Physical size of processors.

Discussion in 'Computer Science & Culture' started by Dynamite Dan, Jun 5, 2011.

Thread Status:
Not open for further replies.
  1. Dynamite Dan Registered Senior Member

    Messages:
    18
    Hello.
    Im in the market to be buying a new processor soon for my aging Pc and i thought about something.

    We all know that processors are getting smaller and faster all the time, Intel and AMD are making the transistors on the processor smaller so they can fit more on.

    What confused me was.. why don't they just make the processor bigger? instead of having a 1cm x 1cm CPU, make it 2cm x 2cm will that make it run 4 times faster?
    I know cooling will be an issue, but im sure a big enough heat sink will manage.
     
  2. Google AdSense Guest Advertisement



    to hide all adverts.
  3. John99 Banned Banned

    Messages:
    22,046
    That the answer, right there. Heatsinks max out, just the law of diminishing returns.

    Threw me a curve ball with multi core, i'll admit that now.

    Just leads me to believe cooling isnt the answer.
     
  4. Google AdSense Guest Advertisement



    to hide all adverts.
  5. Dynamite Dan Registered Senior Member

    Messages:
    18
    so, your saying that heatsinks have limits? over that limit and the ability to take heat away is reduced?
     
  6. Google AdSense Guest Advertisement



    to hide all adverts.
  7. James R Just this guy, you know? Staff Member

    Messages:
    39,426
    The time it takes for a signal to travel from one part of the chip to another depends on how far it has to go. If you double the size of the chip, signals will have to travel, on average, more than twice as far as before, which will take more than double the time. That means the processor will be slower compared to if it was smaller.

    As for cooling, that becomes less of an issue for larger chips, since the heat generated is spread over a larger area and can be dissipated more efficiently.
     
  8. leopold Valued Senior Member

    Messages:
    17,455
    there are two main reasons.
    first is cost.
    second is the statistical failure rates of larger dies.
    let's say current chip sizes fail at 1 out of every 3 off the line.
    if we make the die size 3 times bigger that will mean every chip will be bad.
     
  9. leopold Valued Senior Member

    Messages:
    17,455
    the heat generated from a chip is directly related to circuit density not by size of the chip.
     
  10. Sarkus Hippomonstrosesquippedalo phobe Valued Senior Member

    Messages:
    10,408
    No it doesn't - it means 70% will be bad. Just basic probability, really.

    Please Register or Log in to view the hidden image!



    As for CPU size... speed is one consideration (larger = slower for an otherwise identical chip), but then there are the manufacturing costs (smaller means more chips off the original silicon wafer), power consumption (smaller generally means lower requirements etc).
     
  11. James R Just this guy, you know? Staff Member

    Messages:
    39,426
    Yes. I meant that if you made two identical chips, one with double the circuit density of the other (and therefore 1/4 of the area), then the larger one would be easier to cool.
     
  12. leopold Valued Senior Member

    Messages:
    17,455
    i didn't know how to get my point across.
    yes, increasing the die size will result in a larger number of defective chips.
     
  13. John99 Banned Banned

    Messages:
    22,046
    I cant say what the limitations are for extremely large heatsinks, but they still need to be practical anyway so it is a moot point.

    The heat effects all the components as well, not just the processor. Even if you can increase the heat level a cpu can withstand that processor is dissipating heat to the other components. Larger or more fans just increases the noise and do you want a giant industrial fan blowing?
     
  14. John99 Banned Banned

    Messages:
    22,046
    Let us not forget that heatsinks are really only effective if they themselves can remain cool. And if the end of the hs is cool does not mean the chip will not go above operating threshold. So the larger\further away the hs is from the heat source is where the "diminishing return" comes from.

    Larger cpu surface area would definitely help and have negligible negative effect on speed...if you ask me.
     
    Last edited: Jun 7, 2011
Thread Status:
Not open for further replies.

Share This Page