The folks behind the Linaro open source software project have put a little time into tweaking Google Android to use the gcc 4.7 toolchain. The result is a version of Android that can perform many tasks between 30 and 100 percent faster than the version of Android Google 4.0 Google currently offers through the AOSP (Android Open Source Project).
While it’s not clear whether Google will implement the changes anytime soon, third party developers are already starting to take advantage of Linaro’s optimizations.
The developers behind the popular CyanogenMod custom ROM have started work to incorporate Linaro’s tweaks, and one developer has already compiled his own build of Linaro Android 4.0 for the Samsung Galaxy Nexus smartphone.
Early reports are that it delivers as promised, offering much speedier overall performance — which is pretty impressive, because it’s not like Android 4.0 felt particularly sluggish on reasonably fast hardware such as the Galaxy Nexus.
I’ll be especially interested to see what these improvements offer for devices with slower CPUs, such as the MK802 and other tablets and devices with ARM Cortex-A8 processors.
Sa – weet!
Now waiting for gs2 and transformer prime builds.If they give us these crazy results we’ll finally see some good old Apple stomping!
What the guy in the video says about Google coders is not nice however…maybe it’s time for Google to start taking lessons in proper programming…:) open source FTW!
I don’t think that guy was intended on being nice. You know the saying sometimes the truth hurts and google devs apparently needs awake up call!!!
Waiting for miss Jellybean…Apparently these guys took things to the forefront and didn’t think much of compatibility. Hope specific rom builders can use all that awesomeness( if truly awesome as it seems…) to make me a happier person.
What did he say that wasn’t nice? That Google is more focused on the the ideas and architecture? That’s true. In order for Google to do what Linaro is doing would require a larger Android team than Google currently has. Frankly, they don’t need to do it, since Linaro can do it.
Now *this* is exciting news! Getting twice the framerate on a CPU bound benchmark is a significant speed-up.
Of course, we can hope to these optimizations to be rolled back into the official Android builds at some time in the future.
Intel recently made announcements that they’ve been optimizing the Android software to get speed improvements on x86 hardware. It would be interesting to compare the relative speed improvements, and to see if the two separate optimizations can be combined to give even better performance. Just because the Intel optimizations are for x86 doesn’t mean that they aren’t also applicable or useful for ARM.
Intel is just desperate about getting people to use their x86 hardware, nobody wants to use x86 because it’s closed/proprietary/expensive uses far too much power. ARM allows for competition/differentiation/choice and much lower prices, much better value for money.
How is Intel’s x86 closed/proprietary ?
Cause Intel is just about the only one to sell it, make it. They’ve been convicted of using illegal tactics to push AMD away more than once. Intel’s only dream is to be the only processor maker in the world and that’s more than half of their strategy, using illegal anti-competitive tactics to try to do some kind of trick or other to exclude healthy competition from the market.
Intel’s questionable use of subsidies and any other business practices don’t make the x86 platform closed. Yes they don’t really license their cores in the same way that ARM holdings or MIPS do, but AMD and VIA have made significant profits on the back of x86 clones and there are more besides that in the embedded market.
That said for most embedded uses I prefer to see ARM or MIPS.
I think you can probably look at AMD and VIA earnings reports of these past 5-10 years to see that they have actually lost a lot of money using x86.
Fortunately my memories of those companies, and others goes back much longer. Fortunes hhave been made in third party x86, and lost, but Intel’s technical roadmap doesn’t preclude the success of others.
This has zero to do with Intel. The signs said they were running on Panda boards (OMAP 4460). The optimizations that Linaro (which is mostly made up of ARM corporations, though it is, itself, a non-profit) has done are toolchain related. Specifically gnu/autotools.
As the developer said, this is about ARM optimizations.
Comments are closed.