Advertisement

Blog

Moore’s Law Demise: Maybe It’s a Good Thing!

Moore’s Law will celebrate its 50th anniversary on April 19, 2015. Many say that the statement or theory has run its course due to the next moves toward 450 mm size wafers, extreme ultraviolet lithography (EUV) and 20 nm CMOS. There are many technical problems as well as cost problems in the present development cycle.

Well, that’s what we do as engineers and designers of electronics—we solve these types of problems. And most likely, clever engineering talent will come up with ways to help solve these problems. But maybe the efforts we put into making Moore’s Law happen from year to year might better re-directed with solving other problems and challenges in electronics. How about 5G?

Back in the early 1960s, when President John Kennedy told the world that we would land a man on the moon in this decade, not only did that ignite the spark in me to become an EE, but it set the entire country towards a collaborative effort to reach that goal. When government and industry get behind an effort in such a powerful way as it did then—-technological things will happen which before were only a dream in mankind’s minds.

5G is the new massive effort, in my opinion, that can spur government and industry (Not just in the USA like in the 60s but worldwide) to ante up funds and resources that will propel us towards the goal of reaching “everything everywhere and always connected”—-with 12 year battery life by the year 2020! That’s a key milestone year because of the Olympics in Japan that year which gives industry and engineering development an added reason to reach 5G goals in the next 5 years.

There are many technological needs in order for 5G to happen. If we focus our efforts away from 20 nm CMOS and Moore’s Law and towards next generation technology that can greatly increase speed and lower power in Data converter technology and RF technology (Analog does not need 20 nm CMOS and so what if we don’t have that technology for the digital components either!), low latency and huge number of antennas in Massive MIMO, GaN Technology to improve poor efficiency of Power Amplifiers (PA) in the base station transmit chain and frequencies up to 100 GHz—then maybe our collective engineering efforts will bring about new process technologies and architectures that do not involve going to 20 nm or the next node in semiconductor technology flowing Moore’s Law.

Ideas like maybe using single bit ADC Sigma Delta architectures in the receive section of the base station and even device handsets, tablets and PCs are being considered. This is an old concept, but with a fresh look at it in 2015 and beyond, I am sure we will come up with clever solutions that defy paradigms or advancements in Software Defined Radio (SDR) as seen in Figure 1 below. Test bed challenges are already being addressed by National Instruments since testing capabilities is the area that usually leads advances in technology, or at least walks hand-in-hand with new technology developments and evolves as do the technology prototypes and breadboards.

Figure 1

USRP SDR: Universal Software Radio Peripheral (USRP) Reconfigurable I/O (RIO) (a) Hardware and (b) System Block Diagram (Image courtesy of National Instruments)

USRP SDR: Universal Software Radio Peripheral (USRP) Reconfigurable I/O (RIO) (a) Hardware and (b) System Block Diagram (Image courtesy of National Instruments)

If you think that reaching such high speeds and low latencies are out of the question then just look back to the early 1990s when our landline telephone system carried voice signals only and we needed to find a way to transmit higher speed digital data over the existing 4 kHz copper line infrastructure. Engineers will find a clever technical way just as they did back then so that the Internet could thrive and grow.

So don’t despair over the loss of Moore’s Law whose time has pretty much come to an end (Or maybe engineering will find a way to still get the 20 nm problems solved!) because we have plenty of good technological advances to come because of this thing called 5G. Watch for my feature article on EDN coming soon that will discuss the analog semiconductor design challenges like speed and power and the clever ideas of using beam-forming in antennas to help make 5G happen in the next five years.

6 comments on “Moore’s Law Demise: Maybe It’s a Good Thing!

  1. eafpres
    April 12, 2015

    Hi Steve–way back in 1989, I was trying to get funding at NIST (then, the National Bureau of Standards) for the FEL project–Free Electron Laser.  Why? At that time a few companies, including IBM, were predicting hitting a wall with traditional lithography and UV lithography.  The next logical step appeared to be X-ray lithograpy, and the FEL was a novel idea to generate X-rays without a radioactive source like Cesium.  It wasn't funded; in fact killed by internal competition, and here we are 26 years later and EUV has been pushed way beyond what we thought possible.  Not only that, but step and repeat resolution has been far superior to anything we envisioned.  

    Nonetheless, these days I think that photonics is likely to replace a lot of digtal electronics.  We aren't there yet, but given the achievements already made, and the promise that photonics will dramatically reduce the energy cost per bit, gives me enthusiasm that in the not-too distant future Moore's Law may not be relevant any more.

    One other point; the Moore's Law question may not be the right question.  Take a look at this from EE Times:

     

    http://www.eetimes.com/electronics-news/4408525/Moore-s-Law-trumped-by-Wright-s-Law

  2. Steve Taranovich
    April 12, 2015

    @eafpres1—Wright's Law, on the EETimes article link you provided, really fits the coming generations of the IoT and 5G—excellent observation!

    Your experiences at the NBS are what we traditionally say when things like “going to the moon in this decade” are mentioned. There is no way this is possible most people thought back then just as we think we can't get to a 5G scenario in five years. History repeats itself

  3. madhujy
    April 13, 2015

    great information thank you

  4. praneethgoud
    April 15, 2015

    nice one

  5. D Feucht
    April 15, 2015

    Steve,

    We're synchronized in our thinking regarding Moore's law. It is quickly reaching the point of diminishing returns. In a landfill-driven society, long-term reliability is considered unimportant in design and the decreasing MTTF of shrinking IC geometries will relegate such ICs to the dump all that quicker. Barrie Gilbert pointed out a few years ago that Intel is down to what internally is called “the three-year process”. I would prefer a 1 GHz computer that will last 25 years.

    Moore's law is not the end of misdirection in technology. You mentioned another: “everything everywhere and always connected”. Not enough engineers are watching those sci-fi movies where the highly interconnected, advanced technology fails due to some systemic instability. That part is not fiction. Y2K was but a warning.

  6. Steve Taranovich
    April 15, 2015

    Thanks for your comments Dennis—I hear you loud and clear–unforseen systemic instabilities could be catastrophic—as would inadequate security.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.