This is the generic GPU drivers for adreno GPU found in snapdragon chipsets by Qualcomm.
Although this will not magically make your GPU fast, updated drivers will fix bugs and show improvements all round.
Thanks to @denzel09 who has repacked the may binaries with correct path for Moto G.
Download Link
http://www61.zippyshare.com/v/59501646/file.html
Stock 4.4.4 Adreno drivers (those who wants to go to default)
http://www72.zippyshare.com/v/11705987/file.html
B]Changelogs[/B]
Changelogs for August
@lost101 has posted changelog
Known Issues:
Applications may hang when repeatedly disconnecting and reconnecting Adreno Profiler while plotting GPU Stall Metrics in Grapher.
Capturing OpenCL metrics with Adreno Profiler Scrubber may crash target application.
Fixes and Improvements:
Fix a failure to release performance counter resources in some cases when using Adreno Profiler with certain NDK apps and OpenCL.
Fix a crash when connecting Adreno Profiler to an app that uses the disjoint timer query extension.
Add support to Adreno Profiler for OpenCL kernel type qualifier arguments.
Add support to Adreno Profiler for clCreateImage() 1D image, 1D image array, 2D image array.
Add support to Adreno Profiler for displaying multiple OpenCL queue/device/context combinations.
Fix a potential target application crash when Adreno Profiler metrics are enabled while the GPU is not idle.
Fix for incorrect scissor results if an app calls scissor and glBufferSubdata() between two draw calls.
Fix an issue with binning configuration calculation when MSAA is enabled.
Add support for ASTC texture formats in Adreno Profiler.
Fix potential null-pointer dereference in OpenCL.
Fix rotation issues with glBlitFramebuffer() and glClipPlane() when device is not in native orientation.
Fix for possible crash when using occlusion query while changing render targets.
Fix incorrect Adreno Profiler capture for applications using 2D texture arrays.
Fix potential crash during VAO validation in buffer update calls.
Add Adreno Profiler support for buffer transmission in clFillBuffer() and clFillImage().
Improve performance when invalidating texture attachments.
Click to expand...
Click to collapse
Tested working on[/B]
XT1033 with stock ROM
XT1032 with stock ROM
XT1031 with Stock ROM
Custom ROM people should report back
abhifx said:
The nexus 4 forum is buzzing with new Qualcomm adreno drivers .
Dont know if this is working with Moto G. Maybe a dev can look into it.
also someone has posted zip files for F2FS partition.
thoughts? Link to the post is as below
http://forum.xda-developers.com/nexus-4/general/qualcomm-adreno-gpu-binaries-t2667759
Click to expand...
Click to collapse
This works also on adreno 305?
About f2fs, we have only data in f2fs but i think that these will be in system...
denzel09 said:
This works also on adreno 305?
About f2fs, we have only data in f2fs but i think that these will be in system...
Click to expand...
Click to collapse
since Adreno 305 has feature parity with 320, i was hoping that the same driver should work. i right now cant check this unless i am near my home system.
abhifx said:
since Adreno 305 has feature parity with 325, i was hoping that the same driver should work. i right now cant check this unless i am near my home system.
Click to expand...
Click to collapse
mmm i have some doubt, but i hope ..
abhifx said:
since Adreno 305 has feature parity with 320, i was hoping that the same driver should work. i right now cant check this unless i am near my home system.
Click to expand...
Click to collapse
I installed it. Can't see anything untowards. Antutu about the same result.
lmulli said:
I installed it. Can't see anything untowards. Antutu about the same result.
Click to expand...
Click to collapse
see the zip, i don't think that paths are the same
denzel09 said:
see the zip, i don't think that paths are the same
Click to expand...
Click to collapse
Guys, it seems download link is dead. Can someone re-upload it ?
@denzel09: what do you mean paths are not the same ? Drivers are in different location in moto G ?
Cheers.
Yes they are in different location. You can see them with root explorer.
Can U plz upload zip somewhere? Original link is not working :-/
Sent from my XT1032 using xda app-developers app
Update.
I repacked the zip with correct folders paths.
Last april version.
To verify install osmonitor from PS, go to message tab and see about adreno build date.
http://www39.zippyshare.com/v/58174828/file.html
Tested only on xt1032.
is there any difference after installing it?
No changes in antutu with this new libraries
When you update your nvidia/amd drivers in your pc you will go always to test them with an antutu-like software ? It's not a good point to start i think. If you see the changelog you ll see how many bugfixes i.e., improvements and so on. Drivers should be always updated by the way.
denzel09 said:
When you update your nvidia/amd drivers in your pc you will go always to test them with an antutu-like software ? It's not a good point to start i think. If you see the changelog you ll see how many bugfixes i.e., improvements and so on. Drivers should be always updated by the way.
Click to expand...
Click to collapse
Are you sure this update isn't just for 320 and 330? I'm gonna flash it, I'll report if I notice any difference
Yes, officially only for that i think.
But remember that you can restore a system backup or in next days i can make the same zip with previous older stock drivers
Well, I don't know if it is a placebo or what, but the phone feels actually a little bit responsive and with smoother animations, just that, I'm not saying that it will be a super performance boost though.
denzel09 said:
Yes, officially only for that i think.
But remember that you can restore a system backup or in next days i can make the same zip with previous older stock drivers
Click to expand...
Click to collapse
I can notice the difference
The boot lag bug has reduced (phone lags for a minute in latest cm builds)
Phone is a bit more responsive
Sent from my Moto G using Tapatalk
I am considering flashing these but I still would like some more feedback. Anyone else did it that could contribute with their experience?
Flashed on xt1031. Nothing is broken I am thinking that the content inside of the apps is loading a little quicker such as UI elements. This could be placebo though. As far as antutu goes, no changes really except I saw a new highest FPS in the 3D test I got a 27.8 at one point, which I have never seen.
Sent from my XT1031 using Tapatalk
Related
/!\ disclaimer: this thread is not about a war between android versions, nor a competition of "who has the biggest one". thank you /!\
I just installed an application named Quadrant Standard, it gives good idea of performance of our phone versus different android versions.
Running Froyo 2.2, release 134++, i have a score of 902
(screen capture enclosed, webcam = bad quality sorry)
derefas, running CM6, has a score of 1000
(screen capture enclosed)
CM6 have higher performance because it does not uses HTCSense.
it's interesting, in the version of derefas, to note that we are weak in memory and i/o access versus the other phones.
might be the sdcard latency.
looking at those results, it seems devs have introduced some "lagfix" that boost reactivity of their phones. another thing to investigate.
also, there are some apps to overclock, we could try to overclock to 700 or 800Mhz to see the improvements.
some other benchmark applications here:
http://www.everythingabouttablets.net/2011/05/05/top-6-android-benchmarks/
regarding lagfix, a post in french:
http://forum.frandroid.com/topic/22162-refmaj-22102010-le-point-sur-les-lagfix/
basically it replaces the old filesystem of samsung (RFS) by the well known ext2.
but since we are already using this filesystem, it is not gonna help us here :-D
-r0bin- said:
also, there are some apps to overclock, we could try to overclock to 700 or 800Mhz to see the improvements.
Click to expand...
Click to collapse
I've tried set-cpu 2.2.4 from here, but it doesn't seem to work. On startup, i press 'msm7x27'. It then asks for root permissions, before force closing.
I've also tried overclocking widget, but it says it needs root access, but I'm pretty sure we already do.
Correct me if I'm wrong or some other overclocking app works for you
Thanks devs!
benjanico said:
I've tried set-cpu 2.2.4 from here, but it doesn't seem to work. On startup, i press 'msm7x27'. It then asks for root permissions, before force closing.
I've also tried overclocking widget, but it says it needs root access, but I'm pretty sure we already do.
Correct me if I'm wrong or some other overclocking app works for you
Thanks devs!
Click to expand...
Click to collapse
I tried this too and it didn't work
I guess we don't have the privilege to scale cpu...to make it higher than 600 Mhz or smaller than 600 MHz
you need the "super user" application
it is in froyo release
when you install setcpu, it asks you to grant root access.
it worked for me, i selected "MSM7x27
but it seems the app cannot read the CPU frequency :-/
it shows "0 Mhz"
ok i know why:
our /dev/cpuinfo lacks details.
it shows only this:
# cat /proc/cpuinfo
cat /proc/cpuinfo
Processor : ARMv6-compatible processor rev 5 (v6l)
BogoMIPS : 598.01
Features : swp half thumb fastmult vfp edsp java
CPU implementer : 0x41
CPU architecture: 6TEJ
CPU variant : 0x1
CPU part : 0xb36
CPU revision : 5
Hardware : photon
Revision : 0000
Serial : 0000000000000000
Click to expand...
Click to collapse
to be improved
Hi, i know this feature is disabled in our phone becouse Cotulla tried to add cpu scaling but not finished (he disabled frequency init)... we need to finish his job and enable it!
EDIT:
I compiled kernel with all known fixes, enabled frequency init, so I can confirm about time: FULL working without time forwarding!!! But not know if time fix come from owerclock or by kernel itself, or maybe by gps.conf ?!!
Here is dmesg with owerclocking
that's interesting!
and guess what, stepping for MSM7227 is available
see here:
https://github.com/ladios/htc-kernel-msm7227/commit/935fd0650743f6b68b256d671b2e5446cd90de61
worst a try
and might fix our issues with clock!
I porting ladios 2.6.35, think will be finished tomorrow
munjeni said:
I porting ladios 2.6.35, think will be finished tomorrow
Click to expand...
Click to collapse
you dont need to port this version, it is also available on 2.6.32
https://github.com/ladios/htc-kerne...86e8a995f/arch/arm/mach-msm/acpuclock-arm11.c
Hi, yes I know that, but I need 2.6.35 for Gingerbread becouse if you see that version have new drivers like bluetooth...etc, also have some bugfixes from 2.6.32... we will try, maybe gtt something better tnan 2.6.32!
I finished and now it is in testing... I will upload to git
Here is http://gitorious.org/2-6-35-photon
Hi,
Guys i Found This file : acpuclk_tbl in /sys/acpuclock....
and there are write this values: 245760,480000,600000...
are they the values of cpu speed?
And are they editable to increase cpu speed?
HTCxxxx said:
Hi,
Guys i Found This file : acpuclk_tbl in /sys/acpuclock....
and there are write this values: 245760,480000,600000...
are they the values of cpu speed?
Click to expand...
Click to collapse
yes
......
but...how can i edit the file? it does't change values...can you help me?
HTCxxxx said:
but...how can i edit the file? it does't change values...can you help me?
Click to expand...
Click to collapse
It doesn't work this way
you need a program for that,for ex setCPU available from Market,but CPU scaling feature will be available on next release which is on testing now
tzacapaca said:
It doesn't work this way
you need a program for that,for ex setCPU available from Market,but CPU scaling feature will be available on next release which is on testing now
Click to expand...
Click to collapse
OK...But they are testing MSM 7227 CPU for Mini?
HTCxxxx said:
OK...But they are testing MSM 7227 CPU for Mini?
Click to expand...
Click to collapse
We are testing.
tzacapaca said:
We are testing.
Click to expand...
Click to collapse
True im looking the dedicated Thread...
hurray guys, i managed to overclock our photon!
credits to ladios, the guy who work on htc aria, from which i stole a bit of code
Using Froyo2.2 with HTCSense standard:
Photon original speed: 600Mhz => quadrant standard score = 860/900
Photon [email protected] 768Mhz => quadrant standard score = 1090/1093
it works great!
in addition to that, the power of enabling overclocking is, you can also underclock the CPU to 250Mhz when idle/charging/doing nothing, which dramatically saves your battery life!!
will play with it and commit changes when validated
There's a long-standing bug (or omission rather) in (probably all) the kernels used for the HTC Sensation--a feature of some Qualcomm CPUs apparently--which is that you can independently set the max cpu frequency of both processor cores. Because of the way the CM Performance settings is written, it only sets cpu0 and expects cpu1 to automatically follow suite. This causes the second core to always run at the kernel's default max setting when the screen is on.
For some time now I've been twiddling with a patch for the kernel that would set cpu1(...n) to follow the changes of cpu0. I have not succeeded, and am now entirely fed up with the whole business and will dump my current results here if anyone is inclined.
I register a CPUFREQ_POLICY_NOTIFIER in the mach-msm/cpufreq.c driver and reading the values works just fine, but when I try to set them the second core seems to bog down and the phone slowly freezes.
Please note that I am not a kernel hacker, and not much familiar with Linux so there probably are glaring mistakes and stupid things in here.
This patch is against arch/arm/mach-msm/cpufreq.c and I was hacking at sources from @android1234567 's repo at https://github.com/sultanxda/sultan-kernel-bruce-linaro3
Enjoy or something
I wonder if anyone is going to include this patch. In my eyes its really usefull, but in this thread is no acticity, thats kind of awkward
s0ftcorn said:
I wonder if anyone is going to include this patch. In my eyes its really usefull, but in this thread is no acticity, thats kind of awkward
Click to expand...
Click to collapse
Give it time for a developer to look into it if it's worth it (that depends in the eyes of users / devs) it will eventually be considered.
Far as I understood by default the cpu1 will always max no matter whats defined, is that right?
MidnightDevil said:
Give it time for a developer to look into it if it's worth it (that depends in the eyes of users / devs) it will eventually be considered.
Far as I understood by default the cpu1 will always max no matter whats defined, is that right?
Click to expand...
Click to collapse
Yes, if you go by the CM Preferences GUI. It's possible to set the max frequency by doing
Code:
echo 1 > /sys/devices/system/cpu/cpu1/online && echo <value in kHz> > /sys/devices/system/cpu/cpu1/cpufreq/scaling_max_freq
Or using a init.d script like the one included in some roms, but it's a bit cumbersome
So you need an app like kernel tuner to set each cpu frequency independently is that it?
Sent from my HTC Z710e using xda premium
MidnightDevil said:
So you need an app like kernel tuner to set each cpu frequency independently is that it?
Sent from my HTC Z710e using xda premium
Click to expand...
Click to collapse
Pimp My ROM has been doing fine for me, so yeah, kernel tuner and the alike work properly from my observations.
This is the full patch and the merge error fix, it can be easily merged into AOSP roms with the CM performance built into the settings package. Enjoy
(btw this patch will even work on quad core QCOM devices)
xmc wildchild22 said:
This is the full patch and the merge error fix, it can be easily merged into AOSP roms with the CM performance built into the settings package. Enjoy
(btw this patch will even work on quad core QCOM devices)
Click to expand...
Click to collapse
This is neat, should've known someone would be there before me
I considered solving it through patching the java code, but you'd have to interface it through the sysfs tree, and that leads to a race condition; the cpufreq files only exist in the tree when the cpu is online, so you'd either have to do some kind of wait loop (could be a long time before the cpu wakes up) or somehow force it to come online--something I don't know how to do.
Current state of 3.0 Kernel
WiFi/Thether/p2p - 100%
TouchScreen - 100%
Video Codec/hw 720p/ION stuff - 100%
Display/DSS - 100%
Phone/3G data - 100%
Sound/FM-Radio - 100%
F2FS support - 100%
GPU OC 266mhz - 100%
Bluetooth - 95%
USB - 98%, otg bugged.
Sensors - 98%, Need test all sensors, looks like all sensors works
Camera - 85%, Preview/Photo(Red have random problem)/Panorama works for both lenses. Video rec not work yet.
Overclock CPU- 0%, not implemented.
For CM10,10.1,10.2 < 07.11.13 users:
To install KitKat, need update recovery. Do not forget reboot, after install recovery.
http://defy.bytekiste.de/cm11-nightly-defy+/update-recovery.zip
Tips:
Do not format System and Cache to F2FS!!!
1. To use f2fsl "/system/bin/mkfs.f2fs /dev/block/mmcblk1p25" or format via TWRP (Advanced wipe menu).
2. To restore ext4, in recovery unmount data, mount system, and do via adb shell /system/bin/make_ext4fs /dev/block/mmcblk1p25 or format via TWRP (Advanced wipe menu).
3. If you have 0 bytes free bug, reformat data to f2fs or ext3/4.
4. 1.1 Ghz can be enabled in performance settings.
Click to expand...
Click to collapse
If you liked this project, you can donate
Donate to Quarx
PayPal: [email protected]
---------------------------------------
Donate to Blechd0se
PayPal: [email protected]
Click to expand...
Click to collapse
Download: http://quarx2k.ru/index2.php?dir=buildbot/roms/defy-cm11.0-3.0
Sources of kernel availabe here(branch p-android-omap3-3.0): https://github.com/Quarx2k/jordan-kernel/tree/p-android-omap3-3.0
Changelog:
17.11:
Added timer slack controller
Added dynamic fsync
Added proportional rate reduction for TCP
Added opensource camera driver (mt9p012)
Added syspanic driver for panic_daemon
Added sync framework (together with ION its part of androids graphical "project butter")
Switched to SLUB allocator
Enabled ram console Support
Synced binder to mainline version (~3.10)
Fixed reboot while incoming or outcoming call (fix proximity sensor in general)
Many small performance patches and fixes for better android support all over the place
Logs or it didn't happen! If you want to report a problem, be sure to post a log file!
Here is a small guide when we need which log file and how to get it.
Old Changelog;
Added new extra_free_kbytes tuable (4.4 compatibility)
Added our Interactive Governor from 2.6.32 for additional performance
Added a basic display control driver
Added a dynamic writeback control interface
Enabled cgroups for memory (4.4 compatibility)
Enabled deadline and changed tunables to the same as on 2.6.32
Enabled xattr and fs security for ext3/4
Patched up trim (ext3/4) to the latest mainline version
Switched to XZ compression for a smaller kernel image
Reclaimed some RAM from logger and vram
*Fixed the wakeup delay and the "funny" screen on resume
TODO:
Add Overclock driver
Fix kpanic with pvr and dss (solved?)
Fix sleep problem
How To Build
Code:
git clone https://github.com/Quarx2k/jordan-kernel.git
cd jordan-kernel
git checkout --orphan p-android-omap3-3.0
git pull https://github.com/Quarx2k/jordan-kernel.git p-android-omap3-3.0
export CROSS_COMPILE=/path/to/your/toolchains/arm-eabi-4.6/bin/arm-eabi-
export ARCH=arm
make mapphone_defconfig
make -j12
Sys-Values explanation;
Let me explain some of this features a little bit. 3.0.8 offers completely new opportunities to do some magic stuff.
One thing i developed a while back ago is a dynamic writeback driver with a sys-interface. It uses the earlysuspend kernel api
to flush dirty pages more frequent when in suspend, while display is on it avoids it to give a more fluid user interface.
Can be controlled via the following command;
echo 1 > /sys/devices/virtual/misc/writeback/writeback_enabled
(1 for enabled = default, 0 for disabled)
Click to expand...
Click to collapse
Also new; a basic display driver! As i tried to fix some things in our dss driver i noticed that we have different panel configurations.One is for all defy green and one for defy red lense users. I exposed those values to user space so that they can be changed. The
new default one are those, which are used on the red lense defy. In theory there could be 256 values (0-255), however could only find the
following as useful. This is not the final gamma control i wanted to have, but at least it gives some control. If someone knows the proper
addresses for gamma values, i can add them too
echo xx > /sys/class/misc/display_control/display_brightness_value
(0 will set the screen to almost 0,
9 will set the screen to some kind of blue-washed colors,
31 will set the screen to more warmer colors = default)
Click to expand...
Click to collapse
In order to run 4.4 more fluid, google made a guide and also published some recommended patches. One of them let you change
the VM's behavior to have a minimum amount of free memory. Also controllable over a interface;
echo 8192 > /proc/sys/vm/extra_free_kbytes
Click to expand...
Click to collapse
8192 means 8 MB of free ram (in any case).
Thanks for this information
Gesendet von meinem Nexus 4 mit Tapatalk 2
If I may ask, @Quarx and @Blechdose: which one of these items are the hardest to solve at the moment?
a separate thread was really needed for 3.0.8 kernel.
thank u quarx and Blechd0se
matiasrey said:
If I may ask, @Quarx and @Blechdose: which one of these items are the hardest to solve at the moment?
Click to expand...
Click to collapse
I am guessing it's the Camera?
Honestly, i think its time to stop working on the 2.6.32 kernel and focus on 3.0.x, looking as an atrix user, its definetly the best choice devs could do. Hehe
Sent from my LG-E975 using Tapatalk
Once again, thanks!!!
@Quarx @Blechd0se About the camera... Quarx, you asked in a post somewhere about our opinions on whether you guys should try to set it up with an open-source driver or with the old Moto driver. Maybe you could tell us what would be the advantages and disadvantages of which one and we could even vote if you want us to?
I'm guessing that we could get more out of the camera with an open-source driver, but it would be harder to set-up than the Moto driver?
niksy+ said:
@Quarx @Blechd0se About the camera... Quarx, you asked in a post somewhere about our opinions on whether you guys should try to set it up with an open-source driver or with the old Moto driver. Maybe you could tell us what would be the advantages and disadvantages of which one and we could even vote if you want us to?
I'm guessing that we could get more out of the camera with an open-source driver, but it would be harder to set-up than the Moto driver?
Click to expand...
Click to collapse
its a 5mp camera, so we should go for the old moto drivers, easiest stuff to do right?
Once it works with open source driver for camera, free forever...
Sent from my Nexus 7 using xda app-developers app
I shouted "YES!!" out loud when I saw the commit about headset and microphone fixes! *edit* But the call speaker maybe isn't fixed just yet :/ Still, gotta try the new build when it shows up ^_^
(that's this one - https://github.com/Quarx2k/android_...mmit/8f5a44ca04ba1c005cb1ac2409639b62f6260d4e )
Combined with Blechd0se's stability fixes (especially this one that fixes the "deep sleep death" - https://github.com/Quarx2k/jordan-kernel/commit/f6b6de228a09f46ab7d6eff70f9225f88e7d3515 ) the next build is going to be very usable imho
niksy+ said:
I shouted "YES!!" out loud when I saw the commit about headset and microphone fixes! *edit* But the call speaker maybe isn't fixed just yet :/ Still, gotta try the new build when it shows up ^_^
(that's this one - https://github.com/Quarx2k/android_...mmit/8f5a44ca04ba1c005cb1ac2409639b62f6260d4e )
Combined with Blechd0se's stability fixes (especially this one that fixes the "deep sleep death" - https://github.com/Quarx2k/jordan-kernel/commit/f6b6de228a09f46ab7d6eff70f9225f88e7d3515 ) the next build is going to be very usable imho
Click to expand...
Click to collapse
Yes Quarx is doing a very good job here, still struggling with a last few bugs. Anyway, its progress! Also the huge battery drain should be fixed,
i disabled kmemleak. 3.0 is such a nice base to work with, no need to do hacky things, it just works out of the box.
Next version will have timer_slack controller (its working perfectly). Also many fixes for crashes and memory leaks.
One big thing is still the display. Although wakeup delay is gone for me, dss/pvr crashes sometimes. Its very random.
Also overclocking has two sides; on the one side its much easier to understand, on the other side it doesn't work as expected (can't overclock
to 1200mhz for example). But we have finally a nice voltage/frequency table.
Glad to read the good news, doesn´t seem too far from a usable kernel!
Have you decided what camera drivers you´ll go for?
F* yeah! "isl29030: fix reboot of proximity sensor" commit signed off by Blechd0se
Quarx has been focusing on the in-call sound issue today, so things are looking brighter by the day
*edit* I'm guessing more and more people use the thanks button as the like button
When I wake up in the morning and read Quarx github , appear "Add Micron MT9P012 sensor driver" , so I think we can hope at an camera soon...
OTG!!
In the description it says USB: 98%, "OTG" bugged. Does this mean we can connect pen drives, mouse and keyboards to our defy with this kernel?
mayur2j said:
In the description it says USB: 98%, "OTG" bugged. Does this mean we can connect pen drives, mouse and keyboards to our defy with this kernel?
Click to expand...
Click to collapse
Yes, but with external power.
major9 said:
Yes, but with external power.
Click to expand...
Click to collapse
So it's of almost no use then..
External power? I can conenct mouse/keyboard or flash drives, no need external power.
Welcome to Z-Kernel's thread!
Features
- Base kernel fully updated to linux-tegra-nv-3.1 (bug fixes & performance improvements)
- Simplified board codes
- Cleaned up Acer specific code
Click to expand...
Click to collapse
Changelog
r2:
- Restarted kernel development (r1's features not implemented yet)
- Base kernel fully updated to linux-tegra-nv-3.1 (bug fixes & performance improvements)
- Simplified board codes
r1:
- CPU OC to 1.6 ghz by default (should be stable)
- GPU OC support up to 700MHz (default is 520 mhz)
- Overclocked LP core from 500 to 620 mhz
- Custom user voltage control for CPU and Core components such as EMC, GPU, and LP (faux123)
- Dynamic FSYNC
- Optimized KSM
- Optimized ZRAM
- Compiled using latest Linaro toolchain with optimized compiler flags
- Optimized SLUB and use SLUB by default instead of SLAB
- Glibc memcpy and memmove
- Deadline improvements for solid state drives
- Proportional Rate Reduction for TCP
- Tegra 3 variant display (faux123)
- Optimized swahb32 byteswap helper
- Asynchronous I/O latency improved through removal of plug in do_io_submit()
- allow use of an I/O controller's native max block size
- Optimized ARM RWSEM algorithm
- timer: optimize apply_slack()
- Optimized crypto algorithms
- Optimized AES and SHA1 routines
- LowMemoryKiller fixes and improvements
- Lock performance governor to all 4 cores
- Many scheduler improvements and optimizations
- updated bcmdhd driver (wifi)
- over 450 changes committed, so the above list isn't complete
Click to expand...
Click to collapse
Complete list of changes can be found in the commit log: https://github.com/Ziyann/android_kernel_acer_t30
I'm not responsible if anything bad happens with your device!
This build has been tested on CM11 only, so if you run into something with other ROMs, report it and I will see what I need to change to cooperate with it.
XDA:DevDB Information
Z-Kernel, Kernel for the Acer Iconia A700 and A510
Contributors
Ziyan
Kernel Special Features:
Version Information
Status: Stable
Current Stable Version: R2
Stable Release Date: 2014-07-30
Created 2014-09-23
Last Updated 2014-09-23
Nice, if it's working fine, I hope you will also support the A510/A511
Shreps said:
Nice, if it's working fine, I hope you will also support the A510/A511
Click to expand...
Click to collapse
Of course, I'll see what can I do when I get everything (OC, UV, ...) working here waiting for a tester...
it is indeed slow. even when completely debloated plus stripped of "essential" system components that I personally do not use. I will try this right away. will report back within an hour in this post.
@Ziyan
it boots, runs, everything seems standard. not seeing any performance increases though. same ol' 10 to 15 fps is there anything more you wish for me to test with it? I replaced this tab a while ago with galaxy note pro 12.2 so I can't really use it enough to check for random reboots
Sterist said:
@Ziyan
it boots, runs, everything seems standard. not seeing any performance increases though. same ol' 10 to 15 fps is there anything more you wish for me to test with it? I replaced this tab a while ago with galaxy note pro 12.2 so I can't really use it enough to check for random reboots
Click to expand...
Click to collapse
Thanks, that's good, it means I can continue improving it. I need to check the kernel periodically after a big bunch of commits, because if something gets broken, it can be hard to find what causes it if from a thousand things I'll push a new version and changelog later today or tomorrow.
sounds good. quote this post when it's uploaded, it'll send me am email notification and I can get right on it.
Sterist said:
sounds good. quote this post when it's uploaded, it'll send me am email notification and I can get right on it.
Click to expand...
Click to collapse
Here's a new version. Before testing it, please make some benchmarks with the previous version (mainly 3D), because - if everything works as it should -, the GPU is now working at 520 mhz instead of 416. It should also have CPU overclocking and voltage control support, so please install Trickster Mod, or something like that from Play Store to see if it works (screenshots are welcome). If it works (i'd be a bit surprised lol it was too easy), please run some benchmarks again to see if there's a noticeable difference. I've got 12 pages of commits waiting to be merged, so yeah, if something got broken, I better fix it now :silly:
Ziyan said:
Here's a new version. Before testing it, please make some benchmarks with the previous version (mainly 3D), because - if everything works as it should -, the GPU is now working at 520 mhz instead of 416. It should also have CPU overclocking and voltage control support, so please install Trickster Mod, or something like that from Play Store to see if it works (screenshots are welcome). If it works (i'd be a bit surprised lol it was too easy), please run some benchmarks again to see if there's a noticeable difference. I've got 12 pages of commits waiting to be merged, so yeah, if something got broken, I better fix it now :silly:
Click to expand...
Click to collapse
just woke up and leaving for work in 20 minutes. when I get there and settle in (about an hour and 20min from this post) I'll test it
which benchmark though, antutu?
Sterist said:
just woke up and leaving for work in 20 minutes. when I get there and settle in (about an hour and 20min from this post) I'll test it
which benchmark though, antutu?
Click to expand...
Click to collapse
Yeah, antutu will do, just note the invidual points
Ziyan said:
Yeah, antutu will do, just note the invidual points
Click to expand...
Click to collapse
woah woah... I just went to download the second version and the tab was dead from 77% last night, still warm.
there's a wake problem, screen will turn on one time after booting up but only that one time
and a sleep problem, after manually turning the screen off it will not turn back on (and kills battery very quickly!)
Sterist said:
woah woah... I just went to download the second version and the tab was dead from 77% last night, still warm.
there's a wake problem, screen will turn on one time after booting up but only that one time
and a sleep problem, after manually turning the screen off it will not turn back on (and kills battery very quickly!)
Click to expand...
Click to collapse
Thanks for the report, I'll investigate the problem tomorrow (it's 01:30 AM here :silly. If you could take a dmesg (while turning the screen off, then trying to turn it back on), or last_kmsg, that would be great help! If you don't know how, refer to section 2 and 3: http://forum.xda-developers.com/showthread.php?t=1520508
Ziyan said:
Thanks for the report, I'll investigate the problem tomorrow (it's 01:30 AM here :silly. If you could take a dmesg (while turning the screen off, then trying to turn it back on), or last_kmsg, that would be great help! If you don't know how, refer to section 2 and 3: http://forum.xda-developers.com/showthread.php?t=1520508
Click to expand...
Click to collapse
higher score / lower 3D performance is v1
lower score / higher 3D is v0
recent apps cleared and CPU set to 1400 performance and ROW
no app gives OC/UV options with v1
I don't have access to a pc at work so can't run dmesg properly for this, and I'm having trouble reproducing the sleep crash now
this is probably all exactly what you did not want me to say lol
edit: so, I got it to crash again (yay!) but can't get klast to work (read only fs) I tried mounting various directories as RW but honestly I have no idea which to do... and here's dmesg which may or may not contain what you need.
I know it might be a little early to ask but any chance you could implement Frandom?
Sterist said:
(cut)
edit: so, I got it to crash again (yay!) but can't get klast to work (read only fs) I tried mounting various directories as RW but honestly I have no idea which to do... and here's dmesg which may or may not contain what you need.
I know it might be a little early to ask but any chance you could implement Frandom?
Click to expand...
Click to collapse
Thanks for the detailed report, I think I revert Franco's hotplugging driver for now, seems like it's on a rampage. May try again in the future.
As for the last_kmsg, your first command was correct, it's just that it doesn't always get created.
We'll see about Frandom when we get things working nice
Here's a build with the stock hotplugging driver, report back if it still crashes.
Ziyan said:
Thanks for the detailed report, I think I revert Franco's hotplugging driver for now, seems like it's on a rampage. May try again in the future.
As for the last_kmsg, your first command was correct, it's just that it doesn't always get created.
We'll see about Frandom when we get things working nice
Here's a build with the stock hotplugging driver, report back if it still crashes.
Click to expand...
Click to collapse
ok I let it idle for about 30 minutes (that was enough to crash every time I left it alone at work) and so good so far.
still no OC/UV though
edit... idle 2 more hours and no problems
Sterist said:
ok I let it idle for about 30 minutes (that was enough to crash every time I left it alone at work) and so good so far.
still no OC/UV though
edit... idle 2 more hours and no problems
Click to expand...
Click to collapse
Great, let's see if CPU OC/UV works with this one. If it does, don't forget to run a benchmark, as GPU OC didn't work till now
Ziyan said:
Great, let's see if CPU OC/UV works with this one. If it does, don't forget to run a benchmark, as GPU OC didn't work till now
Click to expand...
Click to collapse
still no CPU OC/UV. also, that 52mhz step has never worked on any ROM or kernel I've tried, that may extend battery life
Sterist said:
still no CPU OC/UV. also, that 52mhz step has never worked on any ROM or kernel I've tried, that may extend battery life
Click to expand...
Click to collapse
Ok, it must work now.
About 52 mhz, it's so low that any small background work would ramp up the CPU to hispeed_freq, so in the end, it would shorten battery life. The same thing happens with Galaxy Nexus if we go down to 180 mhz, which is a lot, compared to 52 mhz
I've added about 200 commits since your last try (this is a great daily activity :silly, the kernel is getting to be on pair with a Nexus 7 kernel soon if we get OC/UV and some more extra things working :highfive:
Ziyan said:
Ok, it must work now.
About 52 mhz, it's so low that any small background work would ramp up the CPU to hispeed_freq, so in the end, it would shorten battery life. The same thing happens with Galaxy Nexus if we go down to 180 mhz, which is a lot, compared to 52 mhz
I've added about 200 commits since your last try (this is a great daily activity :silly, the kernel is getting to be on pair with a Nexus 7 kernel soon if we get OC/UV and some more extra things working :highfive:
Click to expand...
Click to collapse
ok so UV menu now loads but doesn't look to work quite right - not usable. this is the menu I was looking for that did not load before
OC still does not work though
about 52 (51 actually) every time it is selected, after I leave the screen and come back 102 becomes the selected minimum automatically, even if lock frequency at 51
need benchmark? and ty ty your time!
Sterist said:
ok so UV menu now loads but doesn't look to work quite right - not usable. this is the menu I was looking for that did not load before
OC still does not work though
about 52 (51 actually) every time it is selected, after I leave the screen and come back 102 becomes the selected minimum automatically, even if lock frequency at 51
need benchmark? and ty ty your time!
Click to expand...
Click to collapse
At least we're making progress :good: try this one, and also try trickster mod (specific and general menu).
Benchmarks are welcome, those 200 commits should improve performance a bit, though I think most of them can only be seen in real usage (low memory killer, zram, ...)
Ziyan said:
At least we're making progress :good: try this one, and also try trickster mod (specific and general menu).
Benchmarks are welcome, those 200 commits should improve performance a bit, though I think most of them can only be seen in real usage (low memory killer, zram, ...)
Click to expand...
Click to collapse
bootloop, won't get past Acer logo
benchmark is from the last kernel
Hello there, I've been working on this for a few days and I thought I'd finally show it off!
Please note that there is absolutely no graphics, only console output! Also, it's not very fast (only runs at 15fps on a SD625) and it only runs old homebrew (e.g. first libtransistor tests and such).
Discord support: https://discord.gg/fn8VV9T
Source code: https://github.com/Cyuubi/MonoNX
Credits:
Ryujinx Team - For the base of MonoNX, without them this would not be possible! Show them some love over at their GitHub (https://github.com/Ryujinx/Ryujinx).
Xamarin - For allowing us C# developers to program on Android.
Dr.Hacknik - Logo design.
Requirements:
An ARM64 or 64-bit x86 processor
NOTE: If you get a ROM load error, close out of MonoNX and go into app info and turn on and off storage permissions. This should fix the issue.
App download: https://play.google.com/store/apps/details?id=com.cyuubiapps.mononx
Test ROM download: https://drive.google.com/open?id=1cVa5BDn1w6FNb6QEKmch2m6k67t_q5c_
Interesting, I'll be looking forward this project. Thanks.
Great work , but can't load rom.
Pankaj Rai said:
Great work , but can't load rom.
Click to expand...
Click to collapse
NOTE: If you get a ROM load error, close out of MonoNX and go into app info and turn on and off storage permissions. This should fix the issue.
Very nice, BUT, I have a very important question.
Do you use on aarch64 target devices dynamic recompilation or you execute directly the code?
If not, would you mind to give an insight?
VIRGIN KLM said:
Very nice, BUT, I have a very important question.
Do you use on aarch64 target devices dynamic recompilation or you execute directly the code?
If not, would you mind to give an insight?
Click to expand...
Click to collapse
Sadly, Hypervisors aren't really possible on Android from Xamarin... So we translate ARM64 -> MSIL/CIL -> ARM64, this is quite a hackjob but it works however it kills performance. MonoNX also works on 64-bit x86 processors, so the same thing works for this.
Toggling the storage permission back and forth changes nothing for me.
Sent from my crosshatch using XDA Labs
Cyubii said:
Sadly, Hypervisors aren't really possible on Android from Xamarin... So we translate ARM64 -> MSIL/CIL -> ARM64, this is quite a hackjob but it works however it kills performance. MonoNX also works on 64-bit x86 processors, so the same thing works for this.
Click to expand...
Click to collapse
Hmm, that still sounds though like something that CAN be tackled among the line.
Like what if somebody would switch (no pun intended, hahahaha) the codebase to Firebase?
UPDATE: Apparently it works with under 4GB RAM. (probably using swap though)
Hello, the app force close for me (test rom and snake homebrew). I can not launch anything. I tried turn on and off. Doesn't help.
Here's what I get when I load the test ROM.
EDIT: The emulator closes shortly afterwards. I have a OnePlus 5 with Android 9.0.3.
Hope upcoming updates soon.
Can you support google pixel c? The same soc as switch!
Maybe the adaptation speed will be faster
I hope I can play D3 on my phone soon
[email protected] said:
Can you support google pixel c? The same soc as switch!
Maybe the adaptation speed will be faster
Click to expand...
Click to collapse
Yeah then the games can run natively therefore no need for emulation. I would actually buy a pixel C over a switch if it can work. Also what about nvidia shield tv that can run mario galaxy at 1080p
So? Where we can find roms for this emulator?
Cyubii said:
NOTE: If you get a ROM load error, close out of MonoNX and go into app info and turn on and off storage permissions. This should fix the issue.
Click to expand...
Click to collapse
Tried not working , btw m using oneplust 3T (Aarch64)
nintendobuster420 said:
Yeah then the games can run natively therefore no need for emulation. I would actually buy a pixel C over a switch if it can work. Also what about nvidia shield tv that can run mario galaxy at 1080p
Click to expand...
Click to collapse
Any site to download roms from?
What do you think should be the minimum requirements for an Android device to play resident evil 7 using this emulator, considering RE7 is a cloud game for Nintendo switch?
Also, I found its rom whose size is about 40 MB (YES, 40MB).
Only 3 games playable as I see the base project. Need some time but waiting for D3