Recent

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

Intel Corporation Message Board

  • dnenni dnenni Sep 3, 2011 8:00 AM Flag

    TSMC/ARM versus Intel/Atom

    SortNewest  |  Oldest  |  Most Replied Expand all replies
    • I didn't see the ARM based tablet running in the video. It wasn't identified as an ARM tablet.
      ...
      I think that Windows 8 for ARM will be a non-starter. If you want to run Office on it, you will have to buy it. Many people are stuck on their older versions of Office which many Excel users for example like better. When their older x86 legacy programs won't work, I'm afraid they will simply take the tablets and possible laptops back for a full refund. There is a small possibility that they can get something to work, but for the most part Microsoft has a long, long history of poor product launches that get canned a couple years down the road.
      -------

      That was an ARM tablet. Nvidia say it's one of it's SoC's (can't remember if it's the Tegra 2 or 3).

      I agree I think ARM on (full) windows 8 will be a none starter. The slate could be a very interesting product if the price is right and MS 'bundle' office as a freebee.

      We still don't know what MS's strategy for windows 8 will be. We dont know if MS will support fat binaries (which would run on both x86 or ARM) or it could include an x86 emulator for ARM to run some legacy x86 code. It might do nothing at all and confuse the hell out of consumers.

    • ltisteve@verizon.net ltisteve Sep 5, 2011 2:36 PM Flag

      I didn't see the ARM based tablet running in the video. It wasn't identified as an ARM tablet.

      Yes, I know that Microsoft has planned to release Office for ARM based devices. Before you all get too excited about that fact, keep this in mind. There was a time that Microsoft told Intel that their processors weren't good enough and they wanted to develop an operating system for high end work stations. The resulting operating system was "Windows NT" and it was designed for the superior RISC processors. At the very best the RISC processors had a 15% performance advantage over Intel.

      Microsoft had two code bases running at the same time. One was on the Windows 95 platform, the other was on the Windows NT platform. When Microsoft figured out they were wasting their time they decided to merge the operating systems. The new product was called Windows XP. They planned a huge introduction party in November of 2001 of all places NEW YORK CITY. After September 11th it was a somber, scaled down event.

      I think the same industry analysts who criticize Intel for not having a solution to the ARM processors for smart phones also criticize Microsoft for not having a realistic solution for Android and IOS. Windows Phone was a day late and a dollar short. Windows 8 for ARM will be there should this market take off. And, yes you can give credit for Microsoft for being ready for this potential market.

      As an independent software developer, well, you are in another situation all together. If you are Adobe for example you will need to rewrite the code base for Photoshop, Premier, Illustrator and the other 13 large titles. That's a huge task. And it would take your resources away from improving their programs.

      I think that Windows 8 for ARM will be a non-starter. If you want to run Office on it, you will have to buy it. Many people are stuck on their older versions of Office which many Excel users for example like better. When their older x86 legacy programs won't work, I'm afraid they will simply take the tablets and possible laptops back for a full refund. There is a small possibility that they can get something to work, but for the most part Microsoft has a long, long history of poor product launches that get canned a couple years down the road.

      If you want to bet on a long shot the Nokia/Microsoft new phones may be able to spark the market. Currently they are working on a modified version of Windows Phone called "Mango" that will hit the market sometime soon. If they can sell untold millions and millions of Windows 8 on smart phones and open the market up to a third strong phone OS, maybe Windows 8 for ARM will last more than a couple of years and will catch on.

      I am about 95% certain that Windows 8 for ARM will be another on of Microsoft's transitional operating systems that is here today, gone tomorrow.

    • When these devices are actually available in the marketplace at some point next year, the price difference will be much smaller.

    • Are u kidding me? If someone tried to give me that piece of garbage to do my work, I'd laugh at him. Its super-slow, error prone, an interface designed for young teenages (not professionals) etc. I can get the same look-and-feel, size, power, etc. with MUCH better & reliable performance with an ultra-book. Win8 looks to be garbage based on the reviews I've been reading. Another bust for Microsoft.
      -----

      You may be right about the bust for Microsoft...

      You are right about the performance of the ultra-book over this device. That said, the ultra-book will retail for $1000 while that device under $400.

    • Are u kidding me? If someone tried to give me that piece of garbage to do my work, I'd laugh at him. Its super-slow, error prone, an interface designed for young teenages (not professionals) etc. I can get the same look-and-feel, size, power, etc. with MUCH better & reliable performance with an ultrabook. Win8 looks to be garbage based on the reviews I've been reading. Another bust for Microsoft.

    • TSMC better do something...anything. Intel is months away from carving their business up. As if that isn't enough to keep the TSMC boys awake nights, Global Foundries/Samsung is also targeting the TSMC foundry business.

      ARMH also better figure out a way to protect themselves. If low power is ARM's main claim to fame, that advantage is months away from being erased by the Intel 22nm Trigate process. Beyond a certain level, lower power ceases to have value. Intel will pass that threshold before the end of the year.

      After low power is no longer an issue, function will continue to have value. Intel/McAfee have some hardware security tricks up their sleeve that ARM and TSMC will NEVER be able to match. These goodies will be proprietary, valuable, and patented.

      Then we have the Solid State Drive revolution dictated by the physical configuration of tablets, Macbook Airs and Ultrabooks. Intel is the only semiconductor company in a position to integrate the CPU, SSD, and DRAM in a single "Compute Module" package that will provide great performance, and a power level below the threshold of concern.

      High speed communication between the sections of this Compute Module will be accomplished through the Thunderbolt interface that was jointly developed by Intel and Apple..


      Intel is planning/building an additional $40-50 billion worth of annual fab capacity. Filling this with technology that can't be matched will cause pain across a wide swath of players from disc drive companies to favless semconductor companies such as QCOM, TI, NVDA, etc.

      I'll take the Intel side in this wager.

      JMHO of course. Long 8000 contracts of Intel LEAPs.

      • 1 Reply to semiwiz2002
      • ltisteve@verizon.net ltisteve Sep 4, 2011 5:44 PM Flag

        That's a good point about Global Foundries going up against TSMC. I find that there are three different aspects of the ARM processor.

        1. ARMH- They do all the R&D
        2. Marketing firms - NVidia, QualCom, Motorola, TI
        3. The Fabs

        It's true that the marketing firms do a little spin on the original ARMH designs, but in the end they are the ones who take the CPU to market. I have been critical of this aspect of the model because this is the spot where the income is most variable. I am under the impression that Nvidia for example has been pushing it's Tegra 2 ARM cpu at very aggressive prices. It's starting to turn off the other players such as Qualcom.

        The fabs tend to do fairly well because there are fewer of them and they have a bit more control over how much they will charge. The problems I see with the fabs are.

        #1. Enormous capital equipment expenses- It's very expensive to be state of the art and with each new investment there is more pressure for a healthy ROI.

        #2. New competition- Especially from Global Foundries. With new contracts there will be added pressure to win contracts which may lead to reduced margins.

        #3. 28nm- Global foundries is racing down to 28nm. When they got to 40nm they experience problems that I am not entirely certain they fixed. I am not confident that they can get to 28nm processors and make them profitably.

        Bring Intel into the mix, yet taking more business away and you have less margin for error. There is yet another hand in the cookie jar.

        This subdivision of fabs on the ARM side makes me think that not only will they find it harder to compete but also make the huge investments to improve the fabs in the future. Lower sales margins makes it harder for them to invest in their fabs in the future.


        That's why when you look at the ARM model carefully it comes off looking piecemeal. There are a few major weak links in the chain. Not enough for them to crash and burn, but enough for them not to be able to ramp up effectively against Intel.

        The ARM fans have always managed to turn the argument toward power consumption. I believe that the argument will come down to R&D and investment in infrastructure.

        I'm sure the fine people at ARMH can figure out how to make newer generations of processors. But if they call for investments in machinery that Global Foundries and TSMC can't afford, they will fall behind. If they can't master the manufacturing processes that is the most frustrating because it cuts into profitability of contracted jobs. If they can't make the processors cost effectively that will be another problem.

 
INTC
31.57+0.08(+0.25%)May 27 4:00 PMEDT