Global Foundries to start pumping out 28 nm ARM Cortex A9 processors in 2H 2010

[god_43]

Once upon a time there was a company called Advanced Micro Devices. Their largest competitor was Intel. Both companies competed not only on who made the fastest processor, but who made the best processor manufacturing plants that the world has ever seen. These massive, expensive, difficult to build factories are called fabs.

http://www.intomobile.com/2010/02/24...n-2h-2010.html

[ajaidev]

Excellent news

My first netbook /smartbook will have a ARM 28nm processor that is if its cheap enough, i mean if Lenovo/HP offer ARM based notebooks at the same cost as Atom based netbooks i would prefer the x86 alternative or add a few bucks and get one with Dual CULV/NEO X2....

I want a smart book that costs $150-$200 if it costs more there is not much of a point, MSI Wind U100 is avalible around $230-$240 mark.

[god_43]

but ARM is not x86 though? why cant we just get rid of x86?

[saaya]

but ARM is not x86 though? why cant we just get rid of x86?

we are already in the process...
x86 has been popular because of office but mostly because of games.
for years any chip could offer enough perf for office apps and software is cheap, so the driving factor were games. were...
how much market share does x86 have when it comes to games today?

mobile phones - nope
nintendo ds - nope
psp - nope
iphone - nope
ps3 - nope
xbox 360 - nope
wii - nope

if you look at sold games per cpu architecture and game revenue per cpu architecture, x86 is a niche segment...

im wondering what pumping out means though?
mass production?
and H2 when we are already in march? thats a pretty vague timeframe considering H2 starts in only 8 weeks... so my guess is the REAL timeframe is late 2010... which matches GFs 28nm schedule they mentioned before...

[ajaidev]

We can only do that by offering ARM at a lower cost alternative.

For $300 i would not prefer to buy a A9 based system over a Atom based system simply because it x86 and can play nice with most of my app's that i have come to trust.

My Apple iBook G3 was based on PPC and it was my first laptop that did not have Windows and the result was total confusion, i had to search for alternative programs that i had for Windows and then buy them. Now i have Macbook Air which i like mainly because of its weight and sturdiness, it works great with windows and i can use my old programs also.

What Lenovo and others needto do is intro a cheap smart book like around $200 a real alternative, something what asus did with the ASUS eee 7inch pc's and started a moment....

[Particle]

but ARM is not x86 though? why cant we just get rid of x86?

Do you not remember the days when part of the purchasing decision for a PC was how big the "software library" for its architecture and operating system was? It's a distant memory. The only reason architecture is important for us is because of the software we want to run. Windows 7 only runs on x86 CPUs. Most of the software we care about runs on Windows. It's that software dependency that dictates our eternal marriage to x86. It would be a HUGE, HUGE, ENORMOUS effort to move general purpose computing to something new. There would have to be an amazing benefit to doing so before anyone would even bothering to make a serious effort. Take a look at how hard the 64-bit migration has been considering how much easier x86-32 > x86-64 is compared to going from like x86 to MIPS.

The smaller the market the easier it would be. (think Apple's PowerPC to x86 migration versus Earth's x86 to ?) It's obviously possible. It's just not going to happen any time soon.

[saaya]

Do you not remember the days when part of the purchasing decision for a PC was how big the "software library" for its architecture and operating system was? It's a distant memory. The only reason architecture is important for us is because of the software we want to run. Windows 7 only runs on x86 CPUs. Most of the software we care about runs on Windows. It's that software dependency that dictates our eternal marriage to x86. It would be a HUGE, HUGE, ENORMOUS effort to move general purpose computing to something new. There would have to be an amazing benefit to doing so before anyone would even bothering to make a serious effort. Take a look at how hard the 64-bit migration has been considering how much easier x86-32 > x86-64 is compared to going from like x86 to MIPS.

The smaller the market the easier it would be. (think Apple's PowerPC to x86 migration versus Earth's x86 to ?) It's obviously possible. It's just not going to happen any time soon.

exactly... but nowadays most time is spent online... and more and more apps run in browsers.
there was a time when people used outlook on their pcs and laptops... now thats only for business, everybody uses webmail, and actually even businesses are using webmail (their own webmail, but still)
so as long as you have internet access and a platform can run a browser, thats it... thats enough for 90% of what most people do on their netbooks...

well the prototypes ive seen and heard of are amazing...
passive cooling, 8+ hours of battery life + cheaper than atom + thinner and lighter than atom netbooks...
sure, all you can do is browse the web, but hey... isnt that all you really CAN do on a small form factor laptop?
who seriously does production or plays games on a 10" or smaller portable?

[Manicdan]

^ im with saaya, the online world is changing everything, office 2010 is going to have some kind of online stuff too, and with that maybe it will become OS agnostic too. google docs are already a great way to manage alot of things, since you dont have to put the files on a thumb drive.

as we see cellphones merge into personal PCs, and gaming PCs merge into consoles, i hope x86 phases out. the only tough thing is the desktop, where its designed to be you can replace any one part and supposedly keep the rest.

[Chumbucket843]

well, all apps would have to be recompiled to run on a mobile device anyways. you might as well get the cheaper and faster ARM system. more work would have to be done for porting like the operating systems API's i.e. opengl es for graphics. the future of computing is multimedia so graphics will be most important.

[Helmore]

well, all apps would have to be recompiled to run on a mobile device anyways. you might as well get the cheaper and faster ARM system. more work would have to be done for porting like the operating systems API's i.e. opengl es for graphics. the future of computing is multimedia so graphics will be most important.

Isn't this why the web and APIs like Silverlight are a pretty nice option for cross platform and ISA independent development?

[Bobsama]

we are already in the process...
x86 has been popular because of office but mostly because of games.
for years any chip could offer enough perf for office apps and software is cheap, so the driving factor were games. were...
how much market share does x86 have when it comes to games today?

mobile phones - nope
nintendo ds - nope
psp - nope
iphone - nope
ps3 - nope
xbox 360 - nope
wii - nope

if you look at sold games per cpu architecture and game revenue per cpu architecture, x86 is a niche segment...

im wondering what pumping out means though?
mass production?
and H2 when we are already in march? thats a pretty vague timeframe considering H2 starts in only 8 weeks... so my guess is the REAL timeframe is late 2010... which matches GFs 28nm schedule they mentioned before...

Original xbox (pentium iii and geforce4) was x86 AND DirectX.

[Aerwidh]

I do like that the article's source mentions a projected 100% increase in standby battery life

[Manicdan]

I do like that the article's source mentions a projected 100% increase in standby battery life

and its afterburners give it 200% efficiency! (for the futurama fans)

[MrMojoZ]

Original xbox (pentium iii and geforce4) was x86 AND DirectX.

Geforce 3 wasn't it?

[Helmore]

and its afterburners give it 200% efficiency! (for the futurama fans)

You'd have some weird perpetuum mobile.....

[Bobsama]

Geforce 3 wasn't it?

GeForce something. I was originally surprised when the 360's specs were announced and an x86 processor was not included.

[aka1nas]

thats it... thats enough for 90% of what most people do on their netbooks...

That's irrelevant if it can't handle that last 10%. Rational actors don't make purchases that satisfy only a subset of their requirements (with price being a requirement, of course). Atom has shown that x86 can be cheap, and Intel will be forced to continue down that path to maintain marketshare.

[saaya]

well and what, atom can do less than 90% of what people do on their pcs good enough to be considered an alternative... yet its still a huge success thanks to price, size and battery life... all of which arm is even better at

[Speederlander]

well and what, atom can do less than 90% of what people do on their pcs good enough to be considered an alternative... yet its still a huge success thanks to price, size and battery life... all of which arm is even better at

And with Intel's superior manufacturing capability and and nearly unlimited resources, ARM's size and power advantage will be put to serious test.

[Behemot]

And with Intel's superior manufacturing capability and and nearly unlimited resources, ARM's size and power advantage will be put to serious test.

But that's (I mean serious ARM succes) unlikely to happen because of Intel actually has no interest in RISC due to the Atom. So Intel would not only unsupport it, they'll fight against it.

[mstp2009]

I think his point is that we should expect Intel to continue to develop Atom. The first generation was a good product for it's target market. The second generation didn't make too many improvements to the chip itself, but did bring down power levels by integrating more components and reducing power draw for the platform overall.


I would expect the 3rd gen to improve performance at the same power envelope, or perhaps even lower power.


And on top of all of that, it has a huge HUGE codebase of x86 software available to it.



ARM, however, will still continue to be king in cell phones b/c there is minimal x86 presence there.

[zalbard]

But that's (I mean serious ARM succes) unlikely to happen because of Intel actually has no interest in RISC due to the Atom. So Intel would not only unsupport it, they'll fight against it.

Yeah, Intel would rather promoted IA64 than ARM.

[saaya]

And with Intel's superior manufacturing capability and and nearly unlimited resources, ARM's size and power advantage will be put to serious test.

thats what i thought... but look at what intel is doing... pinetrail is still 2 chips instead of one for arm, it still has the fsb interface inside the cpu/nb chip, limiting performance, it still only supports ddr2, it has the most basic igp they could possibly get away with, it only runs at 1.66ghz, its still made in 45nm...

poulsbo + silverthorn was the same perf and power consumption as pinetrail... and they are 2 years apart... intel doesnt seem to focus on atom at all... they are slowing down the rnd cycle as much as possible trying to milk profits...

im sure that as soon as arm starts to ship smartbooks which compete with atom netbooks they will speed things up, but right now they seem to sit back and wait and let arm make the first step... thats a risky move...

[Behemot]

The problem is software as was already said. Every platform running on RISC chips (mainly smartphones) has it's own set of software, but only basic things (browsers, some of which just recenty got the Flash support; file managers; mainla Java games; some basic image and video viewers...). You cannot do some more serious things like playing more than MP3 an some few other (usually unique) formats. No SW for image editing; no networks programs (even FTP clients - totaly out of the game, you can forget on such a things; be glad for some WiFi configuration-only HW things). Nothing. Everything I think about is not possible on RISC machines now. Hoping for everything online is a bit unrealistic, such a portable device like netbook just can't relly on online packages, what if you do not have proper Internet connection somewhere?

All of this is just SW problem on x86, but if you want something on RISC, you usually need to add the right instruction set, maybe some specialized part of core to "accelerate" it (means to be able to make it even run) and make unique SW. Yes, you can try the "brute force" way if you don't have the right instructions (it's alot easier when we have at least C/C++ support now, as e.g. RISC chips in general mobile phones still only support Java or so in most cases), but once again, the RISC chips are usually not so powerfull for this. I mean, there are MMX and SSE instructions, but most PC software runs (or is able to) on x86 only (it is wide set enough), it's just slower, but it will run (try some old PI or PII).

So RISC probably has future, I think in adding more instructions thus closing to x86 (what is in fact the way China is going with their own CPU project). When it will have proper HW base, more people will write SW for it and maybe we'll see some more x86 programs ports. You could try it now, but if the compiler even handles it, it will be too slow. You know - frequency and more cores on A9 is nice, but it's not everything.

But I am only chair guru in this, so if somebody who actually has programming experiences could contribute, I'll be glad to be corrected

[aka1nas]

well and what, atom can do less than 90% of what people do on their pcs good enough to be considered an alternative... yet its still a huge success thanks to price, size and battery life... all of which arm is even better at

Atom can run modern x86 software, albeit slowly for some applications. Most of the things that it can't do passably well are video and gaming related, and is mostly due to the weak Intel video. That's no more or less limited than any other architecture in that space, and we already have shipping hardware that has solved this problem in ION.