What is Moore’s Law Anyway?


Everyone knows Moore’s Law, right? Computers will double in performance every 2 years, or 18 months in some versions? And everyone claims it’s been remarkably consistent and some even claim that it’s Moore’s Law the causes computers to double in performance instead of it being an observation and prediction.

Well that’s not quite right. I read this article because the title caught my eye. While much of the article has food for thought, I was annoyed by the opening paragraph that reads 

“In 1965, Intel cofounder Gordon Moore published a remarkably prescient paper which predicted that computing power would double about every two years. For a half century, this process of doubling has proved to be so remarkably consistent that today it is commonly known as Moore’s Law and has driven the digital revolution.”

But that’s not what Moore wrote. According to Wikipedia, and my own memory of reading Moore’s Law, 

“Moore’s law is the observation that the number of transistors in a dense integrated circuit doubles approximately every two years.”

In the initial paper in 1965, he said every year, but revised that 10 years later to every 2 years. The bit about 18 months wasn’t Moore at all. From Wikipedia:

“The period is often quoted as 18 months because of Intel executive David House, who predicted that chip performance would double every 18 months (being a combination of the effect of more transistors and the transistors being faster)”

Moore was remarkably prescient. The plot in the Wikipedia article is so close as to suggest that he knew something before it happened. And certainly transistor density contributes directly to the steady increase in computing performance. But being the picky person that I am, I’d like articles quoting Moore’s Law to quote what he actually wrote.

Bob

6 thoughts on “What is Moore’s Law Anyway?

  1. True, precision is nice.

    Still, you kind of buried the lede in terms of that article! The reach of physical limits for chip paths is at hand, and so we need some kind of quantum leap (maybe literally) w/ our devices if we’re going to keep getting more powerful in our portable gear.

    I’m not sure which is going to hold us back more, chips or batteries – both are very near being constrained by chemical/physical limits and in need of some out of the box thinking re-engineering if we’re going to get the sci fi future we’ve been promised, never mind the jet cars.

  2. While the end may be in sight for Moore’s Law, there are a lot of other ways to accomplish computer performance. Certainly graphics chips and cards are continuing to be more powerful. Parallel computing allows many tasks to be done concurrently. Granted, programs and operating systems need to be written to take advantage of parallelism, but that is happening more and more.

    Of course, as you mentioned, the next big thing is supposed to be quantum computing where a “bit” can have more than 2 states. Talk about fuzzy logic.

  3. Heh – it’s always funny how LiM not updating gets people (well, me) to dig a bit deeper into the existing posts.

    One thing I’ve never understand about Moore’s Law is if it “feels” more like a series of semi-big breakthroughs that happen to even out to a smooth curve, or if there is some other process making it a gradual but logical progression. I assume manufacturers make more or less the best chip they can afford to at any given time (some trickery not withstanding, like Intel disabling math coprocessors on cheaper 386s to create an artificial price point, that sort of thing) – and also a company can probably make more tightly packed circuits but at the cost of more failures they have to throw away.

    It just is so odd when innovation proceeds in a predictable way…

    1. Mostly it’s the technology of making things smaller. And it does go in discrete steps. The article I just posted talks about 10nm chips. Here’s a list of chip sizes from Wikipedia.

      10 µm – 1971, 6 µm – 1974, 3 µm – 1977, 1.5 µm – 1982, 1 µm – 1985, 800 nm – 1989, 600 nm – 1994, 350 nm – 1995, 250 nm – 1997, 180 nm – 1999, 130 nm – 2001, 90 nm – 2004, 65 nm – 2006, 45 nm – 2008, 32 nm – 2010, 22 nm – 2012, 14 nm – 2014, 10 nm – 2017, 7 nm  – ~2018, 5 nm – ~2020

      Yes, it’s discrete jumps, but it is fairly regular. I worked in a semiconductor division in the late 70s. I was a systems analyst and I remember designing a costing application that had to take into account both lost chips as well as recursive processing. Chips are made on wafers and I remember the big deal when they went from 4 inch to 8 inch ( or was it 4 to 6), and retained the yield.

      That 10 micrometre chip in 1971 was 10,000 nanometres. So 1000 2017 chips would fit into one 1971 chip.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

w

Connecting to %s