Guild Wars 2 Available for Mac OS

by Matthew F. Hibben

Good news for Mac gamers this week as ArenaNet announced Guild Wars 2 will be cross platform going forward.  A Beta Client is available for Mac users, and anyone who purchased GW2 can download this new client, even if they are currently using the PC client.  Anet has stated the features for the new Mac client will mirror in every way those of the PC version of the game.


Motorola Launches RAZR i with Intel SOC

by Mark W. Hibben

But not for the U.S.

Whenever Apple launches a new product, the tech media immediately begin a “due diligence” search for “viable” alternatives, and Apple’s competitors are more than happy to oblige with product announcements, however half-baked.  Motorola’s announcement of the RAZR i (not iRAZR of course) is just one such announcement.  That the phone is only being offered in Europe and Latin America is all the consumer (U.S. or otherwise) needs to know about this phone from a purchasing decision standpoint.  However, it is interesting in other respects.

The most interesting thing about the RAZR i is that it contains an Intel SOC (System on Chip) rather than the usual ARM SOC.  Announced with much fanfare by Intel CEO Paul Otellini at CES in January,  the Atom Z2460  has only attracted a few manufacturers besides Motorola such as China’s Lenovo and India’s Lava, none of whom are pushing their phones into the U.S. market.  Although Motorola was represented as Intel’s strategic partner in this endeavor, Motorola has taken its time rolling out the first product based on the SOC.  With good reason.  There is every indication that the Intel SOC, despite having vastly improved power consumption, still isn’t competitive with run-of-the-mill ARM processors.

Motorola is also not being very forthcoming about what Intel is actually inside the RAZR i, but it must be the Z2460 or subsequent variant.  Even Intel is not particularly helpful when it comes to info on the Z2460, offering little more than the January 2012 press release on their U.S. site.  However I was able to dig up a product brief which I’ll make available for download.  The product brief shows that the 2460 is a single core 32 bit processor with Hyperthreading, giving approximately dual core performance.  Clock rate is dynamically variable with Intel’s Burst Performance Technology (BPT) to allow brief periods of maximum clock rate.  The Motorola RAZR i is spec’d at 2.0 GHz, which is probably the BPT maximum.  The SOC also integrates graphics processing with a built-in HDMI 1.3a interface and video out at 1080p at 30 frames/sec.  The SOC also interfaces directly to HD video cameras at 1080p (30 fps) and still cameras at up to 24 Mpix.  Unfortunately for the 2460, Intel went cheap with a 32 nm process rather than their most advanced 22 nm process.  This, more than anything else, probably doomed the 2460 to being a second rate mobile processor, since it would have needed the more advanced process to compensate the power efficiency disadvantage of Intel’s x86 architecture relative to ARM.  Benchmarking has generally shown the Z 2460 to be slower than the iPhone 5’s A6, but then the RAZR i is also running Android 4.0, and the burden of running Dalvik (Java compatible) virtual machines for every app handicaps Android in any comparison with iOS. 

While the RAZR i may never get to the United States, except as a gray-market import, there’s no reason to expect Intelorola to give up just yet.  Intel’s Atom road map shows 22 nm Atoms arriving in 2013, and these may actually be competitive with ARM, at least the current generation of ARM. 


Is the iPhone 5 Processor Truly Custom Designed by Apple?

by Mark W. Hibben

Information Gleaned from the iOS 6.0 SDK says, “Maybe”

Over the weekend, Anand Lal Shimpi published an article declaring that the new Apple A6 processor in the iPhone 5 was not in fact based on a licensed ARM Cortex A15 design, as he had first surmised, but is in fact entirely new and custom designed in house at Apple, except for being still compatible with the ARM processor instruction set.  What this would mean is that Apple has finally become the mobile equivalent of the Wintel duopoly, able to control almost every aspect of the design of their mobile devices from the processor silicon to the operating system, but with better integration and coordination than Wintel has ever achieved. 

Let me back up for a moment to explain the design distinctions in more detail.  ARM Holdings doesn’t make silicon, they just make designs, and then licenses these to processor foundries like Samsung or Texas Instruments.  ARM provides manufacturers two basic kinds of license.  The first is a license to build processors that use a standard ARM ISA (Instruction Set Architecture).  The ISA is just a set of numerical codes that tell the processor to perform certain actions.  In the end, all apps are just compilations of these codes, drawn from a particular ISA, since that’s all the processor understands.  A manufacturer that licenses the ISA is free to implement the instructions in any way they see fit, as long as the processor performs the action specified by a particular code correctly.  The second type of license pertains to specific processor logic designs, such as ARM’s Cortex series.  Currently, the Cortex A9 is pretty standard, but there’s still a lot that the manufacturer of a Cortex A9 processor must decide on in terms of physical implementation of the logic in the silicon, so a Samsung Cortex A9 might by physically different from a TI Cortex A9 processor.  Apple has both types of licenses, leaving it free to custom design a processor according to the ARM ISA, or license one of ARM’s processor designs. 

Needless to say, custom designing the processor is more difficult and expensive, and heretofore, the consensus opinion has been that Apple was merely licensing completed silicon designs from the likes of their processor manufacturing partner, Samsung.  So Apple was assumed to not even be doing the silicon designs based on a licensed logic design from ARM, but merely pasting in the completed Samsung designs into their “custom Apple” SOCs (Systems on a Chip).  This is in fact what happened with Apple’s very first SOC, the A4, that was put into the first iPad.  Apple could still claim credit as the “designer” of the SOC, since the SOC consists of many design components including the core processor design discussed above, the graphics processor, and various interface functions, all of which have to be integrated together into a working SOC.  But it was known that Apple aspired to do and be more.  Apple had bought a couple of processor design houses that really were doing their own processor core designs, PA Semi and Intrinsity, and both specialized in patented low-power technology ideal for mobile applications.  Apple then hired in 2008 Mark Papermaster to be Senior VP of Devices Hardware Engineering , who had lead the Power PC development effort at IBM.  Then Papermaster left shortly after the blow-up regarding the iPhone 4 antenna design problems, leading to speculation that iP4 had caused Papermaster’s downfall.  Bob Mansfield took over from Papermaster, but Mansfield wasn’t really a processor design guru, and the processor design effort seemed to languish unfulfilled through 2010 and 2011. 

Why does Mr. Shimpi now believe that the Apple A6 contains a custom designed ARM ISA compatible processor core?  The evidence he presents is a little flimsy.  The first and most solid piece of evidence comes from the iOS 6 SDK being seeded to developers.  I should think that showing images captured from the target processor settings window is a violation of the developer license, but somehow Shimpi got hold of said images, which clearly demonstrate that the iPhone processor core is either a Cortex A15 (his initial guess) or some custom processor compatible with the ISA that ARM uses for the A15.  His inference that the A6 core isn’t based on the A15 design is rather speculative, based on a supposedly reliable source inside Apple, as well as performance capability and power consumption data published by Apple.  Given the amount of leeway that manufacturers have in implementing ARM logic designs in silicon, I doubt that there is sufficient information to jump to the conclusion of Mr. Shimpi.  As for the "reliable source", not too long ago such sources were being quoted assuring the existence of an Apple television, until the rumors were deliberately suppressed by Apple management. I always take such claims of insider knowledge with a grain of salt, having been misled myself by "Apple insiders". But now that interest has been sparked, outfits such as Chipworks will undoubtedly get to the truth of the matter.  Chipworks regularly takes apart processors, then grinds off various metallization and silicon layers until they get to the basic logic.  Then they take a picture.  Comparing this picture with images from other processors (such as a Samsung Cortex A15 Exynos) is usually definitive, if there’s a match.