Page 2 of 2 FirstFirst 12
Results 21 to 23 of 23
  1.    #21  
    Quote Originally Posted by NickDG View Post
    This part is confusing. A smaller die size does not make a CPU faster. What it does do is use less power, which in turn produces less heat and allows it to be clocked higher.

    Case in point, a 45 nm single core A9 clocked at 1ghz will have the same performance as a 32 nm single core A9 clocked at 1ghz, but it will run much cooler and use a lot less power saving battery life.
    Nick;

    I wasnt implying that it makes it faster, I was stating that the 32 nm size was indicated to have a 50% performance increase over their own 45 nm sized chip - as they have stated on their site, but, Im sure, as you suggested above, that there are changes to the chip design other than the smaller sizes to provide such an improvement.

    Last edited by LCGuy; 09/11/2010 at 06:19 AM.
    "The more I learn, the more I realize just how little I really do know!" -Albert Einstein

  2. #22  
    Quote Originally Posted by Mike5 View Post
    No problems. Though I did think the yawn was directed at me, I meant it when I said I have learned a lot from you & LCGuy. Keep the discussion going!

    Let me ask you all, though, how far in advance would they have to select & contract/order a processor for a new device? Even if it was still in development? I work in the airline industry & we were promised a lot by Boeing & the 787 Dreamliner--about 3 years later than scheduled, we and the world still await the airplane.

    Thanks for the encouragement LCGuy.

    Cheers
    It depends on what stage of development the device is in. As someone who works in the industry, I can tell you that for most embedded processors, the time period between delivering the first samples to the customer and full production ramp for end-user products is generally at least one year and can be up to two years. I've seen devices take even longer than that especially when a new process node is involved, like going from 45nm to 28nm. That period constitutes a long post-design development process including test development, reliability testing, product qualification etc. etc. before they can get to the point that the product is a) safe and ready for use by the masses and b) is cost-efficient enough to manufacture on a large scale. Of course, if that development period is already complete, then orders should generally take just about 3 months (standard fabrication/testing period), although with a premium payment you can make that go faster.

    In the case of the OMAP4, I'm pretty sure they've completed the development period on the first iteration and are already shipping the product out to customers for building end-products based on all the information that has been published.
  3. #23  
    Quote Originally Posted by LCGuy View Post
    Nick;

    I wasnt implying that it makes it faster, I was stating that the 32 nm size was indicated to have a 50% performance increase over their own 45 nm sized chip - as they have stated on their site, but, Im sure, as you suggested above, that there are changes to the chip design other than the smaller thickness to provide such an improvement.

    I just have to throw this out there because it's bugging me...the sizes given aren't "thicknesses". The quoted "size" is the half-pitch of array elements, ie cells or transistors. In other words, it indicates the linear spacing between transistors/cells, not any kind of thickness.

    I'm sure most people know this, I'm just trying to make sure there's no confusion here.
Page 2 of 2 FirstFirst 12

Posting Permissions