[Tom's] AMD jumping on Andriod bandwagon.

[busydude]

http://www.tomshardware.com/news/Brazos-Llano-APU-AMD-Fusion-Google-Android,12548.html

It was only a matter of time before AMD followed Nvidia's lead and finally jumped onto the Android bandwagon, as the company is now recruiting engineers to create chipset drivers for Android. According to unnamed sources, the nation's second largest CPU manufacturer is now looking to offer notebook and tablet partners chipset solutions supporting Google's popular mobile platform. "The Linux Base Graphics team is looking for Android Driver Development engineers to help us evolve our driver stack for new platforms and in line with the development trends in the Android ecosystem," reads the job description. "Experience with video decode acceleration within the Android web browser or video player application would be an asset. Experience with webm and/or OpenMax would be a bonus."

There was no mention of AMD entering ARM market in this article, maybe this has to do with upcoming Google OS.

Has Andriod been ported to X86?

 

[drizek]

I think acer made an atom android netbook at one point.

 

[Idontcare]

Has Andriod been ported to X86?

Excellent question. How difficult would such a task be? (roughly)

Linux ports to different ISA's seem to happen overnight, I'd think an Android port would be like a 6 month effort if Google wanted it done.

 

[OS]

supposedly acer was already going to do a dual boot android/win 7 netbook as mentioned already

 

[Cogman]

Excellent question. How difficult would such a task be? (roughly)

Linux ports to different ISA's seem to happen overnight, I'd think an Android port would be like a 6 month effort if Google wanted it done.

IIRC, android is pretty heavily based off of linux, so I think that 6 months is a bit of an exaggeration. Since most applications in android are java based, I would imagine things would happen really quickly.

Most android apps should port straight over as well. The only issues are going to be in the applications that use native arm libraries.

http://en.wikipedia.org/wiki/Android_(operating_system)

Seems to confirm this.

Android was built from the ground up to be fairly platform agnostic.

 

[podspi]

Excellent question. How difficult would such a task be? (roughly)

Linux ports to different ISA's seem to happen overnight, I'd think an Android port would be like a 6 month effort if Google wanted it done.

http://www.android-x86.org/ seem to already be on it. Apparently for performance reasons some apps don't use dalvik, which I think is the big stumbling point for being purely hardware agnostic.

Edit: I actually loaded this on my netbook a while ago. Wireless didn't work and the interface was identical (so kind of clumsy) but it works. I'd love to have an x86-based phone from either vendor if they could fit it in the needed power envelope. My D1 @ 1ghz is not fast enough...:twisted:

 

[drizek]

im going to try out the live image in virtualbox.

 

[IntelUser2000]

Has Andriod been ported to X86?

Yes it has, but it needs performance optimizations depending on the build you get.

There's a specific build for the Viliv S5 UMPC I have in my sig. I have tried installing it, and successfully got it working, internet and everything. But the Viliv S5 build wasn't optimized. There's also a generic x86 build up to 2.2 version but didn't know the keyboard commands so I gave up on it for now.

I'm pretty sure you can get it installed working properly on certain Netbooks. Search Android x86 for build.

 

[Dark Shroud]

Too little too late is a joke. AMD's APUs will kill in tablets and could probably do well in cell phones. Just wait until next year when Bobcat cores are in use.

It's much easier for AMD & Intel to build down than the ARM manufactures being able to build up. A low power duel core APU running at 2ghz makes a compelling argument.

 

[BD231]

IIRC, android is pretty heavily based off of linux, so I think that 6 months is a bit of an exaggeration. Since most applications in android are java based, I would imagine things would happen really quickly.

Most android apps should port straight over as well. The only issues are going to be in the applications that use native arm libraries.

http://en.wikipedia.org/wiki/Android_(operating_system)

Seems to confirm this.

Android was built from the ground up to be fairly platform agnostic.

I was under the impression android is based off Linux as well. So, why is porting to x86 even a question?

Should be a breeze.

Too little too late is a joke. AMD's APUs will kill in tablets and could probably do well in cell phones. Just wait until next year when Bobcat cores are in use.

It's much easier for AMD & Intel to build down than the ARM manufactures being able to build up. A low power duel core APU running at 2ghz makes a compelling argument.

Struck me as odd that once nVidia jumped so did AMD, then again, they've had a lot of change's in upper management so change is to be expected. I find it hard to ignore the " anything you can do, I can do better " in this all though. Seems a bit sudden considering how long they've ignored mobile device's.

 

[wahdangun]

good at least it will improve linux driver on pc as well

 

[IntelUser2000]

Too little too late is a joke. AMD's APUs will kill in tablets and could probably do well in cell phones. Just wait until next year when Bobcat cores are in use.

AMD will need 28nm to be really viable in Tablets and at least 20nm for Smartphones. They've been saying 3W is a sweet spot for Tablets. It's still more compared to ARM variants but at that point, going lower isn't really necessary. For smartphones, 2W, at the most. Moorestown's Lincroft wasn't accepted because it was a 2.2W chip.

Oak Trail is 3W now, Android-enabled Tablets from what I read should have competitive battery life, but still needs shrink to 32nm for a good smartphone chip.

 

[ydnas7]

Android is for ARM, X86 and MIPS
AMD has/had products in both X86 and MIPS

 

[busydude]

Yes it has, but it needs performance optimizations depending on the build you get.

This, along with support, has been my main concern all along, I used to be a heavy Linux user when I was doing my undergrad because I could not afford a copy of Win XP(No job and living with my parents) and I had a hard time searching for the right distro that supports my configuration.. finally ended up with Ubuntu as it was the one.. I found that gave me least number of issues.

Given my past experience, I would not touch these with a ten foot pole unless manufacturers provide support for Andriod in x86 systems. It may be a fun little project to kill some time, but right now I don't have the time and patience to play with it.

 

[podspi]

This, along with support, has been my main concern all along, I used to be a heavy Linux user when I was doing my undergrad because I could not afford a copy of Win XP(No job and living with my parents) and I had a hard time searching for the right distro that supports my configuration.. finally ended up with Ubuntu as it was the one.. I found that gave me least number of issues.

Given my past experience, I would not touch these with a ten foot pole unless manufacturers provide support for Andriod in x86 systems. It may be a fun little project to kill some time, but right now I don't have the time and patience to play with it.

That's probably why AMD is hiring people for Android now. Can't expect the project to write drivers (and perhaps ARM emu for native binary apps?) for you. I think 28nm Bobcat could work well with Android on a tablet. I don't think they're even close to being able to put one in a smartphone, though. Especially if they need to do ARM emulation...

 

[Dark Shroud]

Struck me as odd that once nVidia jumped so did AMD

More like returning to mobile because they currently have some financial stability and a good product(s) for tablets. Nvidia doesn't have much of a choice in not going ARM.

AMD will need 28nm to be really viable in Tablets and at least 20nm for Smartphones.

I was thinking 28nm as well for tablets. I'm not sure what they'll have to do to be in phones. But I like to be optimistic.

Oak Trail is 3W now, Android-enabled Tablets from what I read should have competitive battery life, but still needs shrink to 32nm for a good smartphone chip.

We're on the same page here with Oak Trail. I actually expect Intel to continue to do die shinks past 32nm for Atom. Just to stay competitive with ARM and show everyone they can.

 

[podspi]

Does anyone know the breakdown between the power used by the Bobcat cores and the SPs? AMD could always come up with a separate die with very few SPs for mobile use. Let's face it, current mobile screens are very very small, and thus their GPUs do not need to be very powerful... 80SPs is way overkill for a smartphone. Pushing it for a tablet.

 

[IntelUser2000]

Radeon HD 5450 had 59mm2 die with 80SPs and doubling to 160SPs with the HD 6450 resulted in a 67mm2 die, and with same architecture, transistor count is well correlated with power use. So they'll have to cut lot more than just the SPs to reduce power significantly.

It'll likely need a significant overhaul, even if they plan to stick with whatever base graphics architecture they have.

Lost Circuits had CPU power at 10W and CPU+GPU power at 16W. I'm guessing it might be almost 50/50 with 9W chips though.

 

[BD231]

More like returning to mobile because they currently have some financial stability and a good product(s) for tablets. Nvidia doesn't have much of a choice in not going ARM.

All I remember was hand held GPU tech, not quite the same thing.

 

[nenforcer]

Android, being Linux based, should be easily ported to x86 if it hasn't already. I think the bigger question is when will Dalvik, the Android application compiler, have a JIT back end supporting x86, if it doesn't already?

 

[grimpr]

That's probably why AMD is hiring people for Android now. Can't expect the project to write drivers (and perhaps ARM emu for native binary apps?) for you. I think 28nm Bobcat could work well with Android on a tablet. I don't think they're even close to being able to put one in a smartphone, though. Especially if they need to do ARM emulation...

AMD knows perfectly well their position in the smartphone space, its simply "stay out of there and dont play with fire". Windows 8 Tablets are another story...remember, its 28nm HKMG at both foundries, TSMC and GlobalFoundries and the chips will come fully power gated /w turbo core.

Charlie D. chimed some news on this.

http://semiaccurate.com/2011/04/11/guess-what-taped-out-at-amd/

 

[Dravic]

http://www.tomshardware.com/news/Brazos-Llano-APU-AMD-Fusion-Google-Android,12548.html



There was no mention of AMD entering ARM market in this article, maybe this has to do with upcoming Google OS.

Has Andriod been ported to X86?

Maybe they license ARM and add a GPU for their SoC. I'd rather see low watt bobcat based apu's in a dual core 10" ~$400 Android tablet for Christmas.

 

[Martimus]

Maybe they license ARM and add a GPU for their SoC. I'd rather see low watt bobcat based apu's in a dual core 10" ~$400 Android tablet for Christmas.

I would find it surprising if they liscenced ARM again, and developed a SoC using it and their graphics technology again considering they sold that business to Qualcom a couple years ago (Snapdragon). I would think there would have been some language in the contract that would keep them from going back into that market segment, since they obviously already have all the information on the Snapdragon SoC, and could easily just start building their own clone to it if they wanted.

 

[Idontcare]

I would find it surprising if they liscenced ARM again, and developed a SoC using it and their graphics technology again considering they sold that business to Qualcom a couple years ago (Snapdragon). I would think there would have been some language in the contract that would keep them from going back into that market segment, since they obviously already have all the information on the Snapdragon SoC, and could easily just start building their own clone to it if they wanted.

If it was the same CEO at AMD then to be sure it would be handled differently as re-entering the ARM arena would be received as a sign of weakness that the CEO had goofed.

Given that AMD has a different CEO at the helm who can make a clear strategic case for re-entering the ARM marketspace (no sacred cows for him when it comes to reversing mistakes of past CEOs) I'm in the opposite school of that in that I would be surprised if they did not pursue it.

Regarding the possibility of a "non-compete" clause, without question this would have been part of the original deal but these business unit spin-offs usually come with a well-defined time table as to when the non-compete expires. There is no standard, but 5 yrs is certainly expected, Qualcomm would have had to pay a higher price if they wanted an exclusion window that was any longer.

What is more probably IMO is that AMD would simply enter negotiations with Qualcomm regarding the royalties paid for cross-licensing one another's IP portfolios and the non-compete terms would be renegotiated as part of the overall compensation package.

Even TI, who was/is staunch competitive rivals of Qualcomm, has cross-licensing agreements in place. It is unavoidable, all one strives to accomplish is reaching a sort of balance in the IP space such that no one business has the upper-hand at the negotiating table. Very much like the cold war and the policy of M.A.D.

That said, regardless the logistics of the arguments for the case that AMD could or should head into ARM marketspace I just don't see their bottom-line in accounting as being healthy enough to support a further dilution in their already thin R&D budget.

Every dollar that goes to an ARM initiative is a dollar that is not going towards their next-gen x86 CPU development or next-gen GPU development.

 

[podspi]

Every dollar that goes to an ARM initiative is a dollar that is not going towards their next-gen x86 CPU development or next-gen GPU development.

Yes, and we really don't need another vanilla ARM SoC designer...

I'm excited for Denver because it should be a radical departure from everything else out there. I don't think AMD has that kind of R&D money (to spare), at least not yet.