Let’s see what WWDC has. What with me getting appointments duplicated and other hassles, I think Apple has a lot to prove before these ideas are feasible. (and the early status og the I/O betas are as incomplete as GoogleTV was a year ago.)

re: cost of shuffling data around
How can these costs be any different from what netflix and hulu and youtube HD do? Playing a cloud World of Applecraft for 2 hours can’t be significantly different then watching a 2 1/2 hour movie on netflix.

re: cloud not for processing
Try OnLive. The quality is way better then expected and 2 years down the road I can certainly see this being ready for prime time and the entire software market (not just games) being ripe for someone like Apple to come in with a massively disruptive business model.

Again I might be wrong. Thin client consumer laptops may be 10 or 5 years away. But watching Google’s IO I’d say we’re a lot closer.

Presumably they will sell you a cloud solution so your ultralight, ultraportable, ultra long lasting macpad pro can also muscle through tasks in a fraction of the time that a high end local chip would take.

“I think the notion of an ARM controller for a laptop that occasionally wakes up OSX for the heavy lifting is not too bad an engineering challenge… but a UX disaster; the user has to form a mental model of what is going on underneath that screen, what’s in charge at any time.”
.
But, does it need to be a UX disaster? Consider:
.
1) Reports are that Lion provides “gentle persuasion” for the user to move away from the melange of MOWs (Multiple, Overlapping Windows) of OS X towards the single, “in your face” window (with panels) of iOS (like on the iPad). Another Lion feature is the automatic saving of files (in addition to overt saving) where the user need not be concerned with losing files or changes.
.
2) Most laptops have 17″ or less displays — so the MOW capability is less practical on this form factor.
.
3) As suggested by @Eric Dujardin, the ARM would be responsible for the HMI, the UX (UI, View Layer, Presentation Services, whatever) and route “heavy” processing to the x86 as needed.
.
So, one use case is a laptop with a few concurrent apps running — with a tendency for a single app to monopolize the display and the UX at any one time.
.
Say, the laptop is number-crunching a large spreadsheet and/or rendering video on the x86. The user could switch to another window/app to surf, check email, etc. knowing that he will be notified when the x86 apps complete. These “light” apps (surfing, email) could be handled entirely (or mostly) by the ARM (transparent to the user).
.
Should the user finish before the x86 processes complete, he could switch back to, say the spreadsheet window (handled entirely by the ARM HMI. Or, just let the ARM put the display to sleep. As the x86 processes complete, the ARM updates the app windows to reflect this.
.
Without further action from the user, the ARM puts the x86 to sleep, then “instant OFFs” itself..
.
Later, when the user returns, he clicks or taps and “instant ONs” the ARM which shows the app windows just as he left them — updated to reflect completed processes.
.
While the user readjusts to what he was doing and where he was — the ARM wakes up the x86 in anticipation of continued work,
.
.
This is complex to explain, but could be implemented in such a way that it is very intuitive — and the user need not be concerned with what’s happening under the hood.
.
Requisite to this type of solution are better task switching and better notifications (at the very least). No longer would it be acceptable for the user have to select from every app he’s ever run — nor be expected to respond to a modal dialog from a hidden window..
.
The above could be a refinement of the iOS task bar combined with the Command-Tab app switcher — together with an elegant system-wide notifications. Why must I switch to Final Cut, just to see if the render is complete (or not)?
.
We should have [system] apps for that!.

Additionnally new compilation techniques (see Google Native Client use of LLVM) allows late compilation so that the translation to either x86, x86-64 or whatever flavour of ARM architecture could theoretically be done at installation time.

@ Dick Applebaum: bi-processor architectures are hard to program. To deal with it, I’d say the ARM would act as main processor, managing the HMI, delegating intensive-CPU tasks to the x86. Even imagine the x86 acting as a “local cloud processor”, so that the ARM chip would run iPxD-like apps and the x86 would act as the remote server.

Sometimes you go beyond commentary or reportage and add a little unexpected ‘joy and delight’ of your own to your articles – to quote a little from Tim Cook. When I read your reminiscences of the ubiquitous OC71, I almost fell off my chair. We must be of an age + or -. As a young man I persuaded a girlfriend to ask her friend who worked at Siemens in Paris to get me a ‘bunch’ of silicon. I used American part numbers for some transistors and early i.c. gate devices. What I received, among other things, were about 50 Siemens branded OC71s, with their little metal helmet tops. I showed all the signs of a geek in the making, but it was not to be.
Back on topic though, allow me to float a thought or a line of twentyish thoughts.

Many here have said that Apple must move in directions that no competitor can follow.
Hardware ideas are not that direction, since most hardware componentry can be acquired by any competitor at a price.It is not a game of hardware specs. Software is the killer, as you have said.

So, imo, software is the only path where Apple can walk in ways that others can only talk about and struggle to follow. But how to get Apple there, from here?
Well, you have said that ‘He who must be obeyed’ likes a simple world in which less, with some gifted engineering, can indeed become more; sometimes much more.
Apple has two computing platforms: Mac OSX and iOS.
The latter was derived from the former.
Two is more than one, and less is always better.

From the outset, Mac OSX was offered on Power PC but also ran on Intel chips in the lab until the day arrived to make the switch.
iOS devices have a modus operandi that is all the rage in the world and the obsession of all of Apple’s competitors. Everyone says that iOs is the future and the future is post-PC.
The Mac is PC too, so do we anticipate its decline and eventual demise?
Well, we don’t because many people will continue to need a powerful workhorse; the truck that SJ spoke about at All Things Digital.
Less is more.
The son can be the father to the man as the latter ages.
There are many signs that iOS features are being implemented in Mac OSX 10.7 Lion.
Apple seems keen to wean its users away from mice and keyboards for common computing tasks.

Is it possible then that Apple can use its software assets to put itself beyond easy competition for at least five years by doing something rather simple?
iOS came from MacOSX and so there is every reason to believe that there are iOS Macs running in the Lab.

So my point is this.
Instead of a dual chip (x86 and Apple A’X') on the motherboard, why not a dual OS Mac (say an iMac to begin with) that switches from OSX to iOS and back again seamlessly, at the flick of a ‘switch’ or just by tilting the screen from the vertical to the horizontal?
How many iOS customers who don’t own a Mac might be tempted to reach for their wallets to own a large-screen computer on which they get added ROI on their app investments and their iOS skill sets?
And who could compete with that in a hurry?

Just my 2p worth Jean-Louis.

À la semaine prochaine!

Its amazing so many PC and Mobile OEMs do not understand this. …even now.

-Tek

I don’t think it hurts Intel’s ego to fab Apple’s SoC’s. Less than 50% of A5 is the ARM cores. Apple still needs GPU, I/O, RAM, flash, and Intel does all of that, too. And Intel’s ego is attached to volume. In 2015, there will probably be more iPads sold than all notebook PC’s, so does Intel want to see someone else making more PC chips than them, or do they want to do both the notebooks and the iPads?

Apple is about halfway to saving up to buy Intel anyway. Don’t put it past them.

> Do ‘the rest of us’ really need workstation class
> processors and operating systems for our ‘personal
> computing’?

Yes. History shows that we do. The operating system core in an iPhone 4 is workstation class OS X, and iPhone 4 has more computing power than the fastest Mac workstation from a decade ago. And this year the computing power in the iPhone will double and the operating system will improve again.

An iPad with Retina Display is a 2K screen, an iMac with Retina Display is 4K, and both have secondary screens to run. 4K movies are already supported on YouTube. We routinely use large maps and other datasets. We need much more computing power as we go forward. We’re just getting started.