back to article Mad Raspberry Pi boffins ripped out its BRAINS and SHRANK them for your pleasure

The makers of the micro-sized Raspberry Pi teaching computer have come up with an even smaller version of the device, this time aimed at hardware hobbyists and device builders who want to create their own printed circuit boards (PCBs). The forthcoming product, dubbed the Raspberry Pi Compute Module, forgoes the various ports …

COMMENTS

This topic is closed for new posts.
  1. wolfetone Silver badge

    Shut up and take my money NOW!

  2. VeganVegan
    Thumb Up

    This raises a potentially interesting possibility

    Why not take all the obsolescent 'smart' phones, and recycle their compute modules?

    1. DaddyHoggy

      Re: This raises a potentially interesting possibility

      Because they're all non-standard in format; proprietary and non-documented?

      Apart from that it's a good idea.

      1. James Hughes 1

        Re: This raises a potentially interesting possibility

        They're a bastard to unsolder as well.

  3. Zola

    Clustered Pi Blades?

    Imagine packing 4 or 8 modules into a relatively tiny package. Include a single Gigabit LAN port, internally switched between each Pi module. Could be a lot of fun.

    Sadly the reality is that a system based on a decent Octa-core SoC will likely be cheaper and more powerful.

    1. Anonymous Coward
      Anonymous Coward

      Re: Clustered Pi Blades?

      "a system based on a decent Octa-core SoC will likely be cheaper and more powerful."

      Maybe so. But the phone/tablet-based one will be:

      . completely undocumented unless you're a tier 1 design outfit,

      . obsolete every couple of months.

      . unobtainable for Joe Public except as part of a phone/tablet, with prices to match.

      Which hopefully won't apply to the RasPi approach.

      Horses for courses.

  4. Pete 2 Silver badge

    An old dog

    ... see it's latest trick.

    I really can't see the point of this. The Pi's processor is slow, limited and obsolete - why would people want to use it in new designs? Why would anyone expend time and effort in redesigning such a board when they could have done the world a favour: tossed it and come up with a modern design, using contemporary components and with an up-to-date specification?

    Now I know that people get attached to their first car, their first computer and old valve radios. But the only reason I can think of for people to still have some lovin' for the Pi is because they haven't raised their heads up and seen what 2014 has brought in terms of new designs with so much more usability. if they did, they'd never go back.

    1. Mage Silver badge

      Re: An old dog

      A HUGE number of projects & designs do not need even as powerful as the Pi, but well documented dev tools, less than £200 cost and convenience. I've used an embedded 486 kind of thing at £200 because it had the I/O I needed and a single 12V jack.

      There are LOADS of ARM chips sold every week replacing the kind of stuff PIC, Atmel, Z80, 8051 etc is used for. Down to $1 each or less. These can't even run Linux.

      State of the Art and CPU power isn't the only consideration.

      Microchip sells LOADs of PIC chips which have a joke architecture over 30 years old. Probably every IR remote control handset has one.

      Horses for courses.

      1. P. Lee

        Re: An old dog

        > A HUGE number of projects & designs do not need even as powerful as the Pi,

        Indeed, but I think that's the point of the original poster. It looks as though this design is to increase chip density. RPi's are useful for the places you can put them, not compute density. If you need compute density, this isn't the chip you are looking for.

        OTOH, it might just be about repackaging to provide better IO or the start of a standard ARM CPU interface (I jest :) ). Here's Pi v2. When Pi v3 comes out, swap the DIMM or pop an octo-core +GPU on a DIMM and insert.

        1. James Hughes 1

          Re: An old dog @pete

          The huge majority of embedded applications don't need a huge amount of horsepower. But they might want something like a decent display/video for the UI. This is the target for the CM, as the 2835 does have very good graphics and video capabilities, with a slightly weaker ARM (although you can do a hell of a lot in 700Mhz)

          Now, you cannot buy the 2835 in the sort of quantities that most embedded devices are needed in (production runs 10-50k or so), so this module is a quick and easy purchase for that sort of market. It also takes a lot of HW design work away, and also software work as it already run Linux. And of course, you can prototype on the Raspi itself.

          I think it will sell quite well, there's certainly a big market for something like this, that Pete obviously hasn't knowledge of.

      2. Steve Todd

        Re: An old dog

        I'm looking with interest at the ALTERA Cyclone V SE at the moment. A twin core ARM Cortex A9 @800MHz combined with an FPGA on one chip. More IOs than you can shake a stick at and lots of logic resources. You can get a complete dev board, including 1GB of RAM connected to the A9s and 64MB to the FPGA, 85K of logic elements, Gigabit Ethernet, VGA, 24 bit audio in/out, 2 USB host and a USB slave port, a micro SD slot, two 40 pin IO ports and a built-in USB JTAG programmer for less than £170

    2. Anonymous Coward
      Stop

      Re: An old dog

      "I really can't see the point of this. The Pi's processor is slow, limited and obsolete - why would people want to use it in new designs?"

      Because your washing machine doesn't need a 8 core processor with 4 gb of RAM?

      1. DropBear

        Re: An old dog

        Because your washing machine doesn't need a 8 core processor with 4 gb of RAM?

        True, on the other hand neither does it need Ethernet nor an OS (nor a touchscreen for that matter). There used to be this fable about how (not) to (over)engineer a toaster - "the object oriented toaster", I think it was called; the sad part is people these days don't seem to see anything wrong with the approach of the guy the king has wisely ended up beheading.

        1. John 48

          Re: An old dog

          > True, on the other hand neither does it need Ethernet nor an OS (nor a touchscreen for that matter).

          Does the phrase "the internet of things" ring any bells?

          Mundane household appliances with ethernet / wifi and TCP/IP stacks etc are gaining in popularity, so this is quite an elegant solution for DIY embedded applications which nicely circumvents the slightly clunky nature of using a full fat PI for these jobs.

    3. SumDood

      Re: An old dog

      "I really can't see the point of this. The Pi's processor is slow, limited and obsolete - why would people want to use it in new designs?"

      Funny thing, I was thinking exactly the same about the wheel.

    4. SumDood

      Re: An old dog

      You don't join the queue to buy the newest iPhone by any chance, do you?

  5. Jess

    Re: I really can't see the point of this.

    Because if they release a new faster Pi, it will be a simple upgrade for any kit based on it. (Assuming they don't do anything stupid.)

  6. heyrick Silver badge

    Thinking in the future...

    "using contemporary components and with an up-to-date specification?" - because more up to date hardware comes with a price tag somewhat above that of the Pi?

    I note that the core module and the base together will cost a bit more than a regular Pi (possibly near twice as much) while offering the capabilities of a Model A with a heap of additional I/O. While this on its own is nice for people wanting to play with lots of I/O, it is also useful to break away from the association that "a Pi costs this much". In a while, when we can expect a Pi-with-bells to cost a little more than the requisite $35, they might release the Pi "Master" which will cost maybe $65 or so and contain a more up to date multi-core ARM and 1GB, etc etc. And, of course, I would expect the option of a Pi "Master Compact" to be the same sort of thing in this SO-DIMM form factor. ;-) [the hard part will, of course, be making the I/O of completely different devices match up in some sort of sane way!]

    1. James Hughes 1

      Re: Thinking in the future...

      The base is not intended to be a big seller - it's a development platform that people can use to develop their own PCB's. And it won't be particularly cheap, because it's low volume, and not optimised for cost. I'm expecting going on $100. (but I don't know)

  7. Mage Silver badge
    Thumb Up

    I like it.

    The existing board is an awkward layout with connectors all around it, and not ideal for GPIO projects.

    A board to one's own desire with all the HW that is driven by the GPIO and more sane PSU and I/O will be good.

    I look forward to this and the documentation. Where do I find the correct connector for Eagle?

  8. TheOtherHobbes

    It's still too expensive

    I know it's super-cheap as a computer, but if you want to start automating your home >£40/module is still too much for most people.

    It cost me around £80 to automate the home heating and put it online. That's cheaper than a commercial equivalent, but it would be really, really useful to have something cheap enough and small enough to replace light switches and power sockets - ideally something with WiFi or Xbee or even Bluetooth that 'just works.'

    I'd rather pay a tenner for a third of the processor power and a quarter of the size than £50 for this blade-like thing. Because all you're getting is some extra pins. And there's still no A/D. (Yeah, Arduino. I know...)

    1. Anonymous Coward
      Anonymous Coward

      Re: It's still too expensive

      Your tenner would almost stretch to a TL-WR702 if you order online. Much less powerful than the Pi and no video, but it is about half the size, has Wifi and USB and runs Linux, so can host Arduino for I/O.

    2. Yet Another Anonymous coward Silver badge

      Re: It's still too expensive

      So a single one of these as the controller and a bunch of $5 arduinos managing each thermostat

    3. GothBoy

      Re: It's still too expensive

      What system did you come up with to control your indoor environment?

    4. Stoneshop

      Re: It's still too expensive

      I know it's super-cheap as a computer, but if you want to start automating your home >£40/module is still too much for most people.

      If you're designing your home automation with a Pi for each node, you're doing it wrong. Most endpoints can be easily handled by an ATMega, even at several endpoints per. Then a Pi (or BBB, or whatever else you like in that class) to tie it al together and provide a nice UI.

    5. A Known Coward

      Re: It's still too expensive

      For home automation you only need a single Pi with XRF radios (Slice of Radio £9.99 for Pi, XRF £11.88 module + circuitry) dotted around the place to control relays, take temperature readings etc. XRF is cheaper than XBee and incredibly easy to use.

  9. Anonymous Coward
    Anonymous Coward

    Mmmm, I smell a raspberry pi laptop

    * Take one NOS Motorola Atrix dock

    * Insert one Raspberry Pi Compute Module into the flip up bit

    * Solder exitsing wires to the respective pads on the edge connector

    * ...

    * Profit.

    I wonder if a Chinese Allwinner CPU based clone will be available before these are ready for delivery this time.

  10. Anonymous Coward
    Anonymous Coward

    A bit like the SimmStick

    Around 1994 "Antti Lukats, an Estonian"* came up with the idea of the SimmStick, here's some pictures

    http://www.mcselec.com/index.php?option=com_content&task=view&id=22&Itemid=41

    It made use of the 30 pin Simm format of PC memory modules at the time. (the pinouts were different)

    I've still got a DT003 PSU board with a SIMM100 board with an Atmel AT90s8535 on it, in its day it was a well cool way of playing with the AVRs.

    The advantage was that you could mix analog and digital MCUs with relays etc. using the base board as a bus.

    *He was always referred to on forums as such, I don't know why.

    1. Pete 2 Silver badge

      Re: A bit like the SimmStick

      Yes it does. Or (my favourite from that era): the TINI. That was a SIMM format with everything on board - except USB (it wasn't around then) and video. It had I/O serial, ethernet, n/v memory and RAM and a 100MHz 8 bit processor. And you wrote code for it in Java.

      Whaddayamean you've never heard of it? It *should* have been a success, but it wasn't. Just like this proof of concept won't be. The reason: it wasn't picked up and integrated into "stuff". And neither will this version of the Pi. There simply isn't the need for it in industrial systems.

  11. mIRCat
    Coat

    Is that a PC in your pock or are you just happy to see me?

    "on a 67.5mm-by-30mm board that can be slotted into a standard DDR2 SO-DIMM connector."

    Holy miniaturization Batman!

    Nope it's just a computer.

  12. Christian Berger

    We need a "communicator" case for that :)

    So we'll have a real computer in a pocket.

  13. Scott K

    Of all the things Rpi needed

    This is not one of them as I see it. They need to update it so it can be used for a multitude of everything. This to me is a Rpi in a different format to sit on a different smaller shelf gathering more dust.

    1. Slartybardfast

      Re: Of all the things Rpi needed

      If you have a Rpi sitting on a shelf gathering dust then please send it to me as I can make really good use of it. I'll pay postage etc.

  14. DropBear

    If one does only need a headless device...

    ...I still find the Aria G25 SoM for a 24€ unit price rather more appealing - it even includes the Ethernet PHY on-board, while this does not seem to. Granted, you do need your own SD card to run it from.

  15. John Robson Silver badge

    Direct wire to screen etc.

    Now - who has the psion 5 patents again?

This topic is closed for new posts.

Other stories you might like