Sat, Oct 25, 2014, 6:21 PM EDT - U.S. Markets closed

Recent

% | $
Quotes you view appear here for quick access.

Intel Corporation Message Board

  • ideal_invst ideal_invst Feb 13, 2012 1:25 PM Flag

    Android was fragmented because ARM was fragmented...

    Strikes me...in the absence of a completely integrated play like iOS devices (where Apple does processors and OS)...

    Other OS vendors like Google (Android and Chrome OS) and Microsoft (Windows 8) have to go with a one-processor strategy to avoid fragmentation. Working with multiple processors inevitably leads to:
    (1) Fragmentation
    (2) Varied/Compromised User Experience
    (3) Extended time-to-market (working with different strains of ARM processors)
    (4) Much higher development resources
    (5) Higher costs

    When Intel did not have low-power processors, OS vendors had no choice but to go to ARM incurring all the above negatives.

    But if Intel does deliver higher-performance processors with low-power and prices in the same ballpark (and further has a process technology far ahead of any other manufacturer), why would a OS vendor go with any other processor?

    Comments welcome.

    SortNewest  |  Oldest  |  Most Replied Expand all replies
    • II: I meant the latter - in terms of actual SoC differences. An OS vendor would have to dedicate more resources and do more testing to ensure customers have a good user experience across SoCs from multiple vendors - involving higher costs and extended time-to-market. Given a good alternative, I think an OS vendor would try to avoid this.
      -----

      Yes, it's more pain/cost for the OS vendor but they (MS) want choice and competition. Once the standards are set, MS should be able to push much of the pain/cost back onto the SoC vendors for future iterations (assuming WoA doesn't die a quick death).
      -----

      However, Google is into a wide range of devices. Smartphones, Tablets, Google TV...Also, Google is interested in extending its Chrome OS platform into the enterprise and into other home entertainment as well.

      It would be extremely difficult to have all these devices for each architecture. So it makes sense to evolve to one architecture. And that could be Intel if it can prove itself more on SoCs and have an extremely compelling roadmap.
      -----

      Yep, but the momentum is currently with ARM (with Google TV going to Marvell).

      -----
      II: Yes, they can...but would that be a sensible solution...trying to be another Apple?
      -----

      That was one of the idea's put forward by the analysts to explain the move. Although you wouldn't need that kind of licence to do that. It's probably for something else, perhaps the next xbox, or nothing at all:)

    • More on fragmentation from Charles Kindel, ex-Microsoft. Note he advises Google to focus on its Nexus offering which Google seems to be moving towards with the Motorola Mobility acquisition.

      Has Google 'Lost Control' of Android Thanks to Fragmentation?
      http://www.pcmag.com/article2/0,2817,2398944,00.asp

      Google has lost control of Android thanks to fragmentation, but that does not spell disaster for the popular mobile OS. On the contrary, Android will continue to thrive for many years to come, according to former Windows Phone exec Charles Kindel.
      ....

      There are five axes of mobile platform fragmentation, Kindel wrote: user interface; device; operating system; marketplace; and service. It's OK to have a "degree of fragmentation" along these axes (like Apple and iTunes), but "the fragmentation of Android is severe" across all five, he said.
      ....

      So what can Google do? Ultimately, nothing, but Kindel still urged the company to invest in the Nexus brand and hold back access to Google services, including Ice Cream Sandwich. However, that's just a pipe dream since ICS is open-sourced. "More fragmentation. Simply. Will. Not. Work," Kindel wrote.

    • No one in their right mind would risk their business on ARM's fabrication capabilities. Two choices, TSMC and Samsung, and neither will have 14nm for years and years and years.

      Intel will have it next year...

    • =======

      Or, over time, do they
      plan to design and build their own ARM SoCs and/or acquire an ARM SoC vendor?
      To me, Microsoft getting into SoC manufacturing seems to be a huge risk...and so it
      appears that they are muddling through without a clear plan.
      -----

      Microsoft has an Architectural licence so it could build it's own ARM CPU from scratch it it wanted to.

      =========

      II: Yes, they can...but would that be a sensible solution...trying to be another Apple?

      =========


      This is where the importance of the Intel reference design and pre-qualification of the ODM is important. Any new market enterant is faced with the choice of hiring or assigning the staff, or making a phone call to place an order.

      If MSFT were going to build anything from scratch, whatever they built would have to be better than the Intel reference design. The Intel reference design could put on the market in Q2 with a MSFT phone call to the ODM.

      The reference designs provide a low impedance entry into the market and make the choice to build ARM from scratch harder.

    • Appreciate your methodical response. My thoughts are below:


      - Does going with ARM lead to fragmentation?
      -----

      Depends what you mean by fragmentation. There is ISA fragmentation as you don't need to use the full instruction set. When Tegra 2 was launched it causes a few 'problems' as it was missing NEON (SIMD). The OS vendors need to dictate the minimum spec of supported processors. Tegra 3 now has NEON.

      Then there is SoC fragmentation (booting, devices, memory etc etc). Again a bulk of this can be specified by the OS vendor.

      =======

      II: I meant the latter - in terms of actual SoC differences. An OS vendor would have to dedicate more resources and do more testing to ensure customers have a good user experience across SoCs from multiple vendors - involving higher costs and extended time-to-market. Given a good alternative, I think an OS vendor would try to avoid this.

      =======

      - How big an issue is fragmentation? Does choosing Intel help avoid fragmentation?
      What is the value of avoiding such fragmentation?
      -----

      Adding another ISA increases fragmentation. Google now supports ARM, MIPS and x86.

      If in the long run you only had one vendor (say Intel) then this would reduce fragmentation.

      =======
      II: Yes, in the short term it does add one more ISA.

      However, Google is into a wide range of devices. Smartphones, Tablets, Google TV...Also, Google is interested in extending its Chrome OS platform into the enterprise and into other home entertainment as well.

      It would be extremely difficult to have all these devices for each architecture. So it makes sense to evolve to one architecture. And that could be Intel if it can prove itself more on SoCs and have an extremely compelling roadmap.

      =======


      Or, over time, do they
      plan to design and build their own ARM SoCs and/or acquire an ARM SoC vendor?
      To me, Microsoft getting into SoC manufacturing seems to be a huge risk...and so it
      appears that they are muddling through without a clear plan.
      -----

      Microsoft has an Architectural licence so it could build it's own ARM CPU from scratch it it wanted to.

      =========

      II: Yes, they can...but would that be a sensible solution...trying to be another Apple?

      =========

    • - Does going with ARM lead to fragmentation?
      -----

      Depends what you mean by fragmentation. There is ISA fragmentation as you don't need to use the full instruction set. When Tegra 2 was launched it causes a few 'problems' as it was missing NEON (SIMD). The OS vendors need to dictate the minimum spec of supported processors. Tegra 3 now has NEON.

      Then there is SoC fragmentation (booting, devices, memory etc etc). Again a bulk of this can be specified by the OS vendor.
      -----

      - How big an issue is fragmentation? Does choosing Intel help avoid fragmentation?
      What is the value of avoiding such fragmentation?
      -----

      Adding another ISA increases fragmentation. Google now supports ARM, MIPS and x86.

      If in the long run you only had one vendor (say Intel) then this would reduce fragmentation.
      -----

      Or, over time, do they
      plan to design and build their own ARM SoCs and/or acquire an ARM SoC vendor?
      To me, Microsoft getting into SoC manufacturing seems to be a huge risk...and so it
      appears that they are muddling through without a clear plan.
      -----

      Microsoft has an Architectural licence so it could build it's own ARM CPU from scratch it it wanted to.

    • >The shrink from 45nm to 32nm didn't result in any extra CPU performance

      now that you lost the ironics , you are more cute than before. Still you are lowballing me. So, you picked a shrink where lowering the power was first priority. We all (mm, ok not the ones who print "INTC REEEEEEEEEEEEEEEEED" LOL) know that you can give up performance to conserver power and vice versa. So you chose the example perfectly.

      ok, try Sandy to Ivy and exclude the GPU part. Let's talk just CPU improvements. See?
      -----

      I'm not sure what your point is as we're talking about Atom/Medfield?

      I used that example to show that shrinks by themselves do not automatically increase performance. When Intel shrinks Atom down 14nm it will have a transistor budget based upon the cost/power that it wants to hit. What will Intel spend this transistor budget on? Who knows, but it needs to boost CPU performance while still leaving room for more IP (baseband/Wifi etc) and better GPU's. It's a balancing act.

    • Then don't read Anandtech if it's just a bogus source of info...

    • CAN'T YOU READ?
      THIS IS WHAT I SAID 6%
      WHAT PART OF 6% YOU DIDNT UNDERSTAND

      6% IT IS. I SAW IT ON ANANDTECH

      6% NOW, 6% BEFORE , 6% AFTER, 6% ALWAYS

      SIX PERCENT. CAN'T YOU READ? THAT'S WHAT I SAID

      mwahahahahahahahahahahahaahahah

    • "The 4% to 6% estimate from Sept. 2011 is probably obsolteted by the Dec. 2011 actual measurement. Both sets of numbers were published but only one appears to be measured on system where the only thing changed was the CPU."

      Alright... but I also remember reading 6% in extremetech in Dec 2011.
      Anyway, saw 6% enough where it stuck...
      I didn't make it up...

      But tests are tests...
      and those test were with an I7
      Will see if I3 & I5 give similar test results...

    • View More Messages
 
INTC
33.18+0.49(+1.50%)Oct 24 4:00 PMEDT

Trending Tickers

i
Trending Tickers features significant U.S. stocks showing the most dramatic increase in user interest in Yahoo Finance in the previous hour over historic norms. The list is limited to those equities which trade at least 100,000 shares on an average day and have a market cap of more than $300 million.