The Intel-powered Arduino

Dev boards based on microcontrollers and ARM System on Chips are everywhere, but finding a small pocketable computer based on an Intel processor has been difficult to find. [Massimo] of Arduino just unveiled a new Intel architecture Arduino-compatible board at the Rome Maker Faire. It’s called the Galileo, and it has everything you’d expect from a juiced-up Arduino running x86.

The main chip is an Intel Quark SoC running at 400MHz with 256 MB of DRAM. On board is a Mini-PCIe slot, 100Mb Ethernet port, Micro SD slot, RS-232, and USB host and client ports. Here’s the datasheet for the Galileo with all the applicable information.

The Galileo can be programmed with the standard Arduino IDE, but from the getting started guide, it looks like this board is running Yocto, a stripped down Linux for embedded environments.

Realistically, what we have here is a board with about the same processing power as a Raspberry Pi, but with Arduino compatibility, and a Mini PCIe port for some really fun stuff. It will be interesting to see what can be made with this board, but if you have any ideas on what to do with a Galileo before it’s released in two months, drop a note in the comments.

149 thoughts on “The Intel-powered Arduino

  1. CPU looks awesome, and running on top of Linux has some nice advantages. The FAQ mentions 2ms for GPIO speed. The code for digitalWrite talks to a /dev/uio device file, so it’s hard to see what that really does, at least without digging much deeper.

    1. “The GPIO output pins on Intel® Galileo are provided by an I2C Port Expander that is running at standard mode (100 kHz). Each I2C request to update a GPIO requires approximately 2ms. In addition to software overhead, this restricts the frequency achievable on the GPIO outputs to approximately 230 Hz.”

      Yikes, 230 Hz maximum switching rate! An attiny will blow its doors off.

        1. so arduino have sold out to intel then? Any mention of price for this thing? and 230hz, yeah, that’s realllllly going to be worth the effort.

          They’re realistically not going to be running windows on this thing, so why bother ‘arduinoing’ it at all, why not just provide a robust gpiolib implementation like everyone else?

          1. What? How can they be ‘selling out to Intel’ at the same time as they are colaborating with TI to make the Arduino TRE? Unless this article is your sole source of news that makes no sense at all. Actually, it still makes no sense, not unless they are abandoning their whole Atmel microcontroler line to solely be an Intel SoC product from now on.

            As for running windows, are you saying it’s not worth adding something to the Arduino line unless it runs Windows? What does Windows have to do with Arduino? Which version of Windows have you ever ran on any existing Arduino? Why would someone want to embed a crappy low quality desktop OS into an embeded device anyway?

            If you don’t want a Galileo that’s ok, don’t buy one. It’s just another option, relax!

          2. While the GPIO expander is a little slow for real time control I would design my faster signals to use the INT and the 2 UART signal paths. When I was an applications engineer at Microchip I saw and helped design very creative uses for the UART: frequency detection/measurement, pulse counter, data modulator, PWM, video signal generator, frequency generator, and led blink rate controller. Its too bad that the one GPIO direct from the CPU, which may be a lot faster, is connected to the LED instead of the expander port.

          3. They see this as a way to sell chips. They have fabs and need to sell things. They also see the Internet-of-Things as a developing market and want to be in it. They let ARM get into markets and they want to avoid such mistakes in the future.

            AFASIK this entry will still use Arduino IDE, regardless of what is under the hood.

            I don’t look as it being bad anymore than ARM based Arduinos are bad.

          4. Arduino didn’t sell out…. its open source hardware and software meaning anybody can copy,use of modify the design to there liking or needs. and mass produce hardware with there name on it. Intel just simply jumped on the band wagon with this board..

    1. “Not a Pi killer in my opinion.”

      I would phrase it as “Not a Pi killer for your purposes”. Yes it lacks HDMI (though you have PCIe so theoretically video could be added) but it has more IO pins, a lot more PWM pins, and has other advantages. For a lot of people this is better than a Pi, for others not as good. You can’t make a general statement.

        1. It’s really not a general option, the power requirements are not in the same ball park. If your application needs any kind of I/O switching speed it’s not an option either. Price wise it’s probably not in the same ball park either. As an Intel product it’s going to have an Intel profit margin.

      1. I don’t mean it can’t be better at certain things. Most people don’t use their Pi to it’s full potential anyway, they use it to quickly whip up a video player for instance. I don’t think the same amount of people that have a Pi, will get this thing.
        Again, I’m not saying any of the two is better / worse, its more like devices for two audiences.

    2. The micro USB plug is a FEATURE, not a BUG, because you get BOTH host and device capabilities from it. With the blatant Arduino references I don’t think they are aiming at the pi.

        1. In 1992 you wouldn’t have 256 Mb of RAM nor a 400MHz clocked Pentium CPU in a desktop, so not in a laptop either. If I remember well, in 1994 I had a no-brand PC with a 486DX @ 80MHz with 8Mb of RAM and a small capacity HDD (maybe 40Mb, maybe more, I don’t remember). Pentium CPUs have appeared in 1993 and I couldn’t afford one at that time but they were slow and not close to any modern Pentium.

          1. I find it hilarious how you can spot the people born in the mid to late 90s (i.e. the guy you replied to) because all they’ve ever known is 2000s technology and they think it was always that way.

            I graduated high school the year the Pentium finally got its wings (Pentium Pro, 1995) and the computer I got as a graduation present was a laptop with 4MB of RAM a 486SX CPU (that’s right, not even DX; for you young’uns imagine having a single core CPU in a quad core world). Kids these days…

          2. I was actually just being a smart associated.. My family’s first computer was actually a tandy brand, which was Intel 286, that my parents bought in Syracuse NY while there for the weekend for a wedding. It cost over 2 grand back in 89 or 90… Can’t really remember, as a it was over a third of my life ago. It ran ms dose and had a whole 40 meg hdd, that you actually had to run a parking command to the read head before powering down… And no, I don’t remember the ram, sorry for my lack of memory accuracy in details I didn’t know about back then, nor care about. I was more interested in battle chess and f15 strike eagle 2 games…

          3. My first computer was a TI-99/4A. My second was a Xerox 820-II. Third was an IBM PCjr, followed by an IBM 5150 PC, eventually upgraded with a 5 megabyte hard drive. From there it was a succession of XT and AT clones (my ‘peak’ 286 box had 12 megabytes of RAM, 640K onboard, the rest on ISA cards) followed by various 386 then 486 boxen. My First Pentium was a Packard Bell slimline desktop 486 with a PODP5V63. After that it’s been rather nondescript generic systems, with a few year’s stretch of now-classic Macintosh systems in parallel from the late 90’s through early 21st Century.

      1. There is a huge amount of difference between an ATOM and a i5 running at the same clock speed. Trying to compare x86 and ARM while ignoring the implementation details is dumb.

        This chip is supposed to fit into a even lower performance market segment than ATOM, so it is a beast for embedded, but don’t expect PC performance.

        The Intel chip is probably faster, but not because x86 is a superior ISA. The lithography is smaller, the cache is larger, and the gate count is higher. The cost is also, much, much higher.

      1. According to Anadtech’s announcement,
        “Intel will be giving away 50,000 Galileo boards to 1000 universities worldwide over the
        next year and a half or so to spark development. Boards will be available for sale by
        the end of November, at a price under $60.”

        Intel is probably selling these boards at cost to spark interest for their chips.

    1. UNfortunately no… MAc requires a minimum of half a gig of ram. Windows Xp might be able too. Attempting to install XP on my pi. If it works on pi it might be able to work on the Galileo.

      1. Wow, you dont have much of a clue about windows on pi vs windows on this do you.
        Pi has to emulate it, about a 10x performance hit as a general rule, can be worse. Pi cant cope with windows 95 emulated.
        This is native x86, no emulation. Still not likely we will be able to run windows on it, but in theory it would be better suited than the pi

        1. I take what I said back. I had no idea about XP on pi now i have another wrinkle in my brain. (since i learned something new ;) ALthough mac would never be abel to run on 400 mhz and 256 gigs of ram. And even if it could it wouldn’t be effective since it doesn’t have video out.

          1. We once run WinXP on a machine with 64Mb RAM, but for installation it required 128.
            Not an experience to recommend. But we were students with too much time on our hands…

        1. We live in the Google era, people don’t need to be directed to resources anymore, unless of course they are lazy. Laziness is even worse than ignorance… they just end up wasting our time.

  2. No, it does not look good !

    Some words of caution

    Intel wants to eradicate other ARchitecture Manufacturers, be aware of that.

    The market for Arduiono, ARMbased something is flurishing & prosperous, every competitor overpowers the other with interesting things, and Intel just jumps in hoping the X86-compatible will eliminate the ARMs.

    intel == x86 == customer lock in

    1. you are right, i think they do feel threatened. ARMs have become more cost effective and strong enough to replace x86 in many places. There are people trying them in servers as well.

      I’m curious about the price and power consumption of this thingy.

    2. I’m sure that’s what they’d like, but they’re so far from that goal that it’s not a worry. I’d actually like to see Intel’s market share grow in the portable market to give ARM a bit of competition, because in that segment, ARM == customer lock in.

        1. Yeah, I probably misspoke. I was referring to the high-end portable market (cell phones, tablets, etc), where ARM has a death-grip, while the Quark (and the MIPS chip contained in the TL-WR703N) are more for the low-power embedded market, which has many more players (Atlmel, Microchip, MIPS, Freescale, etc).

      1. Not sure how can you speak about lock-in with ARM when you have multiple vendors competing in the market. There is Qualcomm, Nvidia, Broadcom, Samsung, even Chinese Al;winner at the high end, at the low end you have tons more – NXP, FreeScale, ST Micro, TI, …

        ARM is an architecture, they don’t make any chips themselves, only license the IP. And if you don’t like ARM, there are still MIPS chips arround.

        By comparison, x86 is produced only by a single vendor (not counting AMD here, as they have zero presence in the embedded market) and are really struggling to get any meaningful presence in the embedded market since they dropped the XScale line (which were ARM CPUs, btw).

        1. When I said vendor lock-in (in the high end portable market, I later clarified that the low-end embedded market was much more competitive), I meant for the ARM architecture. The Qualcomms and Samsungs can certainly innovate since they design their own chips using the ARM architecture, but the limiting factor would be the architecture if ARM stopped innovating. It’s a different situation from Intel since having various manufacturers would keep the price competitive, but the problem is that a single architecture wouldn’t drive innovation.

          1. That part requires a heat sink and a northbridge chip and a clock generator and expensive DRAM. It only comes in a BGA package that’s impossible to work with unless you have expensive pick and place at your disposal. t’s just not competitive in this arena.

          2. Not competitive in the embedded arena? we’ll just totally ignore the cars, planes and industrial machines using the geode as their CPU, and the set top boxes and TV’s (which is a rarer scene for them but one they have appeared in). Not everything embedded is making little robots which drive around the room avoiding obstacles or making remote garage openers.

          3. Geode was not intel. It was rebranded from a national semiconductor chip which in turn used a cyrix core which also was not intel (had origins as intel pin compatibles but were original implementations). Also currently the Geode NX chip uses the K7 core architecture from an athlon entirely dumping the old cyrix core. So no, it was not intel.

    3. Sure, of course Intel want’s to eliminate the competition. Isn’t that what every company wants? So what? They are no where near in a position to do so. At the most they might inject a little more competion into the embedded market. How is that bad?

      As for vendor lock in… what are you smoking? Several companies have made x86 compatible chips. I’m not sure if any are making embeded ones or not but if not then it’s probably only becauase they don’t see a market for it. More people using embedded Intel x86s would do just that.

    4. This is a last-ditch effort on Intel’s part to save the i386 architecture. PCs are becoming irrelevant in the era of mobile devices, and they had their heads in the sand the whole time ARM took over that market. Too little, too late, I think.

      1. Without all those x86 CPU’s in datacenters your mobile device would be nothing more than a nintendo DS. No Facebook, no Twitter, no email, no Cloud, no Appstore, no… Just a gaming platform where you’re mother can call you on to tell you dinner is ready.

        1. From the intel FAQ:
          The GPIO output pins on Intel® Galileo are provided by an I2C Port Expander that is running at standard mode (100 kHz). Each I2C request to update a GPIO requires approximately 2ms. In addition to software overhead, this restricts the frequency achievable on the GPIO outputs to approximately 230 Hz.

    1. Intel has more resources at their disposal than many small countries, they can make and sell anything they want to, and do it competently. They are new in this arena. They are used to very large profit margins and customers like dell and hp and apple that can make their own computers from scratch. This is a very different experience for them and you can’t expect them to have a broad product line on day 1. They have no idea how many of these they can sell, and they are not going to pour money down the drain on R&D unless they know there is profit in the long run.

      1. Intel has the cash to buy up ARM and ST and NXP and TI and fold them all into a new “embedded” division. They could own the ARM market tomorrow if they wanted to. The reason they haven’t done it is that they prefer markets with higher profit margins. They typically leave the cheap end of the market to the bottom-feeders like AMD.

        1. The flip side to this is that if Intel’s investors are unhappy with the reduced profit margins of the embedded world, they will drop the whole thing like a bad date and go back to what they do best.

      1. Nobody runs the crystal clock at the CPU clock frequency any more, at least in the embedded world. You don’t want 400 MHz signals on your PC board. You don’t want a 400 MHz crystal on your BOM. Everyone uses a low frequency crystal and a PLL multiplier.

  3. Just want to point out… there are a lot of differences between Galileo and the Arduino spec:
    The current supplied on the GPIO is vastly less than Arduino – so some power-hungry devices, sketches, or samples won’t work with this. source/sink is 10/25 mA vs 40/50 mA on Arduino
    The analogs in support 12-bit resolution whereas Arduino is 10-bit. Normally more bits is good, but this means that code written for Arduino that uses analog in won’t work unmodified on the galileo
    analogReference() and AREF are unused in galileo
    SPI doesn’t support slave so no SPI-SPI networking

    So this is cool, and in general a good idea for Intel, but don’t mistake this for a pin-for-pin line-for-line arduino clone

    1. And the inputs are slow. But still, compatibility with a lot of shields is probably a good thing. A lot of applications don’t need speed or the Arduino’s beefier drive capabilities. Intel should create/port some compute-intensive libraries so that Arduino-sketch writers can take advantage of the CPU horsepower available.

      The big question is “how much?”…

  4. One thing I like about this board is the selectable 3v / 5v output. Translating between logic voltage levels is becoming more and more a part of almost every project. At least with this board you can do one or the other (but not both). Wish they had added more GPIO while they were at it. I’m also wary that the power usage is kind of vague.

    Short rant:
    Instead of all those IO connectors, especially the PCI slot, please please please GIVE ME USB PORTS! Almost any device I want to add can be done via USB: ethernet, bluetooth, wifi, storage, audio, mouse, keyboard, displays, etc. But in order to do that I need places to plug them in. Sure you can add a hub, but that adds another power supply, and that doesn’t increase the total bandwidth. A single root port isn’t enough.
    End rant.

    1. Use a proper USB hub.

      Most machines with multiple USB ports are running them all into an internal USB hub so you really aren’t losing out on anything. That internal USB hub is still eating power.

      1. They do make minipcie USB 2.0 and USB 3.0 cards. The beauty of minipcie is that you get to choose what goes in there. WiFi, Video, RS232, SATA, whatever. There is a whole market based on adding ports to the various embedded platforms and laptops with minipcie.

  5. Reading some more, this board requires a heat sink for elevated temp operation. If it’s throwing off that much heat then it’s not competing with ARM. This is just an underpowered PC pretending to be an arduino.

        1. that’s why I said 99% and not 100%

          What’s the point of all that extra bandwidth? This machine has miserable I/O bandwidth so you are not going to max out a 2.4 connection even if you get all of the pins working.

    1. You beat me to it. This looks like a stripped down Alix board with the name Arduino stamped on it. The hardware compatibility with Arduino shields may be a good thing but
      using their awful and dog slow IDE… yikes!

  6. I think the idea here is that Intel is going to listen to everyone whining “but it doesn’t have A” and “it uses too much B” and then they will spin it again after they hear what people want.

  7. Anyone know what that plastic piece on the back is for? It’s at the bottom of the board in the picture. The plastic part has “LOTES” printed on it. I’m not referring to the PCIe connector.

  8. They could have added a SPI expander. This would let them drive multiple
    SPI ports including the ADC & I/O from that single port. This would improve
    I/O bandwidth a lot.

    The expander is just a simple ‘138 decoder/demultiplexer that splits off
    the SPI strobe signal. They could either use I2C port or steal a few MSB in
    the SPI to control that. I have done the latter eons ago and it worked well.

  9. Hmm… I’ve been thinking about building a “laptop” into a briefcase for a while now.

    The idea was to use a beaglebone or a raspi so that I have IO pins and add a breadboard, bananajacks with access to the various voltages in the power supply (i was thinking of going with a pico power) and whatever useful things I can come up wiith.

    The point was to make it a portable development machine.

    Anyway, depending on the price maybe I want to go with one of these? With the mini PCI slot I could have a real video card. I can still do everything else via USB. And.. being x86 would probably open up my options for software. If nothing else I could actually run Flash for websites that still require it. (not sure if there are any I care about anymore…)

    hmm… decisions decions…

  10. I guess the whole point in making these new boards all together is to reduce the easy production of clones by using special parts from their ‘partners’. This trend is significant throughout the new boards:
    Yun- Atheros
    TRE- TI
    Galileo- Intel

    Gone are the old days when we could make something on arduino and then make our own PCB and the ‘friendly AVR’ on it and it would sit there doing the job! Thats a less ‘hobbyist’ friendly approach and might not be inline with arduino`s original goal of ‘everyone`s microcontroller’. They seem to be competing with Raspi.

    1. Hmm. . . you make an interesting point but perhaps a little conspiratorial?

      My perspective is that they’re trying to create better tools by partnering with chip makers and at the same time, playing these giants against each other to foster a competitive embedded development marketplace.

      It is amazing to think that this “toy for hobbyists” is now powerful enough to demand the attention of such giants of industry.

    2. IMHO it is more likely that the chip vendors are licensing the name
      “Arduino” for marketing their stuff than the other way around. This is
      the next step up from the usual eval boards by using a well marketed
      “standard” pinouts, software framework and take advantage of the I/O

      “It’s an Audrino! I know this!” is probably what they are after.

      It is like the Apple market. Trying to market the segment that
      traditionally not technical enough to use their product as a fashion for
      a certain life style.

      Whether or not these small business peddling “Arduino” on the web can
      return a profit for the chip vendors, only time will tell.

      At some point if you want to use the Ethernet or USB ports effectively etc,
      you would want to have some form of OS (e.g. a RTOS or *nix) to manage
      the devices. It gets very messy to do that with a singled thread
      “Arduino” code.

      1. “IMHO it is more likely that the chip vendors are licensing the name
        “Arduino” for marketing their stuff than the other way around”

        Good point and probably more realistic.

        “At some point if you want to use the Ethernet or USB ports effectively etc,
        you would want to have some form of OS (e.g. a RTOS or *nix) to manage
        the devices. It gets very messy to do that with a singled thread
        “Arduino” code.”

        Yes. It seems like Arduino is headed for confusing times where to take advantage of the hardware one will have to depart Arduino land anyway. However, if the IDE were to evolve that would be cool. . . though it would be akin to re-inventing the wheel.

        Perhaps I should spend more time with Linux.

    3. I don’t think they are trying to kill clones. They are just making a SoC board that one can easily transfer knowledge, skills and shields from working with Arduino microcontrolers.

      I’m pretty sure these new system on chip style ‘Arduinos’ aren’t meant to replace the old microcontroler ones. I certainly hope they aren’t! It wouldn’t make sense, microcontrolers are good for some things, SoCs are good for others. It’s apples and oranges.

      Even if Arduino is trying to replace their microcontroler board lines with these I’m pretty sure they wouldn’t be able to kill ‘arduino’ anyway. It would just be forked.

  11. So. . . why hobble such a powerful system with Arduino compatibility? Isn’t the fact that it’s running Linux more exciting?

    I’m both interested and horrified to see what the Arduino kids do with this.

    1. “Isn’t the fact that it’s running Linux more exciting?”

      No, not really. And yes, I AM a huge Linux fan. But… Beagleboard has been around for how many years now? Then came RasPi and now there is another Linux SoC board anounced it seems like weekly.

      Is Arduino compatibility exciting to me personally? No, I see most shields as being way overpriced versions of something I could just buy and wire up myself much cheaper. It’s an Arduino tax! Also, with a full Linux install to work with I could probably do much better than the Arduino programming environment. I’m guessing you feel much the same.

      But.. to someone used to plugging shields together, with a firm grasp of the Arduino environment and not so much Linux experience I can see how this board would be very exciting! Yet another Linux board… not so much. We already have that. A bridge between the hobbyist microcontroler and embedded worlds? We never had one quite like this.

      What does excite me personally though… that mini PCIE slot. Finally! It’s a board that we can expand in pretty much any way we want, USB, PCIE and GPIOs. It’s about time! I’m both eager and dreading learning the price.

    1. meh.. There are too many other options already out there for that. What’s one more? The only place I see this as being exciting is where you have a use for GPIOs (especially Arduino shield form factor) AND Linuxy SoC stuff. Otherwise, I’m not necessarily saying this board is bad but if you are only interested in one side or the other what is new about it?

  12. These days what interests me most in intel x86 compatible boards is running old versions of Micorosoft (or other closed source) OS.

    Any chance this will run Windows NT Server, Windows 2000 or maybe just DOS, Novell,, SCO on some emulation/shim layer (and some remote/virtual framebuffer)?

  13. okay so maybe the ardunioness is lost on me….and I will admit to being 6’2″ and carrying a Galaxy Note 2 so my definition of pocketable may vary from that of teenier men…..

    But I have to challenge this statement, “finding a small pocketable computer based on an Intel processor has been difficult to find”

    Are you sure you didnt mean to write…. “difficult to find documented implementations” or even “too difficult for us to figure out how to utilize the standardized interface of one of the many small form factors” or more likely did you mean “”difficult to afford”? because 30 seconds at mouser disagrees with availability being an issue. And their paltry offerings are but a small scrape of the modules, units and formats available in small factor.

    1. so I just realized a couple of ITX boards snuck into that link…..while cutting them out I also removed the epic boards….even though 165×115 isnt that much larger then my phones 151X80.5

      I dont in anyway think anything in that link is directly comparable to an arduino, a raspberryPi or even this Galileo….I am just in disagreement with the pocketable intel statement.

      However, that said….with no real use in mind..nor need..I have a pile of comexpress and pc104 boards already…..but I am ordering one of these I havent played with the NUC boards yet….and I cant resist an I3 with dual hdmi, thunderbolt, 2 pci express, and 5 usb ports under $300.

  14. I make robots all the time. The number one thing I can’t seem to do yet is connect a web interface and a DB to an arduino yet. This board probably doesn’t have enough HD or RAM to pull it off and nobody that I know has written a bridge between pi and arduino.

    This is a nice step up but it doesn’t solve any practical problems that I know of.

  15. I’d buy a budget laptop running that CPU, but not a Arduino/clone (finally the >$200 x86 PC is doable and instead they make a arduino?).
    Also, is it actually “Arduino Pinout Compatible” (each pin doing the exact same job/format) or just “Compatible”? just for future reference as there have been products touting “Arduino Compatible” when the actual pinout is mostly different but a shield would fit (even if it wouldnt work 9/10 times)

  16. As much as I dislike Intel’s business practices. I will have to say this is really awesome. I wonder if Windows can run on it…add GSM, and a touchscreen LCD. would make for a really nice 3dPrinter

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

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