Actions
Engagement
  • Author Author: rscasny
  • Date Created: Date Created
  • Last Updated Last Updated:
  • Views 861 views
  • Likes 1 like
  • Comments 7 comments
Related
Recommended

Is Moore's Law Still Valid?

Moore's Law is about 50 years old and more or less stated that the number of transistors (and processing power) on a CPU would double every two years. It's held its own for most of that time. But it appears that's being tested as more and more transistors are squeezed on to a die.

 

Poll Question: Is Moore's Law Still Valid?

  • in reply to clem57

    Clem, I'm sorry but MSU is really an IBM box thing but MIPS is the deal, which translates from box to box. Also, there are FFT Butterflies which are a great indication of how your floting point processor is performing, you can use the same test for your GPUs as well

    MIPS takes into account, speed of the overall system. ie, the DRAM, OS, hard drives, and CPU(s).

  • in reply to jw0752

    True. In in fact as performance goes, all machines including mainframes do not use MIPS (millions of instructions per sec), but use MSU (millions of service units) Why service units?

     

    A million service units (MSU) is a measurement of the amount of processing work a computer can perform in one hour. The term is most commonly associated with IBM mainframes. It reflects how IBM rates the machine in terms of charging capacity

    So we see it is the work being done to allow different instruction sets that may exist. RISC vs CISC question affects what an instruction can do...

    I suspect the truth should apply to any MCU or Embedded processor if we want to measure Moore's Law.

  • If one considers only number of transistors and volume they occupy we miss the quality performance aspects of higher speeds and lower power requirements. I think Moore's Law would be better stated with a measure of operations per second in a defined volume as the measure.

     

    John

  • I feel We quickly progressed into a realm of Electronics we actually don't understand fully. Hopefully once the understanding stage is over with, We might see the same scale of progression as Moore's Lay suggested. .

     

    Regards,

    Gill

  • in reply to clem57

    You are correct and that is true no doubt.

    Predicting how long the law will hold is a tougher question.

    I suspect there is still likely some ways to go in this case.

    Moore's Law held up long after the first predictions of it running into a physics limit.

    Scientists and others often believe physics precludes a solution when it doesn't.

    It was thought a heavier-than-air vehicle could never fly.

    I think when the law was postulated it was thought that transistors could not have geometries smaller than 450 nm, because that was the wavelength of light used to expose masks.

    It was thought that white LEDs were impossible because a bandgap generates essentially monochromatic light.

    There are numerous other examples.

  • Physics is the absolute law. So Moore Law cannot hold true long term, only for the shorter term.

    Clem

  • It is sort of still true but not on the 2 year schedule. Quantum (non-linear) leaps in the shrinking of transistor size are still happening and there are more coming. However some of these quantum leaps are bigger than a factor of 2 and some take longer than 2 years to develop. When they actually get commercialized, we will be able to see if they are equivalent to a doubling of the number of transistors every 2 years. Although in the past few years commercialized technology is not quite keeping up with Moore's law, there are already various experimental technologies that can produce transistors all the way down to the size of a single atom, so the potential is there to get back on track. It is just a matter of how long it takes to make commercial devices with these new technologies.