ARM licenses Atom-killer chip to un-named vendor

Tue, 10/27/2009 - 8:40am — Seth Weintraub
2831

Here's a riddle:

What is five times faster than an Intel Atom chip, yet is 60% smaller and uses about the same power?  Give up?

It is a dual-core ARM Cortex A9 chip running at 2GHz...and someone is building something with it as we speak. 

Buried inside ARM's 3rd quarter earnings announced today was that they've given one of their licensees the rights to a chip that is truly a laptop-class processor, yet has the physical and energy footprint of a smartphone:

Twenty-three of the licenses were for ARM’s advanced Cortex and Mali™ graphics processors of which eight licenses were signed for the Cortex-A family processors for use in consumer electronics and mobile computing applications, including a license for ARM’s 2GHz implementation of a dual core Cortex-A9 processor.

We've contacted ARM who've said the licensee cannot be disclosed under their agreement.

We know that Apple is rumored to be an ARM licensee and that that they've reportedly split the PA Semi group up into two parts.  One working on Smartphone chips the other working on tablet processors.  It is now generally accepted that the Apple tablet will run a ARM Cortex processor. It will likely want a bit more horsepower than the ARM Cortex A8-class Samsung chip inside the iPhone 3GS.

So is Apple the mystery licensee of this new high end chip?  Perhaps.  It could also be Samsung, Freescale, TI, or NVIDIA.

Will this be the processor class in the upcoming tablet?  Probably not.  Not only is it much faster than necessary for a tablet, it is just now being produced.  Apple will likely need much more time to play with this design before putting it into production. 

You can bet however, that Apple will be using ARM Multi-core Cortex A9 processors in future products, most likely built by their internal PA Semi team.

( Filed Under: )

Comments

"Not only is it much faster

168

"Not only is it much faster than necessary for a tablet, it is just now being produced" Speak for yourself! I want my tablet to run photoshop!

haha... you made me laugh!

711

haha... you made me laugh!

Shame there isn't an arm

76

Shame there isn't an arm version of Photoshop no matter how fast the chip, I guess Rosetta could be repurposed but that's a bit sloppy for anything other than a transition which this isn't .

> What is five times faster

810

>
What is five times faster than an Intel Atom chip, yet is 60% smaller and uses about the same power?
<

If it uses the same amount of power as an Atom then it wont be used for an Apple tablet. The tablet will need to be thin with better battery life than netbooks.

I think that's where the "60%

119

I think that's where the "60% smaller" comes into play...

Docking iPhone that works as

98

Docking iPhone that works as a desktop with monitor connected to dock anyone?

If this thing truly runs 5x

79

If this thing truly runs 5x faster than an Atom, any future tablet using it would EASILY be able to run FULL OSX with REAL applications, as my Dell Mini9 Hackintosh clearly demonstrates.

Again full osx and all it's

106

Again full osx and all it's built in apps and all 3rd party apps are made for intel, they won't transition to this it's just a sub catagory product, therefore iPhone variant.

What is your brain doing ?

97

As if OS X hasn't been running on several architectures all along. Remember the transition to Intel from PPC ? Apple had OS X running on Intel all along and didn't blab about it. What makes you think they have changed their strategy ?

Besides, running software on more that one architecture helps flush out bugs and bad code practices better than just running on a single one. Can you say robust code ?

I am reasonably certain that OS X and most, if not all, Apple apps run across architectures. If for no other reason than strategic insurance. Steve Jobs is a very clever fellow and really likes crafty engineers.

Not to throw water on your

87

Not to throw water on your fire, but the 2GHZ design released by ARM is a hard macro and not likely to be used by Apple. Apple already has their own design house so they are not the unnamed licensee. Apple already licensed future ARM IP so they probably worked a design based on the A9 Cortex and PowerVR and customize the chip layout. If Apple is manufacturing a product for release in 1st qtr 10, then they already have first silicon in hand and the design is frozen.

Can you explain why you think

610

Can you explain why you think Apple wouldn't use a hard macro ARM device? It's a stable product that's simpler than Intel cores, on the one hand, and it would be nice to hear your reasoning on the other hand :)

 

Thanks!

What I'm saying is Apple will

68

What I'm saying is Apple will create their own design using ARM and PowerVR IP.
When Apple bought PA SEMI, they were making a strategic decision IMO to design their own System on a Chip (SOC) . For whatever reason they looked at the Atom roadmap and what Apple was planning and decided to jump on the ARM architecture for Iphone. In Oct 2007 ARM released initial info on the Cortex A9MP design. In April 2008 Apple bought PA SEMI . The driving feature of the PA processor was performance per watt. The PA SEMI chip used a PowerPC CPU core and custom designed parts for the North/South Bridge. A team with PA SEMI skill could take the ARM Soft Macros and build a physical design and work directly with the foundry for production. Design of a new SOC cost in the range of 50M so unless Apple can sell a bunch they don't amortize the high fixed cost. The hard macro is a complete design including the layout on the silicon which means no customization.

 It does not cost ~$50m to

86

 It does not cost ~$50m to design an SoC if you are an ARM licensee because they give you the cores, the tape-out and the tools. PA Semi is definitely doing ARM work, confirmed by any sources and their own job ads.

That being the case, would Apple use an ARM9 or ARM11 core, or the better performing Cortex series IP? Apple has a history of disabling features in more capable devices, so having a decent SoC design and simply not enabling features in SW seems a likely strategy.

 

11 nm Intel Atom

78

Intel should release a dual-core Atom processor with 11 nm manufacturing process to beat ARM, so that Apple can offer the full Mac OS X (touch) in the tablet.

11 nm
http://en.wikipedia.org/wiki/11_nanometer

When you get to 11nm there

67

When you get to 11nm there will be a bunch more cores on the chip,
you have to do something with all the extra transistors. As the other folks mentioned we got a few jumps to make before 11nm. This year intel is making the jump to 32nm with their CPUs and ARM is at 45NM. In Feb of this year ARM announce 32nm/28nm with High-K Metal Gate with Cortex family cores so designs should move to 32nm in 2010/2011.
At the ARM developers conference they said 32-nm results were showing not only the expected 55% reduction in area, but a 40% drop in leakage, a 30% reduction in dynamic power, and—rather surprisingly—a 24% increase in maximum clock frequency. http://www.edn.com/article/CA6704016.html

 Because, as the wiki states,

76

 Because, as the wiki states, 11nm is likely to tip up in 2022? ;)

Will be in the tablet. A

66

Will be in the tablet. A tablet to separate itself from the netbook and smartphones.

High end graphics --- Apples bread and butter

52

As with the first comment,my wish list is a tablet with full osx, true daylight screen (= pixel qi).  I would be loading geolocated (GE or better) imagery and physically walking through it with real time positioning, site descriptions, actively inputting photographs (Photoshop) and annotate them with (e.g. Adobe Illustrator.)  Probably just dreaming, heres hoping my powerbook lasts long enough to not to not need a switch to Windows which can already do all that (albeit with cludgy solutions).

 

Post new comment

The content of this field is kept private and will not be shown publicly.