I think that this is future of our x10 on CM9, 100% faster than AOSP ICS code. Come on FXP developers, bring it to us.
blog.gsmarena.com/android-ice-cream...nks-to-linaro-the-future-looks-even-brighter/
What do u think about this?
Sent from my "z10i"
BTW not all patches will be merged
http://forum.xda-developers.com/showpost.php?p=27308375&postcount=35875
Yap, but Linaro is based on CPU optimisation, not on GPU, and nightly CM9 with Linaro code for Galaxy Nexus work faster than the one with AOSP code.
Sent from my "z10i"
neoxx3m said:
Yap, but Linaro is based on CPU optimisation, not on GPU, and nightly CM9 with Linaro code for Galaxy Nexus work faster than the one with AOSP code.
Sent from my "z10i"
Click to expand...
Click to collapse
Big part of it be integrated into drivers
No driver sources = no implementation
Galaxy nexus have them(and its nexus devices only which is having all sources)
BTW its won't increase much visible performance and mainly benchmarks are improved
Its proven by AOKP team(if i remember correctly it was AOKP team who released linaro tweaks and toolchain based rom)
Anything newer is better but many stuff can't be implement
Linaro team is working on code for all devices with ICS. On video in first post Linaro ICS is running on SoC same as Motorola RAZR. Find interview with Linaro team leader on youtube and watch it. I think that future of Android and upgrades for devices based on Android is in Linaro optimisations. I hope so that Google will see it and hire those guys on Android development.
Sent from my "z10i"
You don't get it...
Its like you made the cake but forgot to add sugar, now you need to remake the cake but you don't have the ingredients
Razr is having same SoC as galaxy nexus which is open source so they have the ingredients
http://forum.xda-developers.com/showthread.php?p=27323815
Read last post on that page
Sent from my GT-I9100 using Tapatalk 2
Linaro code in 125build of FXP CM9. Is it true? I hope so.
Sent from my "z10i"
neoxx3m said:
Linaro code in 125build of FXP CM9. Is it true? I hope so.
Sent from my "z10i"
Click to expand...
Click to collapse
they removed it due to incompabilities.
THeLogiC said:
they removed it due to incompabilities.
Click to expand...
Click to collapse
who wonders ?
Maybe in 126build.
Sent from my "z10i"
basically i've heard a lot of **** regarding linaro...
i have compiled kernels using linaro, gnu, android's own, and the old linaro toolchain... and i found NO DIFFERENCE AT ALL in daily use...
linaro is not magic... i would just tell guys to use google and wikipedia a bit and get to know what linaro is ....
it wont make things 100% faster... it's just another toolchain...think of it as another version of compiler (just like gcc 4.2, 4.3, 4.6 all can compile the same file and in 99.9% cases all the three binaries will run exactly the same way.... )
linaro just has some optimization options available... which MAY OR MAY NOT make any major difference to the final product...
i would just say... stop overhyping this linaro stuff....
this is what Roman Birg wrote on AOKP website
a note about linaro (warning, tech mumbo jumbo ahead)
These builds do NOT include the new Linaro optimizations. The optimizations don't increase the speed of Android by 100%, nor do they inject butter. They use a newer toolchain to compile Android, which has more optimizations. The Linaro people have also made the proper changes in Android code to allow builds to be compiled with -O3 instead of -O2 flags (think of it as another pass of code optimizations).
Please do not ask us whether we will include them in the future. We like speed as much as the next guy, but we want to make sure that it doesn't break ANYTHING. When/if it's ready, you'll know.
Click to expand...
Click to collapse
Related
Since I got my hands on Epic 4G optimized dalvikVM and libc.so - posted by no2chem, I couldn't help wondering - what stops us, Captivate developers, from recompiling Samsung source code?
Links
- optimized recompiled from sources Dalvik VM libdvm.so - http://forum.xda-developers.com/showthread.php?t=1017781
- Bionic libc.so, GB based, video player not worked, piss poor battery life, may be main reason for GB poor battery - http://forum.xda-developers.com/archive/index.php/t-1018622.html
bravomail said:
Since I got my hands on Epic 4G optimized dalvikVM and libc.so - posted by no2chem, I couldn't help wondering - what stops us, Captivate developers, from recompiling Samsung source code?
developers.com/archive/index.php/t-1018622.html[/url]
Click to expand...
Click to collapse
Not that I am a Dev, but this is a very interesting thread to watch.. I am sure Dev(s) would be interested in taking a shot at this .....
Bravo Bravomail!!
I'm trying it under Windows
Without much luck. Can't even compile a kernel. Cygwin and Codesourcery are not big friends.
So - anyone?
It is pretty much asking to recompile Platform piece of firmware.
Alternatively - back porting CM7 DalvikVM to Froyo will work
bravomail said:
It is pretty much asking to recompile Platform piece of firmware.
Alternatively - back porting CM7 DalvikVM to Froyo will work
Click to expand...
Click to collapse
I am surprised that there is no Dev take on this... may be they are busy cooking
http://www.zdnet.com/blog/hardware/linux-32-kernel-released/17598
Looks like some nice improvements in network connection latency and process handling for the CPU. Anyone think this may help a bit with UI lag in some instances?
gonna update as soon as I get home
cbstryker said:
http://www.zdnet.com/blog/hardware/linux-32-kernel-released/17598
Looks like some nice improvements in network connection latency and process handling for the CPU. Anyone think this may help a bit with UI lag in some instances?
Click to expand...
Click to collapse
They need to restructure the Android framework for that.
Am I missing something here?
Yes, Android is based of Linux the kernels are not the same. Android forked the Linux Kernel in 2009 when some of Androids "code" wasn't approved to become part of the mainline.
With Linux Kernel 3.3 the gap between the Android Kernel and Linux Kernel will be getting smaller but you simply can't just switch them both.
Unless your just talking about booting Ubuntu on your Nexus S or something? Aha.
You can still build a 3.2 Linux kernel for Android.
So everything that runs Linux will be faster?
Even Ubuntu?
Sent from my Nexus S 4G using XDA App. Developer of brickROM, and OP of XDA Thread of The Year 2011.
can`t wait to see the next 3.2 kernels for android
Hope there is some benefits in power-saving ...
LancerEVA06 said:
You can still build a 3.2 Linux kernel for Android.
Click to expand...
Click to collapse
Sent from my Nexus S 4G using XDA App. Developer of brickROM, and OP of XDA Thread of The Year 2011.
LancerEVA06 said:
You can still build a 3.2 Linux kernel for Android.
Click to expand...
Click to collapse
I understand that but I see little point. 3.3 on the other hand, I can.
There is a great description of the ways to get ICS on Atrix from crnkoj:
crnkoj said:
So lets try to get some light into it, the issue about hw acceleration for the atrix is that, even now that nvidia released proper ICS binaries, those are made for recent kernel versions (post 2.6.39 and 3.x versions), those libs are floating around and are accessible to most people who would want to build stuff with hw acceleration, the problem on the atrix however is we only have 2.6.32 kernel sources from motorola, which in term means all the libs floating around are incompatible and useless. Now there would be two general ways of fixing this: 1. get libs that work with 2.6.32 kernels but are ics compatible (most improbable, except if moto leaks them and is still to lazy to move on from the 2.6.32 kernel) 2. get or make post 2.6.39 kernel sources that have the atrix's proprietary drivers including or rewritten (actually more probable, but still quite low chances, except if moto releases and ICS build for the photon/atrix or someone knows, has the time and will to write these code from scratch for the newer kernel versions). So as you can see this is quite a grim outlook, its still being worked on by community devs as far as my information are up-to-date, but i dont know which aproach they are choosing. There is however a minor glimpse of hope, since the move to integrate android drivers in the mainline/mainstream linux kernel is happening in the 3.3 kernel version (there are 3.2.x versions as of now), so this might be actually the best bet, hoping that most of the atrix proprietary drivers will be supported in it, one could just use the "nvidia libs floating around", another thing is nvidia is maintaining their own tegra2/3 kernel sources, so combined with the 3.3 move to android drivers and nvidia implementing support for tegra into their sources, it might not look so grim for us anymore, but this is all a developing story.
Click to expand...
Click to collapse
Also, there is another comment from one of the CM developers:
OK, since so many people are asking: The reason I'm doing the OMAP devices first is because the fine folks at Texas Instruments have, as before, published their reference code. (and Google's current reference device, the SGN, is an OMAP4, which also helps considerably).
The Optimus 2X (and its TMoUS brother, the G2x) is a Tegra2, and nVidia has, as always, published a total amount of zero useful lines of code; at this point, my time is better employed at getting CM9 off the ground with devices in which I can write code than it would be at figuring out how the hell to support old tegra binaries. It'll happen, but not in the near future.
Click to expand...
Click to collapse
According to this, there are some plans of CM-developers to make some kind of wrapper to use GB tegra binaries for ICS on GB kernel. Great!
Also, a question from me. LG has announced ICS on O2X. Will this release help us? I saw that some developers used O2X GB binaries for Atrix, and it worked even better Can we do the same trick with ICS binaries and get fully working ICS on ICS kernel?
v.k said:
Also, a question from me. LG has announced ICS on O2X. Will this release help us? I saw that some developers used O2X GB binaries for Atrix, and it worked even better Can we do the same trick with ICS binaries and get fully working ICS on ICS kernel?
Click to expand...
Click to collapse
You can never be sure on the software world, but I hope so. A Motorola ICS release would definitely be better though.
Interesting information, thank you!
Easier said than done...
But thanks for sharing the info.
Sent from my Atrix 4G using Tapatalk
Hopefully something will get worked out. I have no plans on leaving this phone as I love the FP scanner too much
if i'm correct, samsung's galaxy r and captivate glide are both tegra 2 phones and they run on the 2.6.36 kernel, which supports hw acceleration to a certain degree, does this help us?
Will this help development of ICS and JB ROMS? http://www.engadget.com/2012/09/23/nvidia-to-offer-up-documentation-for-tegra-graphics-core/
This should solve the video camera issues and other issues I think.
It's the documentation for the Tegra chips.
Looks like a good news and a bit of light for ICS+JB for our Atrix. Lets keep waiting...
No idea, but good find nevertheless.
Assuming this is released, what else will be needed to release a stable CM9?
It really depends on whats is going to be released. We don't know if its going to be useful yet.
Sent from my MB860 using xda app-developers app
Yeah it will help even to make better kernels, is a shame that my Atrix touchscreen died totally two days ago (thread reported will move soon, does not belong here).
RAFAMP said:
Assuming this is released, what else will be needed to release a stable CM9?
Click to expand...
Click to collapse
Once Tegra2 graphic driver source is released, Atrix devs then just simply update and compile the expecting module (.ko file) to put into the kernel of their roms
hainguyen273 said:
Once Tegra2 graphic driver source is released, Atrix devs then just simply update and compile the expecting module (.ko file) to put into the kernel of their roms
Click to expand...
Click to collapse
Thank you for your replies, guys!
So, with the graphics drivers into the kernel we would get 100% working roms?
RAFAMP said:
Thank you for your replies, guys!
So, with the graphics drivers into the kernel we would get 100% working roms?
Click to expand...
Click to collapse
Yes.
fviero said:
Yes.
Click to expand...
Click to collapse
I hope it's soon then!
hainguyen273 said:
Once Tegra2 graphic driver source is released, Atrix devs then just simply update and compile the expecting module (.ko file) to put into the kernel of their roms
Click to expand...
Click to collapse
I'm not particularly knowledgable about the ins and outs of Nvidia's modules so I might be wrong here, but I seem to recall a dev on this forum saying that is wasn't just a matter of the binaries being compiled against a different kernel (which could be worked around if it was just version checking), but that it depends upon functionality that only exists in newer kernel versions.
If somebody on this forum were skilled enough/had the time to donate to port the Nvidia 3.1.10 sources to the Atrix, we could have fully functional ICS/JB today. But it's not reasonable to expect the few skilled kernel devs here to make up for Motorola's slack.
Over in the One X forum, richardtrip has ported the reference 3.1.10 kernel using only a few bits of hTC code for the camera, so it is certainly possible for somebody without 'inside knowledge' to do, but it has taken him months of hard work. (we're on 2.6.39 'till hTC's official jellybean drops, which uses a 3.1.10 kernel anyway)
Well.. there is still the matter of the fingerprint scanner beyond the kernel. But we can't get too greedy now can we?
thantos said:
Well.. there is still the matter of the fingerprint scanner beyond the kernel. But we can't get too greedy now can we?
Click to expand...
Click to collapse
The phone would probably be functional without a fingerprint driver, though.
This is not kernel module source they are releasing; they are just saying they are willing to provide more information to the opensource community. NVIDIA has recently indicated they may provide documentation to those working on the nouveau drivers (opensource NVIDIA PC drivers); under an NDA. Meaning that they will provide the documentation to those that work on the opensource driver but not allow the information to be shared. For this to be helpful we would need the documentation provided to someone who is working on opensource Tegra drivers; which I do not believe there is anyone. So this would probably be picked up by those working on the nouveau drivers. Best case if all the information is provided and they decide to work on it it would probably be at least a year before we would have anything stable for use.
thantos said:
Well.. there is still the matter of the fingerprint scanner beyond the kernel. But we can't get too greedy now can we?
Click to expand...
Click to collapse
Yeah, same for the lapdock.
Enviado desde mi MB860 usando Tapatalk 2
Just wondering if anyone knows why we haven't seen a linaro Rom yet?
Is there a special problem that 4.3 presents?
(And for those of you wonderful souls about to write "give them time! Don't you know development takes time! Be more grateful! Etc...", in asking out of curiosity, not out of complaint )
Just interested if there is any special challenge that 4.3 presents!
Sent from my Nexus 4 using Tapatalk 4
linaro..
first of all, when you say linaro, are you asking about the linaro toolchain or the linaro patches? they are 2 different things. what are you asking about?
bongostl said:
Just wondering if anyone knows why we haven't seen a linaro Rom yet?
Is there a special problem that 4.3 presents?
(And for those of you wonderful souls about to write "give them time! Don't you know development takes time! Be more grateful! Etc...", in asking out of curiosity, not out of complaint )
Just interested if there is any special challenge that 4.3 presents!
Sent from my Nexus 4 using Tapatalk 4
Click to expand...
Click to collapse
Maybe this:
http://forum.xda-developers.com/showthread.php?t=2350121
I tried but didn't see significant performance improvement of Linaro. Also 4.3 custom ROM's are still not stable (for me).
Anyway, wait a little bit, you will see linaro builds of nearly all ROM's on the forum.
cewrld said:
I tried but didn't see significant performance improvement of Linaro. Also 4.3 custom ROM's are still not stable (for me).
Anyway, wait a little bit, you will see linaro builds of nearly all ROM's on the forum.
Click to expand...
Click to collapse
Minimus ROM is a perfect stable way to go. PA 3.94 is also stable, although some features are not fully working.
With stable I mean not reboots nor force closes at all, or any other weird behavior.
Most of the AOSP build out there are pretty good and stables.