back to article Intel loves the maker community so much it just axed its Arduino, Curie hardware. Ouch

Intel's flirtation with the maker community appears to have fizzled out, although the chip giant insists its passion remains. After announcing plans to axe its Edison, Galileo, and Joule compute modules later this year, Chipzilla has said it will stop making its Arduino 101 board and its much-hyped Curie module, too. The …

  1. Dwarf
    Pint

    I bet a cold beer that most would just have stuck to the Atmel based boards - given that its about getting people on the ramp of understanding how things work.

    Why add in another variable for beginners?

    Would those starting from page 1 copy the page they are looking at or go - wow, look thats got an Intel badge and a different spec ?

    1. Chronos
      Pint

      @Dwarf

      Well, there's also the ESP8266 in all its guises and the STM32, which runs at 72MHz, is far less fragile, runs on 3V3 and doesn't crap itself after a heavy debug session yet costs around £2 for the F103 "blue pill" modules.

      The ESP has 802.11n built in and massive community support. Both can be integrated into the friendly Arduino IDE. Intel's stuff was barely relevant anywhere.

      Don't get me wrong, the Atmel based boards are great for starting out, they're just rather light on RAM and flash - the STM32 has 20k and 64k (officially, but most blue pills have 128k flash) to the Mega328's 2k and 32k, so much so that you often find yourself running out of dynamic memory on the pro mini and nano, not to mention just two interrupts. Both the SoCs I mention have the ability to attach an interrupt to most of their GPIOs.

      I'll still get a beer in, though.

      1. BillG
        Facepalm

        Re: @Dwarf

        Once again, Intel proves that they are a slave to their own history.:

        Never trust Intel for embedded systems.

        They can't even spell "microcontroller".

      2. Dwarf

        Re: @Dwarf

        @Chronos.

        I completely get your point about other uC's being a lot higher spec than some of the Atmel ones, but you missed the point I raised in my original post.

        Google for "Arduino getting started" and see how many pages you have to go through to find an Intel or Microchip / pic reference. Beginners won't do that as they won't know about the different controllers on the boards and are probably just reading up on this Arduino thing. Will they know or even care that the vanilla Arduino has an Atmel uC ?

        Ask yourself how many beginners would have a starter app that consumes more than say 30 bytes of RAM or even knows what an interrupt is and how it works.

        I'd expect that many of the medium capability users will still not care too much about either, but are probably becoming aware of them and may have used sketches that do use Interrupts.

        Even on my heaviest Arduino project which uses a Mega2560, I only use just over half of the RAM and half the FLASH - due to some careful coding with PROGMEM with EEPROM for things that don't have to be in RAM.

        Its true that more interrupts and GPIO pins would have been nice on that one, but cascaded IO expanders easily handle that, particularly for the low speed stuff and it gave the opportunity to talk to the kids about I2C and SPI and where its used

        1. Chronos

          Re: @Dwarf

          @Dwarf

          I did indeed miss your point and it now makes sense. For educational use, Atmel is the only choice.

      3. wobbly1

        Re: @Dwarf

        Horses for courses . Time critical signalling ATMEL/arduino, net and computationally heavy an ARM SBC, if you need both link ARM and ATMEL systems with a serial connection via GPIO i2c or USB. there was no ecological niche for the Intel offerings.

  2. DCFusor
    FAIL

    They don't get loyalty

    And a few other things, but that's a major show-stopper for any decent engineer. We all look at this before any design-in for something major...

    And Intel, if they don't get major design-ins that generate big volume, drop things like hot potatoes.

    So, any smart engineer wanting to keep his job, won't ever design in an Intel maker type board, ever.

    This isn't the first time they've abandoned everyone on a product.

    Now, you can also look at the fact that they are a poor fit for intended use, draw too much power, and cost too much. For a big design in cost may not matter as much, as your boss will have a custom board made to hold the "Intel or whoever" parts and get them cheap.

    This is why Microchip took such a share (and had enough spare to buy Atmel) - from their earliest series on, they kept some form of easy back compatibility or forward porting toolset available, and in some case you could move to the next model up without even changing the PCB.

    Arduino and Pi are further popular because of the huge community. How many arduino or pi "killers" are themselves dead because it's not just about the hardware? It can be argued that you can do better - and better per buck or watt - for either. But when it comes time to make something actually work, that huge community that has usually already run into whatever problem and solved it is worth a lot more than a buck a part...

    I don't guess Intel will read this... too bad guys, it's not like no one has tried to tell you the truth.

    1. bombastic bob Silver badge
      Devil

      Re: They don't get loyalty

      and the corollary of that, any engineer seeing how Intel is dropping the ball on microcontroller things would realize that the future of their components may be in jeopardy, too, and won't sink any development time into using their stuff [unless it's 100% portable to someone else's chips, like Microchip or ATmel]

    2. Will Godfrey Silver badge
      Thumb Up

      Re: They don't get loyalty

      Hey! Were you looking over my shoulder?

      That's virtually the same as what I was going to post.

    3. Anonymous Coward
      Anonymous Coward

      Re: They don't get loyalty

      I can agree with that. I'm only a begginer hobbiest and getting the raspberry pi working with a tft was hard enough but satisfying.The arguably better orange pi is still in its wrapping.

      1. DropBear

        Re: They don't get loyalty

        To each his own I suppose - I have the Orange Pi on my desk and wouldn't touch an RPi with a bargepole. As for Intel - whipping up a lively community overnight form scratch is absolutely doable (Espressif did it with the ESP8266) but you have to offer a major feature missing from the competition's stuff and you have to sell it _at_ current prices or preferably less. If it costs ten times as much as anything else nobody will touch it. If it's "better" in all sorts of vague ways (like some more RAM, some more CPU cycles, some more integrated peripherals), nobody will bother to learn it all - and most of all, you need to bring something the users really want, not something you think sounds like an improvement. Market polls won't do that for you - you need vision. Espressif had it. Intel didn't.

    4. The Indomitable Gall

      Re: They don't get loyalty

      I personally suspect that the Pi got where it is today because the biggest vendors of Pi hardware and accessories all have "Pi" in their names (well, in the UK at least) -- as well as several of the smaller ones.

      If they started selling the Raspberry alternatives, they'd get shot down for trademark infringement... oh, and probably get pushed down the list when it comes to supplying precious limited Pi stock.

      I've finally got a Pi, because the release of the Zero W means they've finally got a product that is almost (but not quite) what I want -- it's still missing straightforward battery power.

      If C.H.I.P. had a UK distributor, I'd be all over it, but it doesn't, because Pimoroni and The Pi Hut can't touch it.

      1. Anonymous Coward
        Facepalm

        Re: They don't get loyalty

        You don't do battery power on the Pi, for good reasons. If the CPU crashes/halts/you hold it up due to your own buggy code, you don't want that taking out the charge/discharge control of the battery! Plus the different batteries everyone would want. Why force people into 1 type of battery, or worry about loads of SKUs?

        So everyone and their dog sells Pi battery break out boards/addon boards.

        1. The Indomitable Gall

          Re: They don't get loyalty

          " So everyone and their dog sells Pi battery break out boards/addon boards. "

          Which leads to a bewildering array of options in terms of auto-shutdown on low charge, onboard charging vs need for additional charger, compatibility across OSes.

          You're talking to someone who has just said he hasn't bought a Raspberry Pi in the five years they've been available specifically because of this and telling him that his problem isn't actually a problem... well it was for me.

          I've only bought one now because the £9 with onboard wifi del and the particular use I have in mind mean I'm happy to put up with a USB battery pack and frequent battery checks.

          And given that I'll be running a LAMP stack with Python on Raspian, I'm fairly confident I won't be able to block out preemptive multitasking...

  3. John Smith 19 Gold badge
    Unhappy

    Group that came up with it is history

    So is their products.

    Intel don't waste much time airbrushing people (and things) out of the corporate history.

    Something to keep in mind for all developers.

    If it's not very close to the Intel mainstream how long will it last?

    1. HmmmYes

      Re: Group that came up with it is history

      24 months, speaking as someone who's been thru the whole Intel VC buyout, followed by fuckup, followed disengagement.

      Intel are *desperate* to open up new markets. But they are a binch of idiots who dont really seem to grasp how Intel makes it money.

      Intel are very good at fabricating chips and selling them at high margins. Thats.

      Desgning those chips, picking new markets, working with other companies, etc - they are shit at.

      Intel is about 3% world class silicon fbaricators, 10% workers, and 87% seat warming morons.

      1. paulf
        Meh

        Re: Group that came up with it is history

        FTA: "[The New Technology Group] was formed in 2015 through the restructuring of Intel's New Devices Group."

        I bet the PHB MBA who came up with that incredible group rebranding name change got a massive bonus for his/her efforts.

  4. Anonymous Coward
    Anonymous Coward

    Not just hardware either

    They drop software support, often before a device has gone main-stream.

    Look at meta-intel for Yocto Project. Cedartrail support was dropped just as boards were making out in to general availability, which means that any system using it has to take on a third party (which I guess is why they bought WindRiver!) to maintain the kernel and core components. This is feasible for large players, but totally impossible for companies building low-volume solutions.

    Personally, I would like to see support for CPUs / chipsets maintained for at least five years after they go out of production, not just as they start ;-)

    Note: This isn't intended to be a dig at Yocto, Intel or WindRiver - they all do an excellent job and have some very knowledgeable and helpful people who go out of their way to help :-)

  5. imanidiot Silver badge

    well I'm surprised...

    </sarcasm>

  6. EveryTime

    You can't afford a fickel company for long-lived products

    I've put up with lots of software pain on various products because I knew that they would be around for the long term. When there is a long learning curve and a slow move from development to production, you can't risk going with a company that has a reputation for abandoning markets.

    The implication for companies is that they can't make the simple calculation "do we have enough orders to make this product line wildly profitable by next year?" They need to do the harder estimation of "if we kill this product, will people trust us the next time we try to enter a market?"

    When your company culture is such that a manager thinks "screw the rest of the company, this will hurt my profit numbers next quarter" and gets away with it, the decline has begun.

    1. HmmmYes

      Re: You can't afford a fickel company for long-lived products

      This.

      Unless you buying billions of silicon a year then dont get into bed with Intel.

      Even if you are be wary as Intel churn out a lot of silicon that dont work.

    2. paulf
      Alert

      Re: You can't afford a fickel company for long-lived products

      @EveryTime, "When your company culture is such that a manager thinks "screw the rest of the company, this will hurt my profit numbers next quarter" and gets away with it, the decline has begun."

      If you're starting development of new silicon (i.e. not a simple derivative of an existing product) it could take 3-5 years for that to move from initial idea to volume production and [hopefully] profit once it's paid off the R&D investment. If some PHB is worried what his numbers will be in 13 WEEKS time, rather than a few years out, then the decline is so deeply rooted the company is beyond help as those bean counting bonus weasels have infested the whole hierarchy. Exits are here and here - please remove and leave behind any well heeled shoes before jumping down the emergency slides.

  7. Anonymous Coward
    Anonymous Coward

    At least Google puts a 'beta' label on stuff it is likely to kill

    Intel keeps looking for another PC sized market, but they will never find it. They missed their one chance of this century so far by ignoring mobile until it was far too late.

    1. tcmonkey

      Re: At least Google puts a 'beta' label on stuff it is likely to kill

      Yep - I still reckon that getting rid of XScale when they did was one of the bigger mistakes Intel have ever made.

  8. Colin Tree

    cheap arse DIYs

    They just realised DIYs only want to spend $10 for an SBC and they use free Linux or Android, no money to be made here, the money's in high end gear.

    1. Anonymous Coward
      Anonymous Coward

      Re: cheap arse DIYs

      Tell that to ARM - most of their money comes from selling lots and lots of very cheap devices.

      1. Peter Gathercole Silver badge

        Re: cheap arse DIYs @AC

        ARM (Softbank) collect license fees. They rely on other people to actually make and sell the stuff.

        1. Peter Gathercole Silver badge

          Re: cheap arse DIYs @Me

          Sorry, let me refine that.

          ARM (who are now owned by Softbank) create ISAs and core designs, and then licenses these designs to other companies, who actually make and sell the stuff.

          The point I was trying to make is that ARM is not a manufacturing company.

      2. inmypjs Silver badge

        Re: cheap arse DIYs

        ARM don't sell any devices and cut-throat competition means making money isn't easy for their many licensees.

        DIYs (makers) are cheap and don't provide volume anyway. Microchip make more microprocessors in a day than all the arduinos ever sold.

        1. Mikel

          Re: cheap arse DIYs

          >ARM don't sell any devices and cut-throat competition means making money isn't easy for their many licensees.

          ARM commoditized their technology, which makes it more widely available cheaper and in greater variety.

          Water is a commodity. I can buy 47 different kinds of it at the corner gas station - mostly paying more for water than the gasoline from the pump outside. This despite the fact that at home I buy it by the acre-foot. Neither the water vendors nor the retailer are in danger of going out of business.

          Commoditization is not death.

          1. inmypjs Silver badge

            Re: cheap arse DIYs

            "Commoditization is not death."

            Sort of was for Atmel who were bought out by Microchip, you know the Microchip that chose MIPS for their 32 bit processor range because they understood it is hard to make money being just another general purpose ARM core processor vendor.

            Commodotization makes ARM processors cheap and water cost more than gasoline? Dunno WTF you are trying to argue with there.

    2. Richard 12 Silver badge

      Re: cheap arse DIYs

      The Raspberry Pi is now the compute module inside a lot of industrial kit.

      Because it's cheap, can be manufactured in-house if desired and has excellent software support.

      It's the best advert Broadcom ever had - for a mediocre ARM and a mid-range custom GPU. Low margin silicon, huge volume.

      Intel clearly saw what happened and wanted in, but they did not understand why it worked.

      ARM now owns the low-power computing space, and has no competitors left.

      1. Anonymous Coward
        Anonymous Coward

        ARM now owns the low-power computing space, and has no competitors left.

        And the really good bit about that is it's not a monopoly. ARM "only" own the IP in the core and there are loads of companies making the physical devices - which means there are plenty of alternatives to use with a wide range of on-board peripheral options.

      2. inmypjs Silver badge

        Re: cheap arse DIYs

        "Intel clearly saw what happened and wanted in,"

        12 million sold is still peanuts beaten by one design-in for a set top box or router for a major ISP/Cable supplier.

        "ARM now owns the low-power computing space, and has no competitors left."

        I thought MIPS parts were still being made in the millions and isn't it Intel silicon in the Virgin superhub people are currently bitching about (with 5 million Virgin cable customers just in the UK).

    3. The Indomitable Gall

      Re: cheap arse DIYs

      The scary thing about SBCs (from Intel's perspective) is that their ubiquity is making them a replacement for microcontrollers in hobbyist toys, and the hobbyist toys tend to spawn gizmos on Kickstarter, some of which get big. Intel put their money into true microcontroller (Arduino) rather the SBCs, which makes a fair bit of sense, because they certainly weren't going to challenge the Pi -- Arduino is the only other significant single tech used for Kickstarter gizmo prototypes.

      Intel's place in the microcontroller/embedded space was historically based more on the availability of experience x86 assembly coders than any technical advantage. But assembly is mostly a thing of the past in desktop coding, so there's no migration path from Windows to embedded -- the entry to microcontroller coding is Arduino, and if there's no x86 on Arduino, Intel is going to continue to lose market share in embedded.

  9. I3N
    Pint

    After that point, only 'collected' Intel boards

    July 2014

    "So sorry!

    Hi Everyone,

    Due to logistical issues, the Roadshows have been canceled. Thank you so much for registering!

    If you want to check out more about Intel and IoT go here. Don't worry, we're not in the habit of canceling events, so make sure you visit us to get all the deets on upcoming goodness!

    Cheers,

    The BeMyApp Team"

  10. David Pearce

    Intel used to be No2 in the embedded sector with the 8051, only the Motorola 6800 was more used.

    Intel the lost interest because the x86 market was so profitable. Microchip and Atmel muscled in by offering low cost development.

    Now the x86 market is shrinking panic has set in. Right now I am using a year old I5 3GHz, the latest and greatest I5 and I7s are only marginally faster, a little lower power envelope

  11. David Pearce

    The previous post should read >4 year old I5.

    The only real speed gain would be from a SSD

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon

Other stories you might like