Intel Looking to Cut Power Consumption on Future Ivy Bridge Chips

[KnightWRX]

I mean virtualization compatibility as now possible between Mac and Windows or Linux on x86 via VMware Fusion on Mac (run Windows on Mac), for instance.

You want to run virtualization software on an iPad ?

Considering the iPad has 2002-2003 level desktop performance, how do you expect to run software made 10 years later through virtualization ? Not to mention the iPad/iPhone doesn't quite have the input mecanisms to deal with the desktop OSes of then or now.

Or are you talking about running iOS apps on OS X ? Now why would we want to do that ? Bunch of throwaway 1$ apps that while useful on a cellphone, all have much better alternatives on the desktop.

 

[tdtran1025]

Given the current market pressure and the double life Apple had lead in the past, this scenario might be possible. However, when one reads into this, the nuance is that the A6 has so much room to grow; and pending the imminent release of 64-bit ARM next year, the progress forward for Apple is non-regressable. IMO, Intel might be just the fabricator for A series and x86 chip vendor for iMacs.

 

[ifij775]

I'd like to see Intel building Apple's ARM processors way more than Apple using an Atom chip in their product line.

 

[deconstruct60]

Intel could just dust off the ARM fab it once had :

Actually Intel replaced StrongARM with XScale. DEC largely didn't StrongARM. (some of those folks were in one of Apple's acquisitions ). Intel sold XScale off to Marvell.

Technically I think Intel is still holding onto a ARM architecture license but they'd have to ramp up a design team. They don't have any significant ARM design going on right now (other than in microcontrollers for other support chips ).

I doubt Intel is going to go off and become a "for hire" ARM foundry for anyone. They are probably interested in something where they have a "value add" contribution that goes a bit beyond just giving someone access to their "one year ahead" process technology. Customized x86 SoCs are far more likely where intel does the x86 cores and someone else contributes other components of the SoC and the whole thing is fabbed on Intel processes.

 

[Chupa Chupa]

But the Surface RT is a "Windows" machine that doesn't run classic Windows apps....

As long as Windows 8 , Windows RT, and Windows 7 combine to make Windows XP (and Vista) disappear Microsoft is on track to continue to be successful.

If Intel's Atom line continues to make steady, substantive improvements some RT ARM designs may switch back to Windows 8. If not then ARM will get more design wins in 2013-14 as the RT oriented software library builds.

Understood RT is a different beast than the pro version. But it's meant to compete with the iPad and Android, but consumers are not biting.

Large/mid companies are not going Windows 8 in large part, having just made the move to 7 from XP. Consumers are ditching computers altogether in droves, opting for tablets, and the only box maker with any traditional computer sales growth is Apple.

Microsoft isn't fading away, to be sure. It's just going to continue as a legacy brand. It sells plenty of copies of Office and Windows 7, not to mention XBOX, to keep it sustainable for a very long time. But it's not going to grow like the old days. It's best days are behind it. It's stagnant stock price is testament.

 

[KnightWRX]

Technically I think Intel is still holding onto a ARM architecture license but they'd have to ramp up a design team. They don't have any significant ARM design going on right now (other than in microcontrollers for other support chips ).

A design team like Apple's.

Just saying Intel already has ARM experience, maybe it's not current (especially since the StrongARM/XScale stuff is quite old) but it's something they've already pulled off.

Apple has plenty of cash, I'm sure they could persuade Intel into fabbing for them, in exchange for a lucrative Mac long term contract for x86 stuff.

 

[prowlmedia]

iPads with x86 inside would be fantastic for full Mac compatibility!

You are forgetting, the entire reason iOS exists is because a Desktop OS doesn't work on a tablet. You can either design from scratch, like iOS or Android or make stupid compromises like Windows 8.

at the very least there is is no hover / right click etc - the paradigms are all very different.

And 99% of OSX software involves menus and mouse interactions that just won't work with your finger.

That all said - I reckon the Future OSX 11 will be and amalgamation of iOS and OSX with touch from the outset - possibly Gesture and Voice. But it'll still have a keyboard and standard interfaces like wacoms and mice - because to do most kinds of work you will still need direct input.

 

[Mattsasa]

I thought they would be trying to increase power consumption.....(sarcasm)

I think I would prefer x86 architecture on iOS and MAC. Opposed to ARM

 

[griz]

Maybe we will finally see this.

http://www.maclife.com/article/feature/our_favorite_tablet_mockups_and_patents

Although, I doubt it unless the ipad thins down to less than 5mm.

 

[nutjob]

How do you figure ? Mac compatibility has nothing to do with instruction sets. The problem is frameworks. iOS exposes UIKit as a high level framework, OS X exposes AppKit. Both are quite different.

So, you load the frameworks, pretty easy. Third party software requires binary compatibility, so instruction sets are pretty important.

----------

I thought they would be trying to increase power consumption.....(sarcasm)

I think I would prefer x86 architecture on iOS and MAC. Opposed to ARM

You prefer 1 to 2 hours of battery life?

 

[deconstruct60]

Understood RT is a different beast than the pro version. But it's meant to compete with the iPad and Android, but consumers are not biting.

Actually no. The RT devices are meant to compete against the more classic Windows usages. There are and will be no RT phones or 5" devices. 7" devices are pretty unlikely also. Those will be covered by Windows Phone which is an OS competitor to iOS.

RT should sweep up the remnants of the netbook devices with a much better form factor and cost basis. Initially they are priced too high to do the 2nd and 3rd iterations will likely do it for the most part. The sub $300-350 market is likely largely lost to "phones and smaller" devices.

So instead of dumping that netbook for an iPad you dump it for Windows RT. The Surface + keyboard is far closer to being a netbook than an iPad once the pricing gets lowered.

Large/mid companies are not going Windows 8 in large part, having just made the move to 7 from XP.

Large and mid companies don't move to anything. The organizational inertia typically makes them slow, risk adverse adopters of almost everything (except high bonus checks for executives) . So the specifics of RT and Windows 8 are immaterial.

 

[Panther Al]

*snip*


Apple has plenty of cash, I'm sure they could persuade Intel into fabbing for them, in exchange for a lucrative Mac long term contract for x86 stuff.

Well, if Apple still has has 100b in the cookie jar, they could just buy Intel - the market cap for Intel is only 97b - Do they take checks?

 

[KnightWRX]

So, you load the frameworks, pretty easy. Third party software requires binary compatibility, so instruction sets are pretty important.

Load the frameworks... how exactly ? 3rd party software is built on those frameworks, otherwise, iOS/OS X software would be just a recompile away.

No, it's a much bigger issue than just instruction sets.

----------

Well, if Apple still has has 100b in the cookie jar, they could just buy Intel - the market cap for Intel is only 97b - Do they take checks?

I really doubt you've thought this through.

 

[lakshwadeep]

As long as Apple puts the best components in their products, either from external suppliers or their own factories, I'm satisfied.

 

[wizard]

Apple keeps the specs tight to the chest.

So the reality is nobody really knows. I would imagine at full performance it probably runs at 2-5 watts.

I can't find the specs for the current A6 or A6x wattage... It has to be much lower than 10watts....

 

[deconstruct60]

A design team like Apple's.

Intel being purely just a foundry for someone isn't likely going to happen. That would be an indicator of Intel selling away the crown jewels. Intel is not a "job shop" and there is no way they are going to make margins like they currently do being one.

Just saying Intel already has ARM experience, maybe it's not current (especially since the StrongARM/XScale stuff is quite old) but it's something they've already pulled off.

Actually not. Typically when sell off the design and the product the key designers associated with the products also go along with it. When companies sell off designs along with people who did them. So they also tend to also sell off the knowledge in those people's head.

Intel has experience putting CPU oriented transistors on their process tech. But ARM specific optimizers they are not. They could give Apple tools/guides they give to their own x86 teams that aid in getting the most of out the process technology. But that would like Apple handing over sources to key components of iOS and OS X to Intel. That's not the way they typically do business.

Apple has plenty of cash, I'm sure they could persuade Intel into fabbing for them,

One reason Apple has buckets of cash is because they squeeze their suppliers. In other words, they don't spend money unless they have too. There are two other large and far more open Fab vendors out there. Waving their money at them would work much more effectively than waving it Intel. Because Intel is going to want much more money then those folks.

in exchange for a lucrative Mac long term contract for x86 stuff.

It is likely not in exchange but a courtship dance both sides are doing. We "could do" ARM stuff, but in the end it is really about exploring ways to keep the x86 contracts alive and growing.

 

[daywiz]

Smart move by Intel, if its true.Securing a deal from Apple gives them two benefits: 1) immense revenue 2) enough ideas, experience to create Similar to ARM based chips. Apple will indeed relax and may give INtel the job till TSMC is fully capable..but no way an INtel chip is making the cut to the ipad unless u r talking about a repackaged apple designed custom chip that incorporates sum elements of intel chips. I dont see an ios - osx merge ..at least till 2015. apple is smart to claim an opportunity. If intel is willing..apple will utilise them. I am sure intel wudnt mind 2 billion worth of revenue

 

[wizard]

Intel is actually competitive on performance per watt now.

Anyone have any data on the current power consumption of the A6 or A6X chips?

I would also love to see what the computing power/watt is for both IB and the A6/A6X. Now that Apple is designing their own chips, I find it interesting that Intel would be a possibility for future iPads/iPhones. Are they saying Apple would then design based on the x86 architecture? Would Intel even allow that?

With the advent of Ivy Bridge intel actually has better or competitive performance per watt. It is the fact the ARM chips are lower performing that gives them an advantage in wattage. Note this is based on non A6* processors, the additional performance seen in Apples newest chips apparently give a significant boost per watt.

I forget exactly where I saw the testing that revealed just how far ARM has come. Maybe it was on Phoronix. In any event Intel has come a very long way in the computational capability per watt.

----------

I believe they make chips for one of the gate array makers. The difference here is that the margins on such chips is still Intel like. In the case of ARM and Apple they would not be willing to pay Intels traditional margins. This is what makes the idea of Intel as a foundry difficult to digest. About the only positive here is that Intel is going to have idle plants that need to be paid for.

Does Intel produce chips for anyone else? I would find it interesting that Apple would have gone to Samsung in the beginning considering Intel is a whole process node smaller than everyone else (at the moment).

 

[Giuly]

...and if you think about that the A6x costs about $20, while an equivalent from Intel would range anywhere between $50-$150, you immediately know why this story belongs to the realm of DigiTimes.

It's much more likely that a Mac will run a beefed-up iPad chip someday, not the other way around.

 

[Eidorian]

...and if you think about that the A6x costs about $20, while an equivalent from Intel would range anywhere between $50-$150, you immediately know why this story belongs to the realm of DigiTimes.

It's much more likely that a Mac will run a beefed-up iPad chip someday, not the other way around.

Intel took 4 years to get Atom to effectively compete with ARM in wattage and price and that's still using the same architecture from back then. You could get sub-10W Atoms in earlier revisions but the prices were outrageous. 32nm Atom can actually pull it off in the ~$40-60 range.

Failing that people completely overlook ULV Celeron and Pentium models available to OEMs are cut-rate prices. Sure they lack the "Core" marketing name but they are based on Sandy or Ivy Bridge and are dual core.

 

[reallynotnick]

Likely, they're going ARM. Apple has purchased it, and if they weren't going ARM, that's a LOT of wasted money. Intel's time will likely be up in 2014. Then all the Macs will be on ARM processors, and OS Xi (or XiOS, pronounced Zeye Oh Es).

I'm willing to take that bet. iOS devices are Apple's big money maker right now so custom designing ARM processors isn't a big waste of money.

Intel has to slip up really hard for Apple to transition their Macs to ARM, looking at their roadmap that isn't going to happen unless they miss their goals by a long shot.

 

[vanzantapple]

Intel BE worried bout ARM in *****.

 

[iMacFarlane]

Intel architecture in an iPad . . . hmmmmmm . . .

Bootcamp option? Windows 8 on the iPad?

You show me an iPad with the ability to run iOS apps and Windows 8 apps, and I'll show you the end of every other tablet currently in production.

 

[dshan]

Different processors for iPhone and iPad makes no sense at all! Universal iPhone/iPad apps would then need to include two different binaries as well as just different UIs as now. Bigger apps and other unpleasant issues would also result.

A Rosetta-like ARM-to-x86 translation layer would be needed on the iPad for several years, with RAM, battery life and performance hits that would detract from the iPad's popularity.

It would be bad enough switching all iOS devices over to x86 CPUs, but just the iPad would be even worse.

 

[Giuly]

Intel took 4 years to get Atom to effectively compete with ARM in wattage and price and that's still using the same architecture from back then. You could get sub-10W Atoms in earlier revisions but the prices were outrageous. 32nm Atom can actually pull it off in the ~$40-60 range.

Failing that people completely overlook ULV Celeron and Pentium models available to OEMs are cut-rate prices. Sure they lack the "Core" marketing name but they are based on Sandy or Ivy Bridge and are dual core.

Actually, current Ivy Bridge chips don't come below 17W, with clocks ranging from 1.7 to 2.0GHz, so that should be the magic frontier of them. I don't remeber the IPC on those things, but clocking a CISC chip lower than a RISC chip while maintaining the same performance should be hard to do. It didn't work when they competed with the PowerPC architecture, and I doubt that it would work now. The problem with CISC designs is that they're just not as efficient compared to RISC, all things equal.

Anyways, ARM-Cortex A15 wipes the floor with the Atom chips.