Intel looks to build ultra-efficient mobile chips Apple 'can't ignore'

124

Comments

  • Reply 61 of 100
    majjomajjo Posts: 574member

    Quote:

    Originally Posted by bullhead View Post


     




    not sure what data you are looking at but the battery life of that intel based phone is horrible and no where near "toe to toe" with ARM.  Have fun charging your intel phone over twice as often as i will on my iphone.



    First of all, I'm speaking of medfield power efficiency. Second, I'm not sure what data you are looking at, but AT places medfield right in the middle of the pack:


    45987.png


    45988.png


    45986.png


    45989.png


     


     


    Quote:


    Normalizing for battery capacity, the X900 actually does a bit above average. In other words, the Medfield platform appears to be just as power efficient as some of the newer OMAP 4 based smartphones.


     


    Again we see reasonable numbers for the X900 but nothing stellar. The good news is that the whole x86 can't be power efficient argument appears to be completely debunked with the release of a single device. To move up in the charts however, Intel needs to outfit its reference design with a bigger battery - something I've heard is coming with the Z2580's FFRD. The normalized results put the X900 at the middle of the pack:



    So both the data shows, and Anand himself concluded that Medfield can go "toe to toe" with ARM.


     


    WTF data were you looking at?

  • Reply 62 of 100
    bullheadbullhead Posts: 493member

    Quote:

    Originally Posted by majjo View Post


    First of all, I'm speaking of medfield power efficiency. Second, I'm not sure what data you are looking at, but AT places medfield right in the middle of the pack:


    45987.png


    45988.png


    45986.png


    45989.png


     


     


    So both the data shows, and Anand himself concluded that Medfield can go "toe to toe" with ARM.


     


    WTF data were you looking at?



     


    guess you failed reading comprehension.  See the iPhone 4S at the top?  See how long the bars are?  Where is the intel based cloner phone?  Yeah.... So WTF are you looking at?

  • Reply 63 of 100
    majjomajjo Posts: 574member

    Quote:

    Originally Posted by melgross View Post





    I'd like to see a really good implementation of a dual core Cortex 15 later this year, with new Imagination 6000 based graphics for the iPhone (if possible). I would expect the same thing for the next iPad.

    Right now, two really good, fast cores will be enough. It remains to be proven the 4 Cortex9 cores will be better than two Cortex 15 cores. And significantly superior graphics has been very good for Apple, but it costs more, and other companies haven't been willing to take that cost beat down.

    But a year from now, 4 cores will likely, at least for tablets, be required. It could be more efficient for a 4 core device to do better multitasking, and tasks such as movie and photo editing than, 2 core devices. That, plus 3D apps, such as CAD, which I already use on my iPad, can use all the oomph they can get. And Apple, Adobe, Avid and others have multi year experience in using multi core devices to speed up the rendering in their apps. I expect them to be able to use this experience to quickly get up to speed (heh, pun intended!) on this on multi-core tablets.


     


    I think we're in that strange lull period that is the transition from A9 to A15, so I can't really fault samsung for going with a quad A9 design; A15 might have not been feasible when they were in the development stages of that SoC.


     


    People are quick to point to power consumption; but considering that Samsung has stated that each core can be individually power gated, I'm skeptical, at least in theory, that power consumption will that bad. Yes, when everything is going at 100%, power consumption would be higher, but the added performance also means a quicker "return-to-idle-state" ie. for a given task, the quad CPU will spend less time at 100% to complete it than the dual CPU.


     


    Of course, this is all dependent on the software being optimized such that it doesn't wake gated, idle cores when not needed.

  • Reply 64 of 100
    majjomajjo Posts: 574member

    Quote:

    Originally Posted by bullhead View Post


     


    guess you failed reading comprehension.  See the iPhone 4S at the top?  See how long the bars are?  Where is the intel based cloner phone?  Yeah.... So WTF are you looking at?



    guess you failed reading comprehension.


     


     


    Quote:


    we have to compare apples to apples though (pun unintended). because the xolo is running android, we cant directly compare to the iphone in terms of battery life, because we dont know how much of the difference is due to the os and how much is due to the soc.


  • Reply 65 of 100
    tallest skiltallest skil Posts: 43,399member

    Quote:

    Originally Posted by majjo View Post

    guess you failed reading comprehension.


     


    So run a new test with Android 2.2 on the iPhone. Until then we can't really say anything.

  • Reply 66 of 100
    majjomajjo Posts: 574member

    Quote:

    Originally Posted by Tallest Skil View Post


     


    So run a new test with Android 2.2 on the iPhone. Until then we can't really say anything.



    What? that doesn't even make sense.


     


    Do you guys understand the concept of a controlled experiment?


    When we want to test the effect of a variable (in this case, the SoC), we need to do it while keeping all other variables constant.


    That is why when we want to determine if intel can go toe to toe with ARM, we need to compare it to other android phones, to keep the OS variable constant.


    That way, we can be certain that any difference in power consumption is due to the SoC variable, and NOT the OS variable.


     


    And when we make such a comparison, the data shows that the intel phone is right in the middle of the pack.

  • Reply 67 of 100
    bullheadbullhead Posts: 493member

    Quote:

    Originally Posted by majjo View Post


    What? that doesn't even make sense.


     


    Do you guys understand the concept of a controlled experiment?


    When we want to test the effect of a variable (in this case, the SoC), we need to do it while keeping all other variables constant.


    That is why when we want to determine if intel can go toe to toe with ARM, we need to compare it to other android phones, to keep the OS variable constant.


    That way, we can be certain that any difference in power consumption is due to the SoC variable, and NOT the OS variable.


     


    And when we make such a comparison, the data shows that the intel phone is right in the middle of the pack.



     


    disagree. ARM clearly kicks the crap out the intel chip as evidenced by the ARM in the iphone doing the same tasks and consuming much lower power.  Intels chips have always been power hogs..and this shows they still are.  The tests are run across multiple platforms with multiple variables...there is no way to get an apples to apples comparison; you are comparing _different_ hardware architectures, after all.  In most cases there are 50% of android cloner phones way ahead of the intel cloner phone in power consumption.  Each of those phones has differing chips, circuit boards, displays, etc...  Tons of variables.


     


    My point, is an ARM implementation in the iPhone4s has over twice the battery life than the intel power hog.  Therefore, Intel is no where near "toe to toe" with ARM as there exists an ARM implementation which is over twice as efficient doing the _same_ tests.

  • Reply 68 of 100
    tallest skiltallest skil Posts: 43,399member

    Quote:

    Originally Posted by majjo View Post

    What? that doesn't even make sense.


     


    Run a new test comparing Android on this X86 device with Android on the iPhone to get the appropriate numbers.


     


    Quote:


    Do you guys understand the concept of a controlled experiment?


    When we want to test the effect of a variable (in this case, the SoC), we need to do it while keeping all other variables constant.




     


    Which this has not done, as the iPhone is running iOS, not Android. Put Android on the iPhone and run it against the Intel thing and you'll have closer to the right numbers.


     


    To have a truly controlled experiment, you'd use identical hardware SAVE for the X86 chip and ARM chip.

  • Reply 69 of 100
    drdoppiodrdoppio Posts: 1,132member
    Run a new test comparing Android on this X86 device with Android on the iPhone to get the appropriate numbers.


    Which this has not done, as the iPhone is running iOS, not Android. Put Android on the iPhone and run it against the Intel thing and you'll have closer to the right numbers.

    To have a truly controlled experiment, you'd use identical hardware SAVE for the X86 chip and ARM chip.

    Exactly! One cannot separate the effects of software and hardware. Therefore, one cannot claim that the Intel chip is less power-efficient. The lower battery life may be caused by Android and not by the SoC. OTOH, if you compare only phones running Android, you see Medfield doing quite well. So, most likely it is as efficient as your average ARM SoC.
  • Reply 70 of 100
    jragostajragosta Posts: 10,473member
    drdoppio wrote: »
    Exactly! One cannot separate the effects of software and hardware. Therefore, one cannot claim that the Intel chip is less power-efficient. The lower battery life may be caused by Android and not by the SoC. OTOH, if you compare only phones running Android, you see Medfield doing quite well. So, most likely it is as efficient as your average ARM SoC.

    That's a reasonable conclusion. However, it's irrelevant to the topic of this discussion for several reasons:

    1. Apple doesn't have the average SoC. They have tweaked the heck out of it and have one of the best energy efficiencies available. If Intel wants Apple's business, THAT is the target, not the 'average' SoC.

    2. Apple doesn't have any good reason to change. Simply catching up to Apple's chips isn't enough - if Medfield is no better, why bother changing and make the developers go through recompiling and debugging for a new platform? There has to be a significant advantage.

    3. #2 is even more true because Intel screwed Apple with their Ultrabook subsidies for Apple's competitors. Apple would have to consider whether Intel would do the same thing in the mobile space.

    4. There are other advantages to ARM besides the raw performance and raw power consumption. Apple is able to tweak ARM to meet its own needs. It is far less likely that Intel would let them do that.

    I just don't see any way that Apple would switch to Intel any time soon.
  • Reply 71 of 100
    Apple isn't ignoring you, Intel. They're farting in your general direction.
  • Reply 72 of 100
    majjomajjo Posts: 574member

    Quote:

    Originally Posted by bullhead View Post


     


    disagree. ARM clearly kicks the crap out the intel chip as evidenced by the ARM in the iphone doing the same tasks and consuming much lower power.  Intels chips have always been power hogs..and this shows they still are.  The tests are run across multiple platforms with multiple variables...there is no way to get an apples to apples comparison; you are comparing _different_ hardware architectures, after all.  In most cases there are 50% of android cloner phones way ahead of the intel cloner phone in power consumption.  Each of those phones has differing chips, circuit boards, displays, etc...  Tons of variables.


     


    My point, is an ARM implementation in the iPhone4s has over twice the battery life than the intel power hog.  Therefore, Intel is no where near "toe to toe" with ARM as there exists an ARM implementation which is over twice as efficient doing the _same_ tests.



    Once again, the problem with your reasoning is that you're attributing all of the difference in power consumption between the iPhone and the Xolo to the Intel SoC while dismissing the effect running a different OS has on it.

  • Reply 73 of 100
    charlitunacharlituna Posts: 7,217member
    It won't happen. Apple is full-steam ahead with ARM.

     

    Never say never. If Intel can create this promise chip and quickly Apple could switch
  • Reply 74 of 100
    Those unsubstantiated rumors @ Apple wanting to put ARM CPUs inside of their MacBook Airs are ridiculous. Apple dosn't want to cripple the MacBook Air success by outfitting them with smartphone processors. The Ivy Bridge CPUs are far more powerful, providing the raw number crunching power that their portable computers are known for. Only an idiot would think that Apple would trash their successful product to make a lousy netbook. Those died from Apple's 1 - 2 punch the iPad & MacBook Air. Google is now frantically trying to copy the iPad & Intel has copied the MacBook Air & rebranded it the Ultrabook. Apple is still leading the race

    Cheers !
  • Reply 75 of 100
    ksecksec Posts: 1,568member
    melgross wrote: »
    I think you're overly optimistic here. The 14nm node is going to be a doozy. 22nm is late by several months, and Intel will be lucky to keep to a schedule. We may not see production quantities until 2015. Even now, many mobile 22nm parts aren't available, and won't be for a month or two, possibly more.
    That doesn't mean that I wouldn't like to see Intel fabbing Apples parts. I would. But if it doesn't happen, I don't see the problem as being as bad as you are making it out to be, and Apple has advantages that intel doesn't have in making their own OS. Apple's devices have had a performance advantage for several years because of their own IP in the chips. This will just get better as Apple mods the designs more and more over time. Intel can't do that, and will offer, it seems, standard designs for mobile as they do now. All manufacturers using their chips will need to work within the parameters of what Intel sells them.
    But Apple's customizations will continue to become more extreme. That will allow them to continue to offer devices that have better battery life and better performance.
    Right now, Intel's new chip, if people will read the entire Anandtech review instead of quoting just small parts of it so that they would see it, offers just average performance overall. Nothing special. It's a good start, as it shows that Intel can move within the crowd. But they don't stand out in any way. It's expected that the new Cortex 15 SoC's will easily equal anything Intel puts out this year. And its not like the Cortex 15 is the end of the line either.
    Let's not get ahead of ourselves here.

    When Intel mentioned their Tick Tock Strategy, no one would believe it. And yet they have been executing it exceptionally well over the last few years.
    And i thought it was quite obvious 22nm wasn't late, It was the first tri gate Node, so that is complicated in itself, but the main point was they dont want to release it early! You just have to check the channels for the VAST amount of SandyBridge Laptop that are still in stocks and not sold out. Retailers, Distributors were all scrambling to let off these soon to become older models. The amount of Money people spend hasn't go up, but they are putting those money for their Phone or Tablet instead of a Laptop.
    P.S - And you have to blame Intel as well since their 32nm has been fabbing so well they decide to push more Sandy into the market then demand.

    The 14 nm has been going well according to investor's slide. And i believe so since its name has been changed from 16nm to 15 nm to 14nm over the years. So having a 14nm chip in 2014 May ( which is 2 years from now ) doesn't seems too far fetch at all.

    I deliberately left out any conclusion in my previous post. But it seems people are interpreting i am Pro Intel ( Which is not true :) )

    @ Wizard, A Single Core, Same MHz Cortex A15 is expected to beat a Dual Core Cortex A9 in many test while consuming less power. So you only have to look at Anand's Xolo performance charts and it is not hard to see why Cortex A15 will matter.
  • Reply 76 of 100
    drdoppiodrdoppio Posts: 1,132member

    Quote:


    Originally Posted by jragosta View Post





    That's a reasonable conclusion. However, it's irrelevant to the topic of this discussion for several reasons:

    1. Apple doesn't have the average SoC. They have tweaked the heck out of it and have one of the best energy efficiencies available. If Intel wants Apple's business, THAT is the target, not the 'average' SoC.

    2. Apple doesn't have any good reason to change. Simply catching up to Apple's chips isn't enough - if Medfield is no better, why bother changing and make the developers go through recompiling and debugging for a new platform? There has to be a significant advantage.

    3. #2 is even more true because Intel screwed Apple with their Ultrabook subsidies for Apple's competitors. Apple would have to consider whether Intel would do the same thing in the mobile space.

    4. There are other advantages to ARM besides the raw performance and raw power consumption. Apple is able to tweak ARM to meet its own needs. It is far less likely that Intel would let them do that.

    I just don't see any way that Apple would switch to Intel any time soon.



     


    Yes I acknowledged that previously in the thread. Apple isn't likely to abandon its own design and go with Intel. Just a guess though, if Apple did go with Medfield, it would likely have the chance to tweak and optimize it and match up the current performance of A5S, caeteris paribus. It doesn't seem from the data that Medfield is lagging because of inherent shortcomings in the design, but rather from being less optimized for the OS than A5S, for example.

  • Reply 77 of 100
    mdriftmeyermdriftmeyer Posts: 7,503member

    Quote:

    Originally Posted by charlituna View Post





    Never say never. If Intel can create this promise chip and quickly Apple could switch


     


    Hate to break it to you, but Apple controls the release cycles with their ARM SoC hybrid and it's ImgTec selection. They design, internally test and have it stamped out in massive volumes when they want to pull the switch.


     


    Neither IBM, Intel or even AMD offers this and Apple's been burned by decades of waiting for these fab companies to produce for the Desktop/Laptop market. Now that TSMC/Samsung/Global Foundries; especially TSMC and their working relationship with Apple only expanding globally it's rather clear that Intel will never get this business.


     


    The best Intel can do is make sure they don't lose their Desktop/Laptop business exclusivity to AMD.

  • Reply 78 of 100
    jragostajragosta Posts: 10,473member
    sandman619 wrote: »
    Those unsubstantiated rumors @ Apple wanting to put ARM CPUs inside of their MacBook Airs are ridiculous. Apple dosn't want to cripple the MacBook Air success by outfitting them with smartphone processors. The Ivy Bridge CPUs are far more powerful, providing the raw number crunching power that their portable computers are known for. Only an idiot would think that Apple would trash their successful product to make a lousy netbook. Those died from Apple's 1 - 2 punch the iPad & MacBook Air. Google is now frantically trying to copy the iPad & Intel has copied the MacBook Air & rebranded it the Ultrabook. Apple is still leading the race
    Cheers !

    Yes, I agreed - and pointed that out in a different thread.

    Even if ARM were 'good enough' for many people, I don't see Apple putting ARM into a MBA. How in the world do you explain to the market that you're releasing a product that's far slower than the previous version? You can't. ARM would have to exceed Intel's performance by a significant margin before Apple would switch - especially since Intel has proven to be an untrustworthy partner (just being 'as good as' isn't sufficient reason for a change).

    I can picture an entirely new ARM-based product - an iPad Pro, perhaps - which has a fold-out keyboard, but it would be a unique product rather than an MBA.
  • Reply 79 of 100
    derekmorrderekmorr Posts: 211member
    Intel has teased some specs for the next Atom SoC, the Z2580. It's supposed to be dual-core at 1.3 GHz (bursting to 1.8 GHz), with an SGX 544MP2 GPU at 533 MHz. It'll still be on the 32nm process, though. Anandtech did a writeup a few months ago: http://www.anandtech.com/show/5592/intel-atom-z2580-z2000
  • Reply 80 of 100
    solipsismxsolipsismx Posts: 19,566member
    derekmorr wrote: »
    Intel has teased some specs for the next Atom SoC, the Z2580. It's supposed to be dual-core at 1.3 GHz (bursting to 1.8 GHz), with an SGX 544MP2 GPU at 533 MHz. It'll still be on the 32nm process, though. Anandtech did a writeup a few months ago: http://www.anandtech.com/show/5592/intel-atom-z2580-z2000

    Unfortunately that won't be out until the 1st half of 2013 (potentially more than a year away) and the results don't look good enough to usurp ARM's throne. I'm sure Intel will get some vendors on-board, it might be an ideal fit for Win8 tablets, and it's certainly a lot closer than they were to ARM a year ago but will be good enough?

    Another issue I see is the complexity of the ASIC which could be a deterrent for companies like Apple. Apple licenses the ARM reference designs and then build's their own ASIC according to some general specs thus making it a unique and ideal chip that suits their specific needs. I don't think they can technically call it Cortex with the number of low-level changes they make. Will Intel offer them the same options as ARM? Not from what I've seen.

    PS: Your link doesn't work because the period at the end of the sentence is being attributed to the link.
Sign In or Register to comment.