[n9005][2014/DEC/01][TWKK] faux123/Yank555.lu kernel 002b3

Re: [n9005][2014/MAY/21][TWKK] faux123/Yank555.lu kernel 001

Postby ezaechiel » 22 May 2014, 14:03

yank555 wrote:Great news, thanx for confirming that ;)

JP.


You're welcome :)
Note 3 SM-N9005 32Gb knox tripped and so what?
64Gb FAT32 / CM 11 M7 Snapshot / KK bootloader / TWRP / Nova Prime Launcher / 1Gb zSwap
Yank555.lu kernel CM11-20140526-hlte v1.0b-beta8
ezaechiel
 
Posts: 60
Joined: 14 May 2014, 08:28
Location: Paris
Has thanked: 28 times
Been thanked: 18 times

Re: [n9005][2014/MAY/21][TWKK] faux123/Yank555.lu kernel 001

Postby jdidtht » 22 May 2014, 14:16

yank555 wrote:
jdidtht wrote:
yank555 wrote:[quote="jdidtht"]I've noticed the min cpu freq cycling on this kernel , it never rests, constant one core scaling at idle reguardless if mp is on or off and or gov


You mean scaling_min moves around ? If so, that's normal, that's Samsung's touchboost implementation, it changes scaling_min on touch with 2 levels, one touch = low level (1.2GHz stock) and multiple touches = high (1.7GHz stock). That's a plain in-kernel implementation, and it's hardcoded by default.

CPU Hardlimit has tunables for both steps, they are way too high imho. Sysfs paths should be linked to 1-2 pages back.

If you don't mean that, then I'm not sure to get you right ...

JP.

EDIT : Except if you are on interactive, where that's interactive's hasty way to reach it's ideal freq (or whatever the tunable is called), I've calmed it down in my CM kernel using its tunables ;)

, min never stops moving ever even at idle. Thats not normal. Itss always scaling. Every kernel I've ever used the min freq at rest will stay at rest (300) until its told to do so. This will never rest, it will be 300 1194 1728 300 2226 300 1194 1728 over and over at idle, it should be sitting at 300 while not being used. No touch ,no apps, and constantly up an down. I'll flash the newest build to see if fixed.. Core 0 should be at 300 at idle with screen on and the other three cores will be offline. So the issue is at idle its never actually idling, core 0 is scaling for no reason while the other three cores are offline. Their is no screen touching on my end, its just never ending scaling at idle.

Latest is doing the same for me.


That sounds pretty strange, nothing in the kernel is supposed to make it do that, except touch boot and wakeup kick, which both I have 100% identical in my CM kernel, and scaling_min behaves exactly as expected, 300, touchboost low or touchboost high and on wakeup wakeup kick freq...

Do you have Sammy's ROM side dvfs messing still enabled ??

JP.

Send from my 'proudly eFused' Note 3 (n9005) running Temasek v85 on Yank555.lu htle v1.0b-beta3 (3.4.91) kernel.[/quote]

Thats what I'm saying, it is weird. As for dvfs , yes I do have it disabled, and I mean its basically completely gone in tweaked ROM, including the areas in ssrm.jar as well. This kernel, and one other that Im pretty sure is using portions of your source does the same thing. I basically wanted to see if anyone else could confirm idle freqs at rest or if its just me.

Image
jdidtht
 
Posts: 111
Joined: 27 Oct 2013, 17:45
Has thanked: 80 times
Been thanked: 24 times

Re: [n9005][2014/MAY/21][TWKK] faux123/Yank555.lu kernel 001

Postby yank555 » 22 May 2014, 15:56

jdidtht wrote:Thats what I'm saying, it is weird. As for dvfs , yes I do have it disabled, and I mean its basically completely gone in tweaked ROM, including the areas in ssrm.jar as well. This kernel, and one other that Im pretty sure is using portions of your source does the same thing. I basically wanted to see if anyone else could confirm idle freqs at rest or if its just me.


Just to be certain we're talking about the same thing, we actually are talking about scaling_min (as min. CPU freq.), or are we talking scaling_current (as in CPU's real clock at a given time) ?

Hence if you :

Code: Select all
chmod 444 /sys/devices/system/cpu/cpu0/cpufreq/scaling_min_freq
chmod 444 /sys/devices/system/cpu/cpu1/cpufreq/scaling_min_freq
chmod 444 /sys/devices/system/cpu/cpu2/cpufreq/scaling_min_freq
chmod 444 /sys/devices/system/cpu/cpu3/cpufreq/scaling_min_freq


and those rights stay this way unchanged, userspace won't be able to mess with it.

That's the only way to know for sure where to look, kernel or ROM/apps.

But having scaling min jump around without the device being touched shouldn't come from the kernel, my CM kernel is 100% in line in that regard, and it does not do that. So if this does happen, I'd bet it's coming from userspace.

JP.
Nexus 5 (aka. Hammerhead) 32Gb
Note 3 SM-N9005 32Gb Proudly eFused
SGS3 GT-I9300 32Gb
HTC Sensation XE
HTC HD2
TF300TG 32Gb

Image
Credits FAdrums !
User avatar
yank555
-----------
 
Posts: 3520
Joined: 11 Mar 2013, 20:51
Has thanked: 924 times
Been thanked: 4608 times

Re: [n9005][2014/MAY/21][TWKK] faux123/Yank555.lu kernel 001

Postby Julian Jeremiah » 22 May 2014, 17:22

For this mine's are pretty in good shape; core0 stays on 300Mhz and core 1/2/3 stays offline (just staring on the screen while not touching it)
Julian Jeremiah
 
Posts: 43
Joined: 12 May 2014, 14:57
Has thanked: 21 times
Been thanked: 6 times

Re: [n9005][2014/MAY/21][TWKK] faux123/Yank555.lu kernel 001

Postby jdidtht » 23 May 2014, 02:43

yank555 wrote:
jdidtht wrote:Thats what I'm saying, it is weird. As for dvfs , yes I do have it disabled, and I mean its basically completely gone in tweaked ROM, including the areas in ssrm.jar as well. This kernel, and one other that Im pretty sure is using portions of your source does the same thing. I basically wanted to see if anyone else could confirm idle freqs at rest or if its just me.


Just to be certain we're talking about the same thing, we actually are talking about scaling_min (as min. CPU freq.), or are we talking scaling_current (as in CPU's real clock at a given time) ?

Hence if you :

Code: Select all
chmod 444 /sys/devices/system/cpu/cpu0/cpufreq/scaling_min_freq
chmod 444 /sys/devices/system/cpu/cpu1/cpufreq/scaling_min_freq
chmod 444 /sys/devices/system/cpu/cpu2/cpufreq/scaling_min_freq
chmod 444 /sys/devices/system/cpu/cpu3/cpufreq/scaling_min_freq


and those rights stay this way unchanged, userspace won't be able to mess with it.

That's the only way to know for sure where to look, kernel or ROM/apps.

But having scaling min jump around without the device being touched shouldn't come from the kernel, my CM kernel is 100% in line in that regard, and it does not do that. So if this does happen, I'd bet it's coming from userspace.

JP.


I'll run it, but yes you are understanding now, it shouldnt be scaling, more like cycling at rest with no input happening and I know its not romside here on my end, its only this kernel and one other kernel that is very similar released in the past couple days. I've never had an issue before like this on ANY kernel of yours or faux.. I use faux clock, and regardless of gov or mp disabled etc etc. I'll run the script above and see what if I can hold that min freq at rest. Thanks for hearing me

Image
jdidtht
 
Posts: 111
Joined: 27 Oct 2013, 17:45
Has thanked: 80 times
Been thanked: 24 times

Re: [n9005][2014/MAY/21][TWKK] faux123/Yank555.lu kernel 001

Postby jdidtht » 23 May 2014, 02:44

Julian Jeremiah wrote:For this mine's are pretty in good shape; core0 stays on 300Mhz and core 1/2/3 stays offline (just staring on the screen while not touching it)


Lucky man you are, I'll hopefully sort it out here shortly.

Image
jdidtht
 
Posts: 111
Joined: 27 Oct 2013, 17:45
Has thanked: 80 times
Been thanked: 24 times

Re: [n9005][2014/MAY/21][TWKK] faux123/Yank555.lu kernel 001

Postby yank555 » 23 May 2014, 06:56

jdidtht wrote:
Julian Jeremiah wrote:For this mine's are pretty in good shape; core0 stays on 300Mhz and core 1/2/3 stays offline (just staring on the screen while not touching it)


Lucky man you are, I'll hopefully sort it out here shortly.


Hmm ... if it works on one device it can't be the kernel, it's the same binary code on the same CPU, even if I keep reading people saying some devices are 'divas', but electronics doesn't have a 'soul', same code on same hardware will produce the same result. Only thing there is a difference in hardware is stability to OC / UV, because the CPU crashes, but that's something else.

So the difference 'in code being executed' can only lie in ROM and/or apps ...

Have you tested a Nandroid then full wipe, clean ROM flash to see if it still happens ?

JP.

Send from my 'proudly eFused' Note 3 (n9005) running Temasek v85 on Yank555.lu htle v1.0b-beta3 (3.4.91) kernel.
Nexus 5 (aka. Hammerhead) 32Gb
Note 3 SM-N9005 32Gb Proudly eFused
SGS3 GT-I9300 32Gb
HTC Sensation XE
HTC HD2
TF300TG 32Gb

Image
Credits FAdrums !
User avatar
yank555
-----------
 
Posts: 3520
Joined: 11 Mar 2013, 20:51
Has thanked: 924 times
Been thanked: 4608 times

Re: [n9005][2014/MAY/21][TWKK] faux123/Yank555.lu kernel 001

Postby ezaechiel » 23 May 2014, 09:19

Hello,

I finaly change my rom for a CM11, with your kernel yank, all is working fine and very smooth :), thanks and I go to the CM Kernel forum :).
Note 3 SM-N9005 32Gb knox tripped and so what?
64Gb FAT32 / CM 11 M7 Snapshot / KK bootloader / TWRP / Nova Prime Launcher / 1Gb zSwap
Yank555.lu kernel CM11-20140526-hlte v1.0b-beta8
ezaechiel
 
Posts: 60
Joined: 14 May 2014, 08:28
Location: Paris
Has thanked: 28 times
Been thanked: 18 times

Re: [n9005][2014/MAY/21][TWKK] faux123/Yank555.lu kernel 001

Postby cmvcarlos88 » 23 May 2014, 09:24

Julian Jeremiah wrote:For this mine's are pretty in good shape; core0 stays on 300Mhz and core 1/2/3 stays offline (just staring on the screen while not touching it)

For me works exactly as you. If i don´t touch the screen it stays like you. So yank doesn´t need to worry about that.
cmvcarlos88
 
Posts: 1
Joined: 12 May 2014, 09:24
Has thanked: 8 times
Been thanked: 1 time

Re: [n9005][2014/MAY/21][TWKK] faux123/Yank555.lu kernel 001

Postby Julian Jeremiah » 23 May 2014, 18:11

Did you try to tweak the parameters? I've noticed the default sampling_rate and/or timer_rate for intelliactive, ondemand and interactive are too low. Maybe try to raise it? It'd stop erratic CPU frequent scaling.

jdidtht wrote:
Julian Jeremiah wrote:For this mine's are pretty in good shape; core0 stays on 300Mhz and core 1/2/3 stays offline (just staring on the screen while not touching it)


Lucky man you are, I'll hopefully sort it out here shortly.
Julian Jeremiah
 
Posts: 43
Joined: 12 May 2014, 14:57
Has thanked: 21 times
Been thanked: 6 times

PreviousNext

Return to Kernels

Who is online

Users browsing this forum: No registered users and 12 guests

cron