How Intel's faith in x86 cost it the mobile market

Sigh

I always thought the 68xxx range had an edge over the x86 ....

Ever tried to imagine an itanic mobile, anyone?

To me, it's way past time we got rid of that monoculture. x86 has hogged the limelight for far too long. Not that I'm saying it ought to die (it ought to due to excessive staleness, but that's not important for this particular argument), but more that the computing field would be far healthier with at least a few different instruction sets running about. Especially now that we can no longer mindlessly crank the megahertzen and squeeze the processes and claim that's enough innovation for the day. It's not nearly enough, moreso every new day.

Intel, embrace ARM.

Please please please Intel, embrace ARM. It is too late with x86.

Its all about profits

The larger problem for Intel is building profitable CPU or SoC product lines. Consider the iPad 4 and most other tablets where the SoC represents less than 10% of the BOM. Compare with Ultrabooks where Intel sees maybe 40% of the BOM.

With that in mind, its difficult to see how fabbing ARM products or moving to a new architecture during the Otellini era could have helped Intel. x86 compatibility is still significant, certainly not as important to personal computing as 10 years ago but the continuing Intel focus on x86 is at least avoiding making matters worse for the company.

"the continuing Intel focus on x86 "

There isn't a continuing Intel focus on x86.

They've had lots of attempts at getting out of that box.

They've all failed to make any significant impact.

It's not looking good for Intel in the post-x86 era.

Its simple really

Intel are pretty safe until MS produce their next OS, if that is a full version of windows that both consumers and exterprise want to use and also runs on either Arm or x86 then they are in big trouble as by 2014 when they catch up with where Arm are now they will still be 18 months behind and producing more expensive chips.

This is even more so if Arm get a real hold on server / big data storage systems.

Intel need to cut back on x86 research and really go full on for Arm if they want to have a product that will sale in the future as a full fat windows that works on Arm will be a cheaper option than the Intel one when upgrading the hardware.

"liked to see the Z8000 improved and expanded"

You, sir, need help.

a) The Z8001/8002 was/were a dog's breakfast of a chip. And a very poor breakfast at that.

b) You appear to have forgotten, perhaps understandably, about the Z80000:

http://en.wikipedia.org/wiki/Zilog_Z80000

I wish you a safe and speedy recovery from your current condition.

Thin gruel?

There simply aren't the big bucks in the portable device processor market that there are in the tradition PC and server markets. The ARM licensing model has fundamentally changed the economics such that any return Intel will make on the portable device processor market would be a fraction of what they are used to.

Should ARM move into the server or PC market, then that would be Intel's worst nightmare as it would have a catastrophic effect on their revenue model. It's still highly debatable if this will happen - it's unlikely that ARM will ever be capable of matching x86 ultimate performance, but then maybe it doesn't have to. If x86 starts to turn into a niche processor, then it will start to look like the Power, Itanium or SPARC markets. Relatively small volumes and high cost. Should that happen, it will be highly challenging for Intel. Not much margin in a largely semi conductor foundry business.

I'm glad Intel aren't successful in mobile, long may they continue to fail.

It's just a shame we can't grow the ARM business more.

x86 on mobile platforms...

How much on-board storage did Intel expect these x86 mobile devices to have?! Windows alone would eat a large proportion of available flash memory (I'm sure I read Windows RT occupies 12Gb of storage) before adding any applications. Intel should have been helping shrink the storage footprint of x86 software - not just the power consumption...

sc/mp to ns32032

SC/MP to NS32032.

Quite a transition. Both from NatSemi.

The NS32000 family looked quite interesting on paper, perhaps how a VAX would have looked if started from scratch with no need for compatibility and with the benefits of then-current semiconductor technology.

Sadly, the x86 and 68K had already won by the time the NS32000 came out.

Intel aquired the 64 bit RISC Alpha CPU from HP (who got it from DEC via Compaq) in 2001. If they had pursued this CPU design the computer world would have been a very different. I once ran Linux on an Alpha PC.

Licence

The earlier chips were cloned by various manufacturers without any licences. (i.e. they were different internally, but behaved the same.)

http://www.guardian.co.uk/technology/2001/mar/08/onlinesupplement5

"Originally, Acorn planned to use Intel's 286 chip in its Archimedes computer. But because Intel would not let it license the 286 core and adapt it, Acorn decided to design its own. "

Surprised the article didn't include this.

Just remember this...

If IBM had chosen the 68k over the 8088 for the original PC, Intel would probably be dust right now. They wouldn't be making CPU chips, but probably EPROMS, and DRAM chips like jellybeans.

Yes, I believe that the 68k is a much better processor, and might have cured several mistakes that were made back in 1981 for the original PC. Wishful thinking (*SIGH*).

it's the licensing

1st the desktop world is rapidly going 64bit... but that's largely AMD IP, licensed *by* Intel.

2nd a major factor in losing out everywhere else by refusing to license x86 or any other IA *to* competitors, except where they'd already been forced to in court last century. After AMD there was no credible alternative and never going to be another. That's a powerful incentive to switch platform even if Intel managed to deliver the low power CPUs the world needs. Few will willingly tie their future to a single supplier, especially one with Intels history of abuse.

The licensing issues means even if Intel deliver a perfect CPU there won't be mass migration to use it. It cedes control of company futures to Intel, with no backup plan.

Plenty of faith in Atom smashing their profits.

Ultrabook chipsets was their last great cash grab in the mobile market.

Cheap and cheerful Intel Atom always had to be stopped/limited.

Apple stepped in with ARM and the rest is history.

The problem now is that even Apple's Intel Macs have problems selling.

RISCy business for ARM

Intel could do a lot worse than buy up or licence the MIPS architecture. Direct competitor to (and architecturally similar to) ARM, and unlike ARM, has already made the jump to 64 bit some time ago...

Power consumption?

Maybe when passive an Atom can use as low power as some ARMs. But actually running the lowest power x86 system can be x5 consumption of ARM system, excluding screen and Transmitters.

ARM for baseband chips and Mobiles OS's

The success ARM achieved was due to the fact that manufactures could license the ARM CPU and include it with their base band chips responsible for communicating over the airwaves. It is designed to be as simple as possible to aid other manufactures incorporating the CPU on the same silicone with their RF modules.

Integrating many chips is what led to the early mobiles being smaller and less power hungry.

Mobile phone manufactures could create their own custom RF chips and include the ARM core which would enable running the vast software written for the ARM platform.

The ARM CPU would struggle to compete on performance with the highly sophisticated designs from Intel. It has succeeded in being incorporated into the majority of base band chips and running many mobile operating systems that don’t require heavy compute power.