The 'Raspberry Pi' thread

[HyRax1]

David Braben (of Elite fame!) has created a pocket-sized PC the size of a USB stick that can do 1080p output, has OpenGL ES 2.0 support in its GPU and runs Ubuntu as OS of choice.

Dubbed the Raspberry Pi, It's an ARM-based machine with 128MB RAM, has a single HDMI port out on one end and a single USB 2.0 port in on the other end to connect peripherals like keyboard and mouse.

It's priced to sell for USD$25 each (!) and they are aiming for general distribution in about a year's time (unit shown in the video is only a prototype).



News article here.

[JoJoker]

That is the coolest thing I have ever seen. Bravo.

[Menthu_Rae]

That thing is pretty freaking awesome. Sometimes I regret buying my 30" Dell... only because none of the tiny computers I've found can output 2560x1600

We could probably use these at my work.

[SLATYE]

Specifications are available here: http://www.raspberrypi.org/

# 700MHz ARM11
# 128MB of SDRAM
# OpenGL ES 2.0
# 1080p30 H.264 high-profile decode
# Composite and HDMI video output
# USB 2.0
# SD/MMC/SDIO memory card slot
# General-purpose I/O

Interesting. I'm not sure what the USB port is for - in the pictures it's a USB A plug (as you'd find on a device, not a host). Either they've got an adapter cable or it'll need a special hub to provide data + power.

It does look pretty cool, but I'm concerned about the "general availability in a year's time" thing. The ARM11 core is already considered 'slow' and has been replaced by the ARM Cortex A8 (which has in turn been replaced by the A9, and there's the A15 coming out too). 128MB of RAM is also not much for any modern Linux distribution, especially if it's trying to run off a SD card.

Shift forward a year and even the cheap Chinese tablet PCs will probably be running ARM Cortex A9 multi-core CPUs with 256MB/512MB of RAM. This one is obviously cheaper than those, but it doesn't have a monitor included.

[issh]

This could possibility of the future

Projector Monitor

[HyRax1]

Originally Posted by SLATYE

It does look pretty cool, but I'm concerned about the "general availability in a year's time" thing. The ARM11 core is already considered 'slow' and has been replaced by the ARM Cortex A8 (which has in turn been replaced by the A9, and there's the A15 coming out too). 128MB of RAM is also not much for any modern Linux distribution, especially if it's trying to run off a SD card.

It's a prototype. Use cheap and old parts to keep costs down during development and when ready to commit to something more "final", switch to current hardware?

Originally Posted by SLATYE

Shift forward a year and even the cheap Chinese tablet PCs will probably be running ARM Cortex A9 multi-core CPUs with 256MB/512MB of RAM. This one is obviously cheaper than those, but it doesn't have a monitor included.

Good point, but perhaps the purpose of the device is to be a base for larger constructions - ie: a third-party manufactures a screen and a pretty case to go with this device much like NVidia produce GPU chipsets but never make their own gfx cards.

[nanolimit]

Hell for that price I would buy one just to play around with it for a few hours.

And considering the size it would make a pretty awesome car-pc.

[digizone]

So now I know why it's taking so long for Elite 4 to be released !!!!

If this device comes with the original elite game packaged with it I would buy it just for that.

[SLATYE]

Originally Posted by HyRax1

It's a prototype. Use cheap and old parts to keep costs down during development and when ready to commit to something more "final", switch to current hardware?

Maybe, but I'd be surprised if they were able to move to current hardware at that price point. TI's OMAP3530 (Cortex A8, PowerVR GPU) is an 'old' product now (having been largely replaced by the OMAP4xxx series), but the chip alone still sells for over $40 US. Adding in discounts for huge orders might push that down a bit, but not enough to leave much space for other components.

Essentially, at the price point they're aiming for, they're basically restricted to "previous generation" hardware. It might be something faster than the ARM11 in the final version, but it's not going to be anywhere near as fast as the current chips.

Originally Posted by HyRax1

Good point, but perhaps the purpose of the device is to be a base for larger constructions - ie: a third-party manufactures a screen and a pretty case to go with this device much like NVidia produce GPU chipsets but never make their own gfx cards.

Possibly. However, I'd be very surprised if they can get the price of the Raspberry Pi + screen + case below the cost of a cheap tablet that includes all of those things.

[Paronga]

the USB seems pretty obvious to me.

you plug it into a usb power adapter for....power!

sort of like plugging in your ipod/andriod/iphone/ipad/ anything usb powered.

carpc indeed.

[elvis]

Originally Posted by SLATYE

Interesting. I'm not sure what the USB port is for

Power from a workstation. This system is intended to teach real development again, not the "life skills" nonsense that modern computer classes have turned in to. It's cheap enough for schools and students (parents) to afford, and is mobile enough to take home and practice on.

Originally Posted by SLATYE

It does look pretty cool, but I'm concerned about the "general availability in a year's time" thing. The ARM11 core is already considered 'slow' and has been replaced by the ARM Cortex A8 (which has in turn been replaced by the A9, and there's the A15 coming out too). 128MB of RAM is also not much for any modern Linux distribution, especially if it's trying to run off a SD card.

Shift forward a year and even the cheap Chinese tablet PCs will probably be running ARM Cortex A9 multi-core CPUs with 256MB/512MB of RAM. This one is obviously cheaper than those, but it doesn't have a monitor included.

It's not designed as a workstation replacement. It's designed as an educational tool.

Dave Braben, like most old school programmers and computer scientists (myself included), is utterly disgusted with what is taught in highschool ICT classes these days. I've seen "IT" classes that "teach" Microsoft Word, FFS.

ARM is a fairly ubiquitous platform, for learning high level languages, and is a fantastic platform to cut your teeth learning machine level code. Dave Braben himself would have probably learned on similar architecture (albeit many generations older) back in the old BBC Micro built by Acorn (ARM stands for Acorn RISC Machine, for those that don't know).

Moore's law states transistor density doubles every 18 months. May's Law counteracts this by saying software efficiencies halve every 18 months.

Speaking as person who gets paid to attempt to make big applications run in a reliable fashion, modern software is anything but reliable. Every year I see enterprises spend millions on bigger, faster and more dense computing solutions, and all I see is the end user experience get slower and slower.

Those of us frustrated by this have long desired a "return to roots" education system that prevents a system we see today - a system where people are entering the work force as programmers having cut their teeth on horrid, bloated systems like Java and .NET, without any clue what came before that, and the lost art of optimisation.

Dave Braben's invention is designed to be "slow". This is an intentional step to get people back into the mind set of trying to maximise what they can from hardware, rather than throwing more dollars at the problem.

Business people will argue that buying hardware is ultimately cheaper and easier than optimising code. That might be true for bad code that's already written, and already in the market. I posit to these people that, were programming education standards higher, and people were writing lean code "by default" with the same time and resources, then when that code finally goes to production it'll run a hell of a lot faster.

Good education is the key here, and that won't happen as long as schools are teaching Java and .NET on workstations with multi gigahertz processors and multiple gigabytes of RAM.

[stmok]

Originally Posted by elvis

Dave Braben, like most old school programmers and computer scientists (myself included), is utterly disgusted with what is taught in highschool ICT classes these days. I've seen "IT" classes that "teach" Microsoft Word, FFS.

Agreed.

When I was High School in the mid- to late-1990s we cut our teeth on various solutions in order to get a feel of what word processing, spreadsheet, etc was about. The point was to teach fundamental or universal concepts.

Today? I went back to my old High School to see what has improved over time. Suffice to say, it wasn't for the better. They don't teach fundamental concepts. They teach future consumers for a particular brand.

Originally Posted by elvis

Speaking as person who gets paid to attempt to make big applications run in a reliable fashion, modern software is anything but reliable. Every year I see enterprises spend millions on bigger, faster and more dense computing solutions, and all I see is the end user experience get slower and slower.

Because the folks "upstairs" believe in the flawed notion that automation and new technologies will improve production/quality without them needing to put effort in. ie: Sit back and let the money come in...They don't understand that successful, long term management requires constant work and active effort to improve the processes of development, production, etc. (Its universal and not specifically tied to the software industry.)

Originally Posted by elvis

Those of us frustrated by this have long desired a "return to roots" education system that prevents a system we see today - a system where people are entering the work force as programmers having cut their teeth on horrid, bloated systems like Java and .NET, without any clue what came before that, and the lost art of optimisation.

The more automation there is to lower the bar, the more people will need to re-discover the lost art of programming...The old-school engineering way of creating robust and lean solutions. (When you read past stories of IBM engineers developing air traffic control solutions and software for the space programs that are often re-used today; you realise the current trend in software quality is an embarrassment to humanity itself.)

Originally Posted by elvis

Dave Braben's invention is designed to be "slow". This is an intentional step to get people back into the mind set of trying to maximise what they can from hardware, rather than throwing more dollars at the problem.

Its intentionally trying to force students to code in a restricted/limited environment, in order to encourage better coding practices.

Originally Posted by elvis

Business people will argue that buying hardware is ultimately cheaper and easier than optimising code. That might be true for bad code that's already written, and already in the market. I posit to these people that, were programming education standards higher, and people were writing lean code "by default" with the same time and resources, then when that code finally goes to production it'll run a hell of a lot faster.

When I ran into this situation, I was outright blunt about it. (They got sick of all that marketing smoke blown into their faces.)
=> I'm not going to BS you, so I'm going to put it to you straight. You either put the effort in now to do it properly, such that running costs will be low. Or you can do a half-arsed rush-job that will cost you more in keeping the bloody thing up in the long run!

Originally Posted by elvis

Good education is the key here, and that won't happen as long as schools are teaching Java and .NET on workstations with multi gigahertz processors and multiple gigabytes of RAM.

Its not just the engineering level, education also needs to be at the business level. There are certain key principles from the WWII-era that nearly all management have never heard of or have completely forgotten over time. (I highly doubt its taught in business schools of today. As they are short-term oriented in thinking.)

The current trend in commercial software development is to push crap out as fast as possible to get into the market. Then fix the bugs after the customer has paid for the solution. ...We see it in the never ending supply of system updates on PCs and game consoles, downloadable content in games, etc. (You need patches in games?!)

People are being pushed into believing that getting into the market first is more important than releasing a quality solution to the customer.

The key issue goes all the way back to short-term VS long-term thinking...
Short-term => Easy/fast now. Hurts you later.
Long-term => Hurts/slow now. Easy later.

Each approach has incalculable consequences. ie: The former will blow back in your face, when your customers find or get info about someone else doing a better job. The latter will earn a reputation that keeps customers and bring you more, as word spreads.

[Anomalous]

I'll take 5!

[ThunderbirdMoz]

This shouldnt just be a educational tool, its the right size and price for smart programmable appliances to be everywhere around the house.

[BlueRaven]

Found this via slashdot, probably already seen by many but apparently not posted here yet:

DEMO - Raspberry Pi running Quake 3 @ 1080p with 4x AA

Credit-card sized Full-HD media player + casual gaming machine anyone?