Intel is a big profitable company, but is being left behind by a major technology paradigm shift.

The future, and present, is mobile devices.  Intel has built its business on making fast processors faster.  Employing this approach meant all power consumption considerations were swept aside in the race for ever more speed.  However to most users, computers are “fast enough.”  Do you really need that hot new multi-core Pentium to run Word or Excel?  PC upgrade cycles have been lengthening.

Computing on the go

Carrying your computer with you has become the new job and life requirement.  Even before the iconic iPhone of 2007, people started using their mobile devices for text messaging and email on-the-run.  The explosion of form factors from smartphones, airbooks or ultrabooks to tablets of all sizes shows portability and connectivity are the prized features now.

Microsoft understand this trend and has tried to carve a new path with its Windows Phone line, so far with limited success, but since Nokia has bet its farm on Windows Phone it is still too early to tell how this will fully play out.  The crown jewels of Microsoft are its Office suite of software products: Word, Excel, PowerPoint, Outlook.   Microsoft isn’t taking any chances and is planning on a tablet version of Windows 8 this year.  Windows president Steven Sinofsky made it clear that x86 Windows apps won’t run on ARM Windows tablets.  You can read that quote here and here.

Windows 8 tablets will be ARM-based to take advantage of ARM’s superior power profile.  Battery life is the new megahertz.

Atom is too late

The mammoth Intel ship has been turned to try to meet this threat and to reduce its CPU power consumption.  Atom chips are starting to get reasonable, but especially in graphics their hunger for power lags far behind the power nibbling ARM chips.  So what can Intel do?

The Third Way

Intel is already an ARM licensee and manufactures some of their chips.  Why not support the ARM instruction set on an x86 chip?

Lest ye think this is heresy, there is precedence for this path.  TransMeta pioneered the concept of using a VLIW architecture chip to recompile, optimize and translate x86 opcodes at runtime into the CPU’s internal machine code.  They actually termed this technique “code morphing.”

Intel could take the same path and code morph ARM opcodes into x86.  Intel’s Itanium architecture is also VLIW.  ARM is a RISC style architecture with regular size instructions and could be mapped onto a VLIW word.  The challenge is to design a compiler or binary pre-processor that is intelligent enough to know how to build the very long instruction words.

Essentially this is just a translation step, best done with software compilers to translate binaries.  At this stage of VLSI life, the x86 instruction set is just one design mapped onto an internal, simpler CPU.  Why not also add the ARM instruction set and gain the ability to run ARM binaries?  ARM libraries that manipulate and push pixels in power friendly ways would be immediately available and Intel could go head-to-head with mobile big boys like Qualcomm and Texas Instruments in seeking mobile design wins.

How long would you push a shopping cart with a stuck wheel uphill?  That’s what Intel is doing with its Atom architecture.  Emulating ARM could revive their fortunes.  ARMulation can save Intel.

 

Leave a Reply

(required)

(required)

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>

© 2022 pdxmobile Suffusion WordPress theme by Sayontan Sinha