experienced just 4 hours of battery because a buggy kernel patch.
They're not placebo when your device start to last 19 hours instead of just 13 like it did since you bought It.
lovetatfitties said:
What are you guys talking about??. No, I'm not a kernel dev BUT I have tried so many ROM's and kernels on different devices for a long time now and by experience I can say that they DO influence battery life. I have seen %~50 increase (compared to stock) and aso have experienced just 4 hours of battery because a buggy kernel patch.
They're not placebo when your device start to last 19 hours instead of just 13 like it did since you bought It.
Click to expand...
Click to collapse
your cpu is set up differently with the custom kernel. thats why. its not the kernel itself. you can set up your cpu on stock so itll be like the custom kernel. its very easy if you are familiar with a few scripts and the root filesystems. for example, i run trinity kernel. with the way trinity kernel sets up my cpu, i get better than stock kernel in battery. trinity set my cpu to run all four cores always on, no hotplugging. it also disables mpdecision. i can set up the stock kernel that way as well. i can do it using a root file explorer, manually, or i can do it via script. and i do know people that run the stock kernel that just get fabulous battery life.
---------- Post added at 10:21 PM ---------- Previous post was at 09:55 PM ----------
there are things that you can do with kernel control apps as well.
Sometimes when you flash a kernel...and many times we flash a ROM right before.....you are only getting rid of a battery issue you previously had before you flashed. So with the clean start, of course you're battery will be much better (especially if you had an unfixed or unnoticed issue before)....and sorry to put it this way, but don't know how else to put it....but the less knowledgeable people think its the new kernel that's made such a huge. Pretty much all the most experienced people know kernels don't make a very big difference to battery life. You can flash clean stock with stock kernel and see a huge gain too if you had an issue before.
Test it yourselves. Flash your favorite ROM and some kernel one day. Go a few battery cycles without changing much or installing much. Then flash the same ROM and another kernel. Same changes and same few apps. And so on...
Eventually you'll get the point.
KJ said:
Sometimes when you flash a kernel...and many times we flash a ROM right before.....you are only getting rid of a battery issue you previously had before you flashed. So with the clean start, of course you're battery will be much better (especially if you had an unfixed or unnoticed issue before)....and sorry to put it this way, but don't know how else to put it....but the less knowledgeable people think its the new kernel that's made such a huge. Pretty much all the most experienced people know kernels don't make a very big difference to battery life. You can flash clean stock with stock kernel and see a huge gain too if you had an issue before.
Test it yourselves. Flash your favorite ROM and some kernel one day. Go a few battery cycles without changing much or installing much. Then flash the same ROM and another kernel. Same changes and same few apps. And so on...
Eventually you'll get the point.
Click to expand...
Click to collapse
That right!
It's all about how you feel about your devices.
simms22 said:
your cpu is set up differently with the custom kernel. thats why. its not the kernel itself. you can set up your cpu on stock so itll be like the custom kernel. its very easy if you are familiar with a few scripts and the root filesystems. for example, i run trinity kernel. with the way trinity kernel sets up my cpu, i get better than stock kernel in battery. trinity set my cpu to run all four cores always on, no hotplugging. it also disables mpdecision. i can set up the stock kernel that way as well. i can do it using a root file explorer, manually, or i can do it via script. and i do know people that run the stock kernel that just get fabulous battery life.
---------- Post added at 10:21 PM ---------- Previous post was at 09:55 PM ----------
there are things that you can do with kernel control apps as well.
Click to expand...
Click to collapse
Ok man whatever, I don't like to argue about the color of the sky so.. Yeah..
Also, I think you should e-mail all the kernel devs and tell them that they're full of bs and that they all are wasting their time improving nothing. You should try to reach Linus Torvalds and tell him that he has to stop developing worthless software, that we all can mod our systems the way we want'em to behave and that we don't need support for newer technology. :good:
lovetatfitties said:
Ok man whatever, I don't like to argue about the color of the sky so.. Yeah..
Also, I think you should e-mail all the kernel devs and tell them that they're full of bs and that they all are wasting their time improving nothing. You should try to reach Linus Torvalds and tell him that he has to stop developing worthless software, that we all can mod our systems the way we want'em to behave and that we don't need support for newer technology. :good:
Click to expand...
Click to collapse
It's a good job YOU know what you're talking about, otherwise this thread would have been worthless
The point is you can do some customizations with custom kernels. Gamma control, usb fast charge, etc etc.
But battery life? I never see the difference even with the similar usage that I always do every single day, really, I'm living in Indonesia and the data connection is the battery killer that makes me swearing aaaaall the time lol
I have tried that kernel this kernel (except caf kernel), you name it, but still nothing different, except each of them has their own characteristics.
My own solution? I bought a powerbank, that's my magic pill when my phone is out of battery.
Please, don't think I'm bashing kernel devs out there, they do fantastic job with their creations and I hope they don't stop doing that, I myself using ElementalX, because I need the features that the dev offers to user like me.
Sorry if my English is that bad :/
lovetatfitties said:
Ok man whatever, I don't like to argue about the color of the sky so.. Yeah..
Also, I think you should e-mail all the kernel devs and tell them that they're full of bs and that they all are wasting their time improving nothing. You should try to reach Linus Torvalds and tell him that he has to stop developing worthless software, that we all can mod our systems the way we want'em to behave and that we don't need support for newer technology. :good:
Click to expand...
Click to collapse
Sent from my Nexus 5 using Tapatalk