The First 5nm Chip

For almost forty years, integrated circuits have become smaller and smaller. These chips started out with massive transistors in the early 1970s. They shrank to less than 1μm by 1990, and shrank yet again to less than 100nm by the turn of the last century. Now, Imec and Cadence are experimenting with 5nm technology – the smallest technology available for any mass-produced integrated circuit.

The history of microelectronic fabrication over the last decade is a story of failure. Something happened in 2005, and although chips could be designed at ever-smaller technologies, the transition to these smaller manufacturing processes didn’t go as smoothly as in the 70s, 80s, and 90s. Just a few years ago, Intel said 10nm chips would ship by 2015. These chips are nowhere to be found, and even 14nm technology is still catching up to the yields found in 22nm technology. In 2009, Nvidia said their flagship graphics card would be built with a 11nm process. The current Nvidia flagship desktop graphics card is built with 28nm technology. Moore’s law isn’t 18 months anymore.

While Imec and Cadence have completed the tapeout on a 5nm device, it’s just a test chip. Before starting manufacturing on a single process node, Intel and others will tapeout a simple test chip to verify their latest process. This 5nm tapeout will not become a manufactured chip, but it does mean we’ll see more talk about the 5nm process in the future.

68 thoughts on “The First 5nm Chip

  1. flameshield+=1;
    Brian, based on the articles that you’re publishing, it looks like you’re getting more and more interested in general technology news. Even though that’s interesting, it’s got nothing to do with hacks. Hackaday is supposed to be for hackers. If you really want to do something else, then you should do it on another website. Now you’re just trying to fit a round peg in a square hole.

    I would be more than happy just to see a daily pick of interesting hack projects on hackaday.io. Surely there are enough of those, and posting an article about them saves me from rooting through mediocre ones.

    1. Pretty sure a greasemonkey script that exists where you can block his posts from showing..

      Or, just don’t waste the brainpower reading it in the first place. Pretty simple concept.

    2. Yes same here I found the article interesting even though it was low on content but really it’s not Hackaday more like EETimes where at least one Former Hackaday hack has moved too!

    3. That’s pretty fun how the relation to the reader is slowly changing here.
      Your comments are less and less friendly. But here he was not flaming or trolling.
      Yes those post make Hack A Day less Hack A Day and more EEtimes Each Day.
      And you should understand that people taking time to write that are maybe your
      most devoted readers…
      People who want to see only hack (and original stuff) here, come here because they
      are searching for something you don’t see anywhere else.

      Sad to say but it seems the buy of Hack A Day is not going as well as expected:
      “A directive came down, and we need ‘news’. That means general technology news. You think I like writing this stuff?”

      1. This. I religiously read Hack a Day multiple times per day. I didn’t know about the thing regarding the posts that Brian said though. If I understand correctly each post is grades as (number of views + 200*comments)?
        I find that grading a bit strange, but maybe that’s a directive too, so not open for discussion.
        I’ll do my $0.05 and will post a comment from now on every now and then, for good posts.
        Perhaps a “like +/-” voting system per post would work instead/in addition), for people like me who usually don’t write comments.

        I appreciate that Hack a Day needs money to run, but when you’re adding news that is not hacking related, at least for me it makes the website less interesting.

      2. One fundamental from the “theory of business” is product distinction: what is it that makes your product different from everyone else’s product.

        If your product is no different from anyone else’s, then there’s no incentive for people to buy your product over theirs, and you’ll have to compete over price.

        That’s a hard way to do business, but a lot of people do it. For example, it’s impossible to make money selling regular printer paper because there’s so many people already doing it. If you want to sell paper, you need to be distinct from the existing players in ways that people want: specialty paper, for instance, or recycled paper.

        The overlords see other tech sites like EE-times with a lot of readers and a lot of ad revenue, so it seems like the way to get more revenue is to be like them. It seems obvious to the overlords.

        I notice that HAD has spent a lot of time, effort, and money in building the Hackaday.io infrastructure, which gets them lots of fodder for posts, but has never addressed the business issues with the blog. There are easy and obvious ways to bring more readers to the site, but they just don’t bother with them.

        They’ve never tried to make a better experience or expand their product distinction.

        Expect a slow and painful decline, or perhaps being put up for sale by Supply Frame.

        1. You are right about distinction. However, Exclusivity is a distinguishing feature of product differentiation. Consider Vuitton handbags, which have limited runs and sell for extremely high prices.

          When HaD did their own thing, they did have exclusive content, which appealed to their advanced, albeit finite viewership.

          Unfortunately, this meant that HaD was not living up to its profit potential. Given the amount of time and effort expended to write one of the older articles, I can see how the staff could feel short-changed; because, they lost all those hours, which could have been used to develop their own projects.

          For this reason, I appreciate the inclusion of these newer, more general articles; because, the staff does deserve remunerations’ for their work. So what if this requires some occasional “dumbing-down”, deal with it. At least you still get the exclusive content.

          After all, the change to the business model could have been much worse.

          /two-cents

      3. Hey, be nice. One day you’ll get old and retarded too. You won’t be able to stay on the hamster wheel of tech forever… Besides, [Brian] forgot Voyage of the Mimi had Ben Affleck as a main char.

        No, I meant the paycheck comes from.

        http://hackaday.com/2015/10/06/password-haters-can-now-use-parts-io/

        yeah, we have to deal with shilling b.s. retardation..

        This aint eevblog youtube/google bux someone has to sponsor the HaD domain.

        1. Re-purpose, Reuse, Renew WAS IN THE PAST HaD.

          Let us NOT compare Mondo 2000 to Wired OR TIME mag to shit tabloid rags. Or old HaD to new HaD.

          Take the gems when and where you can.

    4. Thank you for the suggestion, but no thank you. I looked at your link and it seems that Grease monkey filters out keywords. That suggests that in order to filter this news article, I should have filtered for any of the tags ” 11nm, 14nm, 5nm, chip, foundry, or process”. How am I to know that you’re going to write an article about it beforehand?
      The article is posted in “misc hacks, news”, it would be more user friendly if you removed “misc hacks”, and instead added the keyword “general tech”.

      1. Brian can be abusive and sarcastic when people try to give feedback.

        He wrote in a previous post thread that he sometimes posts misleading or ambiguous information (for example, about contest rules or other business functions) just to provoke someone into asking for clarification so that he can make fun of their response.

        Don’t feel bad, it’s nothing to do with you, it’s just his way.

    5. And yet there is almost NO increase in CPU power ! In the 1980s they had CPU like the Z80 running at 1MHz or the 6809 at about 1MHz they finally got up to 3 GHz for the Pentium 4 in 2002, in 2015 they are STILL at 3GHz !!! 3.6GHz for the I3770 or the I4770 ! Why have we been STUCK at 3GHz for 13 years ? We should be up to about 100GHz by now !

      Hard drives too ! The first 2TB came out in 2009, they STILL sell TB in 2015, 6 years later ! WHY ?!?! We should be up to about 32TB to 64TB by now !!!

      RAM size has the same problem but we have 5nm chips but no more RAM density ? What is going on and who is stopping the increase of the technology ?

      Moores Law says CPU speed should about DOUBLE in 24 months, same for storage of RAM and hard drive space so what happened ?
      CPU speed in 2002 3.09GHz for Pentium 4
      should 6 GHz in 2004, 12GHz in 2006, 24GHz in 2008, 48GHz in 2010, 96GHz in 2012, 192GHz in 2014 and 384GHz in 2016 ! But we are STUCK at 3.6GHz ??? Yea right.

      Hard drive 2TB in 2009, 4TB in 2011, 8TB in 2013, 16TB in 2015, does anyone have a 16TB hard drive ?!?!?

      RAM limits, 2007 32GB, should be 64GB in 2009, 128GB in 2011, 256 in 2013, 512GB in 2015. Does ANYONE anywhere have a 512GB machine ?

      Why are we still using I7 series from 2007 that hasn’t improved much since then, 8 years ago !?!?!!? And more cores doesn’t equal more speed unless the program is written for it. What is going on and who is stopping the improvement of the computers ???

      1. And Intel announced a 1THz chip, yes 1000GHz would be ready by 2005 !!!! Ten years ago they could do 1000GHz so why are we stuck at 3 GHz !?!?!?!? Think it is impossible ?

        Here is DARPA with its 1THz chip too !!!
        http://www.gizmodo.com.au/2014/10/darpas-chipset-runs-an-astonishing-one-trillion-cycles-per-second/

        Here Intel says they will have 10GHz by 2011
        http://www.geek.com/chips/intel-predicts-10ghz-chips-by-2011-564808/

        So in 2001 they say 1000GHz and 10 years later they might be able to do 10GHz, 10% of that speed !?!?!?!
        So what happened to the 1THz chip in 2005 ?

        Someone complained in 2002 or 2003 that computers were TOO FAST and they couldn’t read everyone’s email, guess who, three letters starting with N…… Maybe they had something to do with computer power stopping in 2002 !

        Either way computer speed shouldn’t be the same as 13 years ago !!!! And we shouldn’t use the same CPU since 2007. Yes it has SLIGHTLY improved speed but the speed should GREATLY IMPROVE !

        Here is Intel with its new 1 THz transistors for its 1THz chip in 2005, but we can’t even do 1% (10GHz) 14 years later !!! Why !?!?!
        2001 techtv clip – intels new “terahertz transistor”

        1. …right. No improvement between a modern x86 at 3GHz and a P4 at 3GHz. Totally. Yep. Clock speed is all that matters.

          Dude. Transistors on CPUs cost energy each time they switch states. If you change states at 1THz but you spend most of your time waiting for RAM to catch up with your face-melting CPU doing nothing but check to see if the RAM has filled the cache yet, you do nothing but waste power.

      2. Clock speed != CPU power

        Why? Modern computer architectures get a lot more done in one clock cycle than a Z80 or 6809.

        This is why we have benchmarks for computers (instead of simply comparing clock speeds). Benchmarks measure how much “actual work” done per second.

        Physics is a bitch. Electrons propagate through most semiconductors at a pretty constant rate. Increasing clock speeds means your pipeline gets super deep. You get less and less done in each clock cycle. At a certain point, you’re just making higher clock speeds for marketing purposes (Reference Intel’s old and forsaken “Netburst” architecture:
        https://en.wikipedia.org/wiki/NetBurst_%28microarchitecture%29#Hyper_Pipelined_Technology)

        Super high clocks also use a ton of power. And because the pipeline is necessarily deeper, branch misprediction penalty is much higher.

  2. Thank you for a new (for me) word.

    https://en.wikipedia.org/wiki/Tape-out
    “In electronics design, tape-out or tapeout is the final result of the design cycle for integrated circuits or printed circuit boards, the point at which the artwork for the photomask of a circuit is sent for manufacture.[1] Some sources have indicated that the roots of the term can be traced back to the time when paper tape and later magnetic tape reels were loaded with the final electronic files used to create the photomask at the factory.”

    1. Another use of a coated plastic strip media term hanging on long past when it’s been all but eliminated from use. Can the computer industry please stop calling things “tape” when none is involved in the process?

      Same for video. Who uses tape anymore? How about film? It’s rapidly declining in use, especially at the production end. Most projection film copies for theaters start as digital. More and more theaters are switching to digital projectors, yet movies are still “filmed”.

      Use the right words for the technology actually being used.

  3. I don’t think that having news in and of itself is a problem. A lot of the ‘news’ articles I’ve seen so far have been quite informative and dive deep into the engineering side of things, and not as much the sensationalist scratch the surface drivel.

    This article is light on the engineering, but it gives a history and context for what it means at a level that can be appreciated. There are many hackers that may not know the history, especially now that there’s a lot of young folk involved.

    Basically what I’m rambling on about is that bring on the news, so long as its written at a journal level and is full of tech information that I can take away from it and act upon it. There’s not enough of that going around.

      1. So what’s with folks posting comments that are just “+1”? It didn’t used to be a thing. Like, I don’t remember it happening before this week. I’ve been hitting “Report” on all of them but none got removed despite other comments on this article getting moderated, so I guess they’re not seen as worthless spam??

  4. Maybe Brian just answers to his overlords who sign his paycheck? Maybe those overlords have assigned certain article types to different people. I see certain people always doing things down the same lines as their last article, maybe it’s no difference with him?

    Instead of attacking him (which I know I have in the past for other reasons) we should all bitch to whomever owns the site and tell them the content we would like and that general sciencey type stuff is not hacks and this is HACK-a-day. Not “famous science” or whatever.

    Then maybe it writes about the hacks, or else it gets the hose again.

    1. “I see certain people always doing things down the same lines ”

      I can be naive at times, but I think many writers stick to their areas of expertise (maybe instead of “expertise”, familiarity would be a more accurate term).

  5. I’m probably in the minority here, but it seems like over the last several years, hackaday started off pretty good, got crappy, and has improved a lot lately. The general interest pieces are actually pretty OK, and the hacks seem to be better quality – blinking an LED just doesn’t get on here nearly as much as it used to. Hat tip to the editors – we’re a tough crowd to please on a good day. :-)

    1. As a general rule, atoms start at half an angstrom and go up to 2.5 angstroms. Silicon comes in at about one angstrom, or 0.1 nm.

      A 5 nm wire would be about 40-ish atoms wide, depending on what it was made of.

  6. If I wanted to read media releases I can get time magazine or something like that. Please stick to hacks. Or if you want a section on technology news, create a separate section.

  7. Isn’t any time you try something that can’t be done a hack. 5nm will be a hack until they master it then it will be old hat. One thing I don’t want to see on this site is regurgitated press releases of a company’s bullshit.

    1. Clifford Stoll in his TED talk quipped something to the effect that “The first time you do something is science, the second time is engineering, the third time makes you a technician”

  8. Sheesh, I see the complainers are out in full force today.

    How about this: If you don’t like a given article on HAD, or any other site for that matter, simply don’t read it. There’s no need to gripe and complain that the article doesn’t fit the paradigm that you have imagined for someone else’s site.

    Meanwhile, the rest of us will simply enjoy all of the articles here, because learning new things is fun.

    1. Spoken like a true sheep. Listen, there’s nothing wrong with the articles on here and I won’t complain about tech articles when they are generally appropriate. The point everyone is making is a noticable shift away from articles about the garage engineer, hacking together things that are inventive or normally economically prohibitive for the general public. That’s what HaD is famous for. It’s like going to KFC and they start offering cheeseburgers instead of chicken because that’s the highest selling item in the fast food industry. People go to KFC to get some damn chicken. They go elsewhere when they don’t. But because funding a business on a one trick pony is short lived, you can’t expect everyone in town to clamor over your chicken when they have had it every day for years. They want a cheeseburger. You see it cut into your pockets. So you start acting competitively. You don’t sell the damn burger. You find a way to make the chicken popular again. That’s what HaD is trying to do. I’d suggest they start posting more articles with challenges. Build something using only this or that. Offer prizes. Instructables has been doing this for years and it works. The trip to space thing is cool but it’s too big of a challenge. Too many people will shy away because of the nature of the competition. It’s too steep. You won’t get the small guys working on stuff. You are going after the bigger groups, people with extensive backgrounds. These are the people who take the money.. HaD needs smaller weekly competitions, using categories related to hacking things together using guidelines. Challenge them with requirements, like the device must not use more than so much power and it must do blah blah. I know it sounds like kiddie hour but how do you bring fresh crowds in when you have old tech geezers bitching and moaning about the good old days.

      1. You make good points, some of which I agree with, but calling me a sheep because I happen to enjoy the content of this site is a bit harsh. My point is that almost every single response to this article was a complaint that it was a news item rather than a “hack” , which seems a little ridiculous.

        1. Fair enough, I won’t call you a sheep. Bahhing like a sheep won’t get you in the barn. The argument I’m making is that I have seen too many people use the “if you don’t like it don’t read it” argument. But just a week ago (or so?) they published an article about click baiting. The author even mentioned that it worked enormously to get views, and to expect the technique to be applied to future articles. This is what it has come to, readers are being tricked into reading articles they probably would have skipped over in order to help HaD generate revinue. Once I click into it thinking it will be a great read, I find out I was scammed. It’s starting to burn some of the long time dedicated fans of this site.

      2. “HaD needs smaller weekly competitions…”

        Yes, tiered competitions.

        Although I have been a HaD viewer for some time, I am only now getting into design/prototyping… So, it would be handy if, for example, the Square Inch Contest would have had multiple, experience based tiers… Because I have a novel product idea simple enough to be bulk produced and sold as an Arduino module by the likes of AF or SF. But, due to its simplicity, I wouldn’t dare submit it to the current competition fear of being laughed off of the Internet.

        1. That’s what I mean, this is a great example. I have ideas of my own, but the caliber of some readers on here is light years ahead of my abilities. The HaD grand prize winner from last year was doing something I couldn’t even dream of, with a team of highly skilled individuals. They had a sophisticated platform already in place, almost in the range of commercial level stuff you’d find large companies buying into as a solution. I’m not expecting to enter into contests to win a $5000 scope. But hey a t-shirt, a cheap dev board, or a gift card would be worth it. People aren’t going to submit satellite communications infrastructure systems for a t-shirt.

  9. I read plenty of Hackaday articles and enjoy some of the non-hack one like this. However for the sake of everyone’s time I suggest two rules: 1 the team of writers put [nohack] in the url and title of every text that is not a hack. 2 commenter’s who wish to express their belief that the article they’re commenting on is not worthy of being called a hack must put the tag [nohackcomplaint] first in their comment. This way everyone easily script their browser/RSS-reader/whatever to filter out 1 and/or 2 content going forward.

  10. The parallel to Moore’s law is that every new feature size node requires exponentially more money to implement, which requires more chips to be manufactured and sold to make it economically feasible.

    That means the industry will first merge into one big monopoly – because only by cornering the entire market can they sell enough – and then when there’s no more room to expand on the market, they either start expanding the market or taking exponentially longer to make their return of interest.

    Makes you wonder why the industry is hyping Internet-of-Things so much.

    1. I agree with you, I’d rather not have to discuss the content suitability of a website I frequent. I find myself questioning this daily though. I think a lot of readers are starting to see that as well. I wouldn’t want HaD to tank simply because I wasn’t happy with the random tech articles they submit, but I feel a radio shack-esque revolution in the works. I remember when you walked into that store and it was mostly electronics components and a smart staff. As the years went on I’d walk in, baited by the expectation that my friend will still be there on the corner block, only to find that its been over run by tech garbage I don’t need and a small selection of actual hobby electronics stuffed into the drafty corners of the store. I do it every time, to be let down. It’s like looking for love in a loveless relationship. I want it so badly that I’ll give it without getting it.

  11. “Wass da big deals on deez chipos brah?”

    Physics modeling. Deck out a tiny chip run the sim “Oeh kays”. “So you ran it against Gaussian, Hall, and Lorentz forces mapped out?”,”*Derp* Allwinner is great companies. WE WIN ALL.We worked with Foxxconn. PRODUCT NUMB ONE”

    Great, just f’in great.

Leave a Reply

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