I'm really excited by this project. I think it's a great idea. I very much hope they sell them to anyone, maybe at a slightly higher price to subsidise the others.
But there's a tiny[1] hint of regret that modern programming is necessarily abstracted away from the hardware. When people were using the popular home computers of the 1980s they would be able to use some version of BASIC, and then also some machine code if needed. And that wasn't something just limited to the brightest smartest young programmers; it was above average but nothing extreme. I learnt Z80 ASM on a suitcase instructional computer, turning fans on and off or stepping through traffic light sequences on LEDs. Then I made a resistor ladder DtoA convertor and pushed data out the parallel printer port for audio. (Something also done by No$GB emulator.)
That seems to be something that children will be missing out on, unless this project gives them enough boost to then investigate things like arduino or similar.
[1] Not a criticism in any way of the Raspberry Pi project.
One of the operating systems that may be ported to the Pi is RISC OS, which includes a reasonably decent implementation of BASIC (http://en.wikipedia.org/wiki/BBC_BASIC) which itself contains a built-in ARM assembler.
A student could quickly and easily start writing graphical programs (plotting randomly coloured circles onscreen, for example) or simple games, without needing to download, install and use a compiler.
We're literally returning to the world of 10 PRINT "Hello" : 20 GOTO 10... in the good "very accessible programming" sense, not the bad "...considered harmful" sense.
They could have achieved higher density, but it would have been more expensive. The density is in service of low costs, so that wouldn't have made much sense.
Yup, it was interesting reading some of the discussions about the design of the device. PCB cost was a major consideration. They aren't exposing some of the I/O pins on the SoC because it would require a more expensive PCB to do so.
Custom small-time boards get expensive quickly when you shoot for tighter tolerances, which are necessary for truly tight spacing. If Raspberry Pi expects to sell 10 million units, then even a fancy board will cost nothing, but I doubt they are planning on such numbers...
$100? heh the 16gb 4s is $200(+fees, tax, etc.) because its subsidized with a 2 year contract. I think it's something like $650 if you buy it unlocked. compared to a $25 - $35 computer? Apples and oranges ;)
What's their initial production run going to look like? There's a whole lotta geeks who will want this - and at that price, they'll want 10. ($250 for more computers than I'll know what to do with? I'll take 20!)
What would be the best way to interface a 2.5" hard drive with one of these? A USB to sata controller? That seems bulky if you want to have it in one case. I'd like to find a compact way to add some storage to the device as I am interested in using it as a media player.
This is quite awesome. I can imagine many great things coming out of this. Such cheap computers, will bring computers (and therefore the internet) within range of many people in developing countries.
One potential problem will be the cost of monitors, which might exceed the cost of the computers itself :-( But hopefully someone can make cheap monitors soon..
What doesn't look legit? The PCB silks look typical, the board size is about right for a small SoC-based design. You just need a lot of space for those large connectors like SD/MMC and HDMI.
There are other designs recently launched of the same size and capability like BeagleBone or the Gumstix line.
The part that's suspicious is the $25 price point (the BeagleBoard is $125, the BeagleBone is $90, the Gumstix is $150-$250 depending on configuration). How many products can you name for under $50 (2x the cost of the Raspberry Pi) that approach its specs and capabilities? Also, they're aiming for user-friendliness to the point where it is supposed to be an educational product for young people.
These are both really, really hard to do. The foundation seems to be approaching this goal fairly well; if it is vaporware they have documented it quite well indeed. Having volunteers like their director and trustee Eben, who is an architect at Broadcomm, and trustee and operations director Pete, who is the managing director of a PCB manufacturer, can't hurt on getting this thing spec'ed out. Whether or not those relationships will last once this gets rolling is the shaky part.
Don't get me wrong, the team seems like a great group of people, and I trust them to have good intentions. However, I don't give anything approaching trust to Broadcomm.
Roku has something with similar specs for $50 retail (which means they sell it to retailers at $30 at most). And Apple has something called "AppleTV", which has a little more hardware on it (wifi, power supply, ir remote receiver + remote, 8GB of flash), and sells for $90.
The tplink wr703n is a 400mhz MIPS computer with a USB port, an Ethernet port, and wireless networking which sells for about $20 (on volumerates.com, which is unfortunately down at the moment). It has a completely functional Linux port and is supported by OpenWRT. Not as fast, and fewer peripherals than the Raspberry Pi, but an indication that the $25 price point on the pi is doable, I think.
Amazing to think the impact of these low-cost computers.. goodbye expensive routers, NAS, set top boxes, PVRs, media centres. It can't be too long before they (ARM devices) start eating into the low end laptop market.
My hunch is that the final price will be something slightly more, or else Broadcom is going to subsidize the project to some degree. They're already easying buyers into the idea of a $35 "Model B" version that has the ethernet magnetics on it and more memory etc.
Why don't thes people release their files as they work on them? Why don't they resell the bare SoCs at cost or a slight markup so we can use them for other projects?
Something has always struck me as fishy about this whole thing.
You couldn't build one yourself anyway, because Broadcom won't sell you parts or even talk to you unless you're doing volumes like 100k/quarter. And then you'll probably need to sign an NDA to get the datasheet.
I see the Raspberry Pi project as an attempt by Broadcom to build some goodwill - but they seem more interested in doing so among society at large by producing a super-affordable computer for kids, than among nerds by doing an open source project. That's Broadcom's prerogative and I applaud what they are doing for what it is.
Personally though, as a hardware hacker, I am more likely to buy similar products using TI (BeagleBoard) or Freescale (Chumby Hacker Board) or Atmel parts, as those manufacturers have histories of being a bit more friendly to hackers, hobbyists, and small volume buyers. The BeagleBoard in particular is open source, there's development chatter on an open mailing list, and TI sells the same OMAP SoCs to small volume customers (right down to the possibility of a hobbyist buying one or two through a distributor, although at that volume you'll pay $50 for just the bare SoC).
Rasberry Pi is not a project run by broadcom, just using their SoC (and featuring one of their employees).
They are selling the first round to nerds / hardware hackers, and I doubt many others will choose to pay 4x the price to keep using the beagleboard.
I understand the project isn't actually run by Broadcom but I think the fact that Broadcom is selling them chips indicates some level of interest in what they are doing. They aren't exactly 'just another customer'.
Why aren't they doing those things? Perhaps because that isn't what they set out to do:
We plan to develop, manufacture and distribute an ultra-low-cost computer, for use in teaching computer programming to children. We expect this computer to have many other applications both in the developed and the developing world.
How does selling SoCs at cost or a slight markup serve their aims? My guess, it doesn't.
But what does that get you? Still need to plug into a monitor, keyboard, mouse. So you can only use it where those things are.
For real portability, how about an SD system thumb drive I can plug into any computer, and it boots as my computer. That would be portable. Useful for a student lab, at my Mom's house, in a hotel.
> For real portability, how about an SD system thumb drive I can plug into any computer, and it boots as my computer.
Linux distributions have been doing this very successfully for the past 9 years (knoppix paved the way, but almost every linux distribution today can run live from a CD/SD/DiskOnKey, and store your data on an SD/DiskOnKey (the same one, if properly partitioned).
And it's unlikely a hotel or your student lab would actually let you reboot into your own distribution; once you have control of the hardware (which you must have), you practically own it from that moment.
CherryPy is actually a much better solution than yours:
Your lab, and your hotel room, is likely to have a screen that can take VGA or HDMI input - so you've got a display; You can get foldable silicon keyboards for $10 or so, and a small travel mouse for $10 more. So, for $45, you have something you can _already_ use at your lab or hotel room, with no one having to fear you're putting viruses on their system.
A comment like Joe's really surprises me. Time and again I see brilliant coders who for whatever reason cannot see the path to making their own live USB/SD sticks.
But there's a better solution than beagle's suggestion.
Sell hardware that has no bundled OS.
Sell different OS's on removable media (e.g. memsticks).
Or users can create their own OS memsticks.
Apple is primarily a hardware company, as evidenced by the value they place on their design team and design patents, and where they derive the lion's share of their revenue.
If I could buy Apple hardware without it being tied to Apple's OS's, I would be willing to pay Apple hardware prices more frequently.
I've read all of Joe's comments in this thread, and I'm not really surprised. He is either a troll (and not a very successful one if he is), or, more likely, a bigot who believes that the any software worth running costs >$100 (office, photoshop).
Unfortunately, he represents a non-negligible majority of users out there. But I care not discussing anything with him and his like on HN. I've got better use for my time.
Maybe he works for a company that sells software priced at over $100?
I have no doubt such companies have employees, or contractors, who not only monitor message boards on behalf of the company, but who post comments on them in a deliberate attempt to influence discussions in ways favorable to the company's interests. Whether they are effective is another question. But they are trying.
Younger readers should be aware of this as they are the ones being targetted.
Perhaps it's not that a majority of users are convinced price equates to quality control so much as a majority are aggressively manipulated in terms of what information about software they are exposed to.
It's actually pretty easy to do. I've seen several Windows, several BSDs and linuxes as well as some open-solaris derivative on x86 Macs.
The PowerPC macs are a bit less well supported just because there are fewer distros for PowerPC, but on the other hand, since apple's pretty much abandoned them, going with linux or a BSD has its own advantages
_I was not aware it's "easy" to run other OS's on Apple hardware. Knowing Apple, I'm still not sure I believe it._
I run linux (arch now, but once ubuntu) on a MacBook5,4. It's somewhat harder to get set up if you want to dual boot (the dual partitioning scheme is finicky, and although linux doesn't need it, OS X forces you to use it if there's a foreign OS, ugh..). See http://wiki.freebsd.org/AppleMacbook and https://help.ubuntu.com/community/MacBook .
Only one proprietary driver (for wireless) is needed - the rest Just Works. (And flash plays smoother than on OS X :P )
One thing that could be easier is booting MacBook Pro from USB stick. A Unetbootin that works on Mac, and builds usb sticks bootable on Macs, would be good.
My preference is to use only the bootloader for the OS I'm booting, as opposed to a "boot manager" that presents a menu of OS's. I avoid using chaining techniques the way something like GRUB does. And I don't use MBR's if I don't have to. Disklabels alone will suffice. So the stick contains only an OS-specific bootloader and the OS, usually just a kernel with embedded ramdisk or a loadable kernel module containing a filesystem and userland. I keep it very simple.
This works well for me with PC's. Would this work with today's x86 Macs?
I installed Windows as dual boot on a friends Mac.
It's far easier than installing it on a windows PC.
Put in the windows disk and follow the Apple instructions - it includes all the drivers and even gave advice about sharing files
.
Joe here - I am unaware that USB/SD sticks are fast enough to run as my system drive. Ok, cool. I haven't used a stick in years (ever since this new thing called the Internet showed up).
Yes, agree with Mr Binary - in a perfect world, computer hardware and software are completely divorced. The hotel wouldn't worry about viruses, because there would be no storage on their computer at all - I provide that with my stick. All I need is a 'kiosk computer', hardware that I can repurpose.
If what you're using works for you there's little reason to switch. I have had certain consraints in how I use computers that motivated me to explore different alternatives that "do more with less". But as users demand smaller devices and more portability I'm beginning to wonder if these "unconventional" alternatives are not generally useful for others besides me.
Not only do I divorce the OS from the hardware, but I separate the OS from my personal data. Unless I have both wholly residing in RAM (which I find generally faster than HDD or USB/SD), they are not on the same media. As much as possible I try to make the OS read-only and the media used for data (e.g. RAM for short-term data storage or HDD for long-term data storage) read-write. Perhaps there are parallels to certain object file formats and their separation of code and data.
I've also thought a lot about and experimented with the use of separate, simplified, fast-booting OS's for different purposes. I am forced to use different OS's to perform certain tasks. But the popular all-purpose OS's are huge and often I only need a small fraction of their functionality.
What if we did not think in all or nothing terms about OS functionality? What if some OS's were small and only limited-purpose?
Rebooting is the slow step. But these limited-purpose OS's, being small and simple, can be very robust and responsive once booted.
The smaller the kernel and userland, the more RAM I have for storage.
What it gets you is the ability to give it to a household with a TV, and they can have a computer to play with cheaply.
It's not meant for us, it's meant for people with no computer, or maybe one computer they aren't allowed to (or don't dare to) fiddle with in case they break it. It's for educational use.
Well, I think the idea is that you will plug it into a TV that you probably already have in the house. Thus both the HDMI port and composite TV-out. Keyboard and mouse still will be required though.
Rather like some of the early, low-cost home computers which came out of Cambridge in the 1980s; see ZX Spectrum. It almost feels like history coming back on itself.
Au contraire, it's perfect for kids. First off (as a parent of two toddlers) it's CHEAP; if a kid manages to destroy it (the little entropy generators kids are), so what - get another. Next, it starts so simple (and CHEAP) that you'll have no qualms about running simple cheap software on it. Then, you'll think nothing of putting money into a computer for any/all TV in the house (I'm waffling about a Mac Mini, but a RasberryPi is a no-brainer). So there's an easy, cheap, nigh-unto-disposable (we're talking kids here) computer that can run powerful stuff yet you'll have no qualms about devoting to a young mind.
BTW: I started programming in 4th grade, with printing terminals and acoustic-coupled dial-up mainframes. A few years later I had a then-equivalent of the RasberryPi: Dad soldered together a $100 Sinclair ZX-80 (likewise, the screen was your TV), and I programmed that thing darn near non-stop. Heck yeah the RasberryPi is "gonna be much use to a primary school student". Can't wait to get my kids going on it.
Sounds idyllic; but what are you going to run? The latest Office suite? Photoshop?
Whatever you answer, the cost dwarfs the device, and swamps its processing power.
So, we're talking about teaching kids to use some special kid-computer with software/tools they will never encounter ever again. What tools are these? A BASIC interpreter?
Why don't you go look up Linux, that way I don't have to list tens of thousands of applications, including dozens of language implementations. If you want someone to learn software development, or computing in general, Linux is a far better starting point than Windows or OS X.
Yeah, yeah. Lots of open-source half-baked projects with no support and incomplete features. Advertised as "photoshop replacement" or "complete IDE" but turns out, missing critical features, crash right and left, won't work on modern hardware etc etc.
Been down that road, burned more times than I can count. I wish it were true, but I don't fool myself into believing it.
You do realize that with the exception of one or two of the IDEs, the "mac dev tools" run or originated on Linux or other similarly now free systems, right?
In 6th grade I learned programming on a $100 computer with 1/1,000th the speed of this and 1/1,000,000th the memory and, yes, a BASIC interpreter.
Load DSL (Damn Small Linux) on this, get gcc set up with a decent minimal text editor, and yes I'll start them with C++ from the ground up - akin to how I started, but with industry standard tools from square 1 and more introduced as they progress.
I bought my first PC when I was eight (at a very discounted price from my parents). I already knew how to do trivial graphics programs in Logo. I had a "100 in one" electronics kit as well, but didn't have anyone to teach me how to read schematics. So, I got the example circuits running by following the numbered wiring sequences printed below the schematic. I learned how to use DOS before that by watching the output of my parents' first computer as it ran through AUTOEXEC.BAT (I was insanely excited when I typed "prompt SpSg" into a command line and the computer actually did something -- later on I figured out that the S should have been a $). By the time I was 13 I was buying scrap computer parts, assembling working computers, and selling them. Don't underestimate the potential of children.
You seem to have fixated on the portability. Understandable, but I don't think that is the point. The point is to build a cheap but capable device that kids can hack on. The portability is a byproduct of that focus on cost: the fewer parts, the smaller the PCB, the cheaper it can be.
Keyboards and mice are pretty cheap these days. Monitors, not as much, but this can work with TVs, which are surprisingly common, even in the developing world.
If an aspect of this is hackability, I'm wondering if a schematic or a datasheet for the CPU will be released. Under "schematic", in the wiki right now I see a pcb "gerber" layout, and Broadcom isn't one of the more open companies with datasheets. But, I figure they're in a pre-shipping crunch right now, so we'll all wait and see...
This thing does 1080p through hdmi and has a hardware video decoder. An xmbc port is being worked on for it, i plan on getting a few of these and using them as xbmc server's on my hdtv's. Can't beat it for $25/$35 + a few bucks for a 4 gig memory card
My thoughts, exactly. That's what I have in mind as my first project with these babies. It's the easiest thing to get right the first time around (as it's mostly playing around with software).
But there's a tiny[1] hint of regret that modern programming is necessarily abstracted away from the hardware. When people were using the popular home computers of the 1980s they would be able to use some version of BASIC, and then also some machine code if needed. And that wasn't something just limited to the brightest smartest young programmers; it was above average but nothing extreme. I learnt Z80 ASM on a suitcase instructional computer, turning fans on and off or stepping through traffic light sequences on LEDs. Then I made a resistor ladder DtoA convertor and pushed data out the parallel printer port for audio. (Something also done by No$GB emulator.)
That seems to be something that children will be missing out on, unless this project gives them enough boost to then investigate things like arduino or similar.
[1] Not a criticism in any way of the Raspberry Pi project.