[Kernel] ICS (3.1 and 2.6.39 versions) (V25.5 and V1.1)

Search This thread

GrijzeGeest

Senior Member
Dec 14, 2010
125
24
Tried the oc1200 for a while, both with setcpu and CPU master pro. Seems to work for the most part (after erasing cache and dalvik cache) but makes the browser unstable: crashes and freezes.

Sent from my Transformer TF101 using Tapatalk
 

Mikey

Retired Senior Moderator
Oct 10, 2007
6,262
1,743
Wheaton, Il
Google Pixel Fold

oifi

Senior Member
Nov 22, 2010
309
118
Vienna
installed the 1200 kernel and max was 1200MHz, but after a reboot it only clocks at 1000MHz max according to cpuspy
 
Last edited:

kxnzxmori

Senior Member
Jun 25, 2011
396
160
Will you add support/compatibility for AOSP roms? Thanks.
smile.gif
 
  • Like
Reactions: dbass2002

Caramel

Senior Member
Little feedback after one day, no freeze, no reboot, wifi and BT are OK

Using no-frills Min 456 Max 1200 Governor On Demand /Noop on ARHD 3.1.0

Something strange, when i submit my poor 5533 antutu score :D i found a Transformer under ICS O/C at 1592 :eek:
 

j3tt3

Senior Member
Feb 21, 2012
340
49
After one day, I have no reboots,battery drain nor sods even switching from stock to arhd to revolver. 1.2mhz is stable on my TF.
Next level pleaze... 1.4 or1.5mhz



Sent from my Transformer TF101 using xda premium
 

SiNJiN76

Senior Member
Sep 14, 2009
3,993
626
Just tried this about 30 minutes ago. OC version. Hot Reboot. While reading email. FYI...my TF is stable up to 1.544 or so.
 
Last edited:

AllanJacques

Senior Member
Nov 27, 2007
593
51
41
Vancouver
For me it is a no-go, as far as I flash this kernel TF hangs at asus screen, if I manage to turn it off and on again it boots but hangs as soon as system loads...


Sent from my Transformer TF101 using Tapatalk
 

kevonomics

Member
Jul 27, 2011
19
4
I'm going to try wiping caches because when I flash the kernel SetCpu nor Voltage Control show me the possibility to clock it at 1200 mhz.

I currently have that bootloop CWM problem that I should fix through ADB.. So I cold linux boot it after flashing. That could be it.

What app are you guys using? Any other way to fix bootloop without ADB, I am not around any free using computers most of the day to fix it.
 

Danzano

Senior Member
May 21, 2011
3,075
463
Auckland
Little feedback after one day, no freeze, no reboot, wifi and BT are OK

Using no-frills Min 456 Max 1200 Governor On Demand /Noop on ARHD 3.1.0

Something strange, when i submit my poor 5533 antutu score :D i found a Transformer under ICS O/C at 1592 :eek:

Lol was the nickname danzano with a picture of auckland city if so that was me the other day testing blades kernel but had to go back to stock after random reboots :D

Sent from my Transformer TF101 using XDA Premium HD app
 

gVibeR

Senior Member
This is a kernel on which I am doing some tests.
As I not found the source of the latest Blades kernel and wanted to do some testing, I used the asus base source and made ​​my own modifications.

My tf101 suffer of restarts and sleeps of deaths but as do not occur as a controlled manner, I don't have clear whether there will be improved with this kernel (for the moment I have none), so if someone wants to try, can comment how it works.

I use Revolver rom (with some custom modifications). The kernel is only tested with this rom, but should work on other (supports init.d).

At the moment I have two versions, one with overclocking limited to 1.2Ghz and one without overcloking.
The issue that max frequency is set (although we configure a lower frequency ) continues, resulting in more restarts if the frequency is very high, so I have limited to 1.2.

Clemsyn's kernel sources are right here: https://github.com/clemsyn/android-tegra-nv-2.6.39-rel-14r7

Not sure that Blades is even the latest, but Clemsyn's are.
 

guevor

Senior Member
Aug 20, 2010
361
1,104
Valencia
I have updated the first post with a new test version.
Also created a github repository.
The new version can be overclocked upto 1640.
I also increased the voltages a bit, because they were too low and could cause instability.
The problem with frecuency limits was also solved, thanks to Roach2010 fix.

I'm thinking of adding another version that boot with the upper limit to 1.4Ghz and the min at 456Mhz and using ondemand to test if is more stable as Caramel suggests.
 

msticninja

Senior Member
Mar 16, 2007
731
174
I have updated the first post with a new test version.
Also created a github repository.
The new version can be overclocked upto 1640.
I also increased the voltages a bit, because they were too low and could cause instability.
The problem with frecuency limits was also solved, thanks to Roach2010 fix.

I'm thinking of adding another version that boot with the upper limit to 1.4Ghz and the min at 456Mhz and using ondemand to test if is more stable as Caramel suggests.

Lmao. We must have been working on this at the exact same time. I'm looking at your github, and it says we both pushed our last commits exactly an hour ago. Nice work, nice work. :)
 

guevor

Senior Member
Aug 20, 2010
361
1,104
Valencia
Compatible with cyanogenmod?
I can not say because I have only tested with Revolver, but some previous comments indicated that on AOSP roms failed things like wifi.

Lmao. We must have been working on this at the exact same time. I'm looking at your github, and it says we both pushed our last commits exactly an hour ago. Nice work, nice work. :)

Best two people doing something interesting that none.
On Monday I did a lot of tests and traced the problem to cpufreq.c, but today when I came to have time to do more test, I looked before other cpufreq.c versions and found Roach2010, he did it before.


I updated the first post with a test3. The previous version has bad 3d clock that caused bad 3d performance, sorry.
 

msticninja

Senior Member
Mar 16, 2007
731
174
I can not say because I have only tested with Revolver, but some previous comments indicated that on AOSP roms failed things like wifi.



Best two people doing something interesting that none.
On Monday I did a lot of tests and traced the problem to cpufreq.c, but today when I came to have time to do more test, I looked before other cpufreq.c versions and found Roach2010, he did it before.


I updated the first post with a test3. The previous version has bad 3d clock that caused bad 3d performance, sorry.
I agree. Two interested, but probably tired, people.
Same timeline for me. Started working on it on monday, then I saw Roach update his git, so I started over with a merge from his ics branch.
Good luck, I'll keep an eye on your dev. I probably won't be doing a lot in the next few days, but feel free to fork my git or pull changes if you see something you like. Not much changed yet, though.
 

SWEagle

Senior Member
Nov 16, 2011
321
62
Machico
I'm thinking of adding another version that boot with the upper limit to 1.4Ghz and the min at 456Mhz and using ondemand to test if is more stable as Caramel suggests.

That would be great :D but don't forget to low the voltages of the 456Mhz to the lowest as you can... or, as i requested earlier, leave us the option of managing our mV...

I've been using the 1.2Ghz since it came out. Great Kernel. No reboots and i think Wi-Fi is more stable...

Thanks
 

hachamacha

Senior Member
Jun 15, 2010
1,368
751
Honor 8
Nice job. Thanks for taking this on and setting up git for the changes.

I've had ICS on since it arrived, and for whatever reason haven't been unlucky with it, so haven't had a crash yet (~3-4 weeks).

I just tried your test3 kernel, only for the last hour using 1.2G as default + interactive gov and a few profiles. It seems solid to me at this speed. I think my box can only manage ~1.5 but to spare the battery I'll keep it here.

Another thing I like is that this is the first batch of modules I've seen built for ICS that seem to all work properly. I'm sure someone has done that before but they've caused me troubles till now.

Thanks again - Hashi

PS: Eventually I suppose ICS will start crashing for me like everyone else, so will report back if so.
 
  • Like
Reactions: mrirvs

Top Liked Posts

  • There are no posts matching your filters.
  • 141
    Initially the idea of ​​this kernel was to test some things to see if it improved stability. I would have liked to start with the kernel of Blades, but I don't found the source of the ICS version, so I started from the official source of asus.
    With the versions, more people used my kernel, so the initial idea to make a kernel for my personal needs has become more generic, so to some extent, try to apply the suggestions that other do.

    I must make clear that this kernel does not exist without the work of others and thanks to free software. Just trying to be a bit, over a huge pyramid.
    Thank Blades, Roach2010, clemsyn for his work and all who have commented for your help.

    If you want to donate, there is a button on the left and a link below. Do not feel obligated at all and only in gratitude for my work if that has been useful to you.

    Donation Link

    In this post I left the files of old versions.

    Github repository: http://github.com/daliguro/tf101-kernel-test
    117
    [Kernel] ICS

    Changelog:

    Test2 version with 1.64Ghz Max Overclocking. I have increased the voltages (the previous 1.2Ghz was very low). The problem with the frecuency limits was solved by Roach2010. Give thanks to him. No need to have several kernel with different frequencies.

    Test3 Fix bad 3d Clocks on test2. md5 ce39c26f037d9721a5f1071a55a02456

    Test4 Added support to lowering the voltage. Works well with Pimp My Cpu (no displacement). md5 006914bcc531e092bab3e127d6fc7797

    Test5 Increased voltage at higher frequencies (1.5Ghz and above). Some clocks changed (Perhaps best performance). md5 0d0a76d8c04e8c9c19807286666c6cee

    Test6 1.7Ghz Max Frequency, added SIO, CFQ and BFQ i/o schedulers. Boot at 1.2Ghz. I do not recommend work at 1.7Ghz

    Test6b Fixed 1.7Ghz clocks. md5 3025f2dd99e9f5f818606ad03204e6f2

    Test7 - Rebased on V9.2.2.3. md5 47cd318fe6872b665bf96c7d31db4eb0

    Test8 - Fix for stability and 1.7Ghz voltage increased. md5 e8f513de01755cf875ad083a4aa9ec88

    Test9 - Backports from Linux 3.1, added cifs and xpad (xbox) modules. Having introduced many changes, it may be much more problematic than previous or perhaps more stable if all went well. Take care. md5 417f587b0bfba9d7c55f227b10d746ea

    Test9b - Wifi now seems to work on Revolver and CM9. Please test on others and report. md5 8b31d1c03425c3df5efb771ec26f598c

    Test10 - Backlight code changed. More backports from 3.1. Wait for others to report if it works correctly if are not willing to endure problems. md5 06bcf3b654a728039b161ade070c6a98

    Test11 - HAS A BUG. A lot of minor changes, and some cpu errata fix (more details). md5 d3f5ea57e5b57e504de3fd63d94e8f3b

    Test11b - Bugfixed. md5 2dd7130491e750e5230f03e058154047

    Test12 - Added governors wheatley, SavagedZen, lazy, lulzactive and lagfree. It should be more compatible with various roms (please test wifi and BT). md5 9e6a712d5937dbbcc528eb8db50722b9

    Test12b - Changed voltage/frequency tables and added 1640. You need to adapt your undervolting settings. md5 8836aa5ab48cadff9b49d29156597fca

    Test12c - Fixed a problem with undervolting. I had forgotten to fit the new frequency table. md5 6dcf20cbe608365f0bd862c16ad0a654

    Test13 - Another big changes. The most important on the suspend code where reside the principal factor to random reboots. May be .... md5 865aa7db4f8b6625af444adf701841ec

    Test13b - Frequency/voltage tables of test11. Only for people that have more problems overclocking with test13. md5sum 88557afa3c52db356576f0a99f5d8cf4

    Test14 - More backports and fixes. md5 b0780e56ab2b65d311b5a3d16ee53943

    Test15 - Usb working again, sd card and more fixes. md5 f80642e9d3ae405edd0ea4f6e34a1e4a

    Test16 - Changes on timers and other fix to SOD. Compiled with VENO. md5 392b66ea63dca265d391992e8af0c34f

    Test16b - Modified SOD fix to be faster when not docked. Other optimizations. md5 e3f1450caf982af040d278d3c4058ed5

    Test17 - Different SOD fix, updated with 9.2.1.21 asus source and others. md5 7bf83bbf6f7d68de2814d88b5c24ed05

    Test17b - Small modification to try to handle better when can not suspend and when he wakes up alone. md5 73c4e80a29532552807790b330d4a70c

    Test17c - Reduced consumption if you use a microsd all time (like me). Changes in the configuration of bcmdhd. Usb serial changes may affect modems. md5 f9163588e10ba570870a1d83f4a9046d

    Test18 - Several changes on suspend,pm,dc,timer,usb,mmc. Reverted some changes, included 17c ones. Will test one by one to be sure can be re added. md5 76c181b30886aa9ae22f8ba2ff80d5dc

    Test18b - Reverted a change that seems unstable. md5 8a8b989a85b9622f7ea5b8501d6e6b6d

    Test18c - More changes reverted and a new try with suspend and usb. md5 106626e6dee091d3ccc28e58da6adee6

    Test18d - Minor delay change. md5 f08a8bddab9ec8e78618f070af1714c6

    Test18e - Reapply some v18 changes and suspend code. md5 3c2ec029133c274746cbf2a56f1e6c90

    V19 - Eliminated the changes that this time I hope were the bad and re-added the good. md5 66e4f9be8650cb7676c2748578f26c82

    V19.1 - Fixed a bug. md5 aa67a380b3ccc77afd791f267de435ab

    V19.2 - Another attempt to return to stability. md5 34bc9721647c8a316b5c352108ef0a22

    V19.3 - Changed compilation options, suspend code and some modem options. md5 2217da228d9323aae42b835735a7ba4c

    V19.4 (#504) - Fixed a problem that can cause to not deep sleep. md5 bd08cd3cc2a20829d69f00b302e6f870

    V20 (#505) - Try to avoid ram fragmentation and new compilation options for better performance. md5 a2a43d5eac6cca6f1045d90925532a11

    V21 (#527) - Usb problem fixed, iptables problem fixed and i2c changes. md5 0f3e339ff7157850619cea19ec8a772c

    V21.3 (#559) - Fixed some stability problems introduced. md5sum 0758f26e3ef9e56f25a9b2a77098c9c1

    V21.4 (#564) - Minor graphic drivers change. md5sum a6f6a4a9aaaf464c5b02a087b48b0cf7

    V22 (#607) - More graphic drivers changes, ram timings, compiler and optimization options, pci drivers and other bugfixes. md5sum 0635a12362b47a5798fc8f48cd4439c4

    V22.6 (#645) - More performance and several fixes.

    V23 (#689) - Some memory, video drivers and wifi fixes

    I created a new branch based on original 2.6.39 code. Use it on TF101G or if have problems with the 3.1 version.

    V1.1 (#28) - Initial version. CPU and GPU overclocking and other performance changes.

    Current default voltages: 1.7Ghz 1475mv - 1.64Ghz 1450mv - 1.6Ghz 1425mv - 1.5Ghz 1350mv - 1.4Ghz 1275mv - 1.2Ghz 1100mv
    I currently use voltages: 1.6Ghz 1425mv - 1.5Ghz 1350mv - 1.4Ghz 1150mv - 1.2Ghz 1050mv
    103
    [Kernel] ICS

    Downloads

    For a Megatron Version you can use this link (thanks to ricardopvz)

    The New codecs from Nvidia (video decoding) (NOT INCLUDED ON ANY KERNEL)

    Download this to enable io as busy on interactive governor (better video playback)

    3.1 based

    Kernel-ICS-V22-OC1700-fb_607.zip same file as V21.11
    Kernel-ICS-V22-OC1700-fb-AOKP_606.zip
    Kernel-ICS-V22-OC1700-fb-megatron_605.zip

    Kernel-ICS-V22.6-OC1700-fb_645.zip md5sum cf90635049728aef79c4cf72dfc83b8a
    Kernel-ICS-V22.6-OC1700-fb-megatron_645.zip md5sum 4ad30d5b7f031c09ca46a5d676a20c79
    Kernel-ICS-V22.6-OC1700-fb-AOKP_646.zip md5sum d257aed3b8933003b1b7a7310d30db6b

    Kernel-ICS-V23-OC1700-fb_689.zip md5sum db62b36c6fb110902c57f549232888e0
    Kernel-ICS-V23-OC1700-fb-megatron_689.zip md5sum ffaec5a8646c05c5106d4c8f0543a576
    Kernel-ICS-V23-OC1700-fb-AOKP_690.zip md5sum 8d6051faf4059324a91b51031bce8522

    Kernel-ICS-V25.5-OC1700_741.zip md5sum 7b154df80a3e96fe2f425939c3611898

    2.6.39 based

    Kernel-2-ICS-V1.1-OC1700-fb_28.zip md5sum e8d15d5168587dec00e228dc2a72cf2f
    Kernel-2-ICS-V1.1-OC1700-fb-megatron_28.zip md5sum 7c82713293bc7226a301e60d2dea2796
    Kernel-2-ICS-V1.1-OC1700-fb-AOKP_29.zip md5sum 6fd2a8d8db9dade682114e3c18682d87
    19
    Sorry, but I've been sick and I could not use the computer. The truth is that has been a difficult year and seems the doctors did not match with the solution (the problem is aggravated from time to time).
    I will try to rest a few days and hope to return with more energy to continue collaborating.
    18
    I compiled a new version for CM9 with small changes in the configuration. If someone want to test it, please confirm if it works or not.
    I have also compiled another version for JB eos with best compiler options similar to what I use on my other versions (the former was very similar to the configuration used in eos)
    Sorry I can not spend more time at the moment but hopefully I can get in a few days another version for other JB rom.