Читать книгу Raspberry Pi User Guide - Eben Upton - Страница 4

I started work on a tiny, affordable, bare-bones computer in 2006, when I was a Director of Studies in Computer Science at Cambridge University. I’d received a degree at the University Computer Lab as well as studying for a PhD while teaching there, and over that period, I’d noticed a distinct decline in the skillset of the young people who were applying to read Computer Science at the Lab. From a position in the mid-1990s, when 17-year-olds wanting to read Computer Science had come to the University with a grounding in several computer languages, knew a bit about hardware hacking, and often even worked in assembly language, we gradually found ourselves in a position where, by 2005, those kids were arriving having done some HTML – with a bit of PHP and Cascading Style Sheets if you were lucky. They were still fearsomely clever kids with lots of potential, but their experience with computers was entirely different from what we’d been seeing before.

The Computer Science course at Cambridge includes about 60 weeks of lecture and seminar time over three years. If you’re using the whole first year to bring students up to speed, it’s harder to get them to a position where they can start a PhD or go into industry over the next two years. The best undergraduates – the ones who performed the best at the end of their three-year course – were the ones who weren’t just programming when they’d been told to for their weekly assignment or for a class project. They were the ones who were programming in their spare time. So the initial idea behind the Raspberry Pi was a very parochial one with a very tight (and pretty unambitious) focus: I wanted to make a tool to get the small number of applicants to this small university course a kick start. My colleagues and I imagined we’d hand out these devices to schoolkids at open days, and if they came to Cambridge for an interview a few months later, we’d ask what they’d done with the free computer we’d given them. Those who had done something interesting would be the ones that we’d be interested in having in the program. We thought maybe we’d make a few hundred of these devices, or best case, a lifetime production run of a few thousand.

Of course, once work was seriously underway on the project, it became obvious that there was a lot more we could address with a cheap little computer like this. What we started with is a long way indeed from the Raspberry Pi you see today. I began by soldering up the longest piece of breadboard you can buy at Maplin with an Atmel chip at our kitchen table, and the first crude prototypes used cheap microcontroller chips to drive a standard-definition TV set directly. With only 512 K of RAM, and a few MIPS of processing power, these prototypes were very similar in performance to the original 8-bit microcomputers. It was hard to imagine these machines capturing the imaginations of kids used to modern games consoles and iPads.

There had been discussions at the University Computer Lab about the general state of computer education, and when I left the Lab for a non-academic job in the industry, I noticed that I was seeing the same issues in young job applicants as I’d been seeing at the University. So I got together with my colleagues Dr Rob Mullins and Professor Alan Mycroft (two colleagues from the Computer Lab), Jack Lang (who lectures in entrepreneurship at the University), Pete Lomas (a hardware guru), and David Braben (a Cambridge games industry leading light with an invaluable address book), and over beers (and, in Jack’s case, cheese and wine), we set up the Raspberry Pi Foundation – a little charity with big ideas.

Why “Raspberry Pi”?

We get asked a lot where the name “Raspberry Pi” came from. Bits of the name came from different trustees. It’s one of the very few successful bits of design by committee I’ve seen, and to be honest, I hated it at first. (I have since come to love the name, because it works really well – but it took a bit of getting used to since I’d been calling the project the “ABC Micro” in my head for years.) It’s “Raspberry” because there’s a long tradition of fruit names in computer companies (besides the obvious, there are the old Tangerine and Apricot computers – and we like to think of the Acorn as a fruit as well). “Pi” is a mangling of “Python”, which we thought early on in development would be the only programming language available on a much less powerful platform than the Raspberry Pi we ended up with. As it happens, we still recommend Python as our favourite language for learning and development, but there is a world of other language options you can explore on the Raspberry Pi too.

In my new role as a chip architect at Broadcom, a big semiconductor company, I had access to inexpensive but high-performing hardware produced by the company with the intention of being used in what were then very high-end mobile phones – the sort with the HD video and the 14-megapixel cameras. I was amazed by the difference between the chips you could buy for $10 as a small developer, and what you could buy as a cell-phone manufacturer for roughly the same amount of money: general purpose processing, 3D graphics, video, and memory bundled into a single BGA package the size of a fingernail. These microchips consume very little power, and have big capabilities. They are especially good at multimedia, and were already being used by set-top box companies to play high-definition video. A chip like this seemed the obvious next step for the shape the Raspberry Pi was taking, so I worked on taping out a low-cost variant that had an ARM microprocessor on board and could handle the processing grunt we needed.

We felt it was important to have a way to get kids enthusiastic about using a Raspberry Pi even if they didn’t feel very enthusiastic about programming. In the 1980s, if you wanted to play a computer game, you had to boot up a box that went “bing” and fed you a command prompt. It required typing a little bit of code just to get started, and most users didn’t ever go beyond that – but some did, and got beguiled into learning how to program by that little bit of interaction. We realised that the Raspberry Pi could work as a very capable, very tiny, very cheap modern media centre, so we emphasised that capability to suck in the unwary – with the hope that they’d pick up some programming while they’re at it.

After about five years’ hard grind, we had created a very cute prototype board, about the size of a thumb drive. We included a permanent camera module on top of the board to demonstrate the sort of peripherals that can easily be added (there was no camera when we launched because it brought the price up too much, but we’ve now made a separate, cheap camera module available for photography projects), and brought it along to a number of meetings with the BBC’s R&D department. Those of us who grew up in the UK in the 1980s had learned a lot about 8-bit computing from the BBC Microcomputer and the ecosystem that had grown up around it – with BBC-produced books, magazines and TV programmes – so I’d hoped that they might be interested in developing the Raspberry Pi further. But as it turned out, something has changed since we were kids: various competition laws in the UK and the EU meant that “the Beeb” couldn’t become involved in the way we’d hoped. In a last-ditch attempt to get something organised with them, we ditched the R&D department idea and David (he of the giant address book) organised a meeting with Rory Cellan-Jones, a senior tech journalist, in May 2011. Rory didn’t hold out much hope for partnership with the BBC, but he did ask if he could take a video of the little prototype board with his phone, to put on his blog.

The next morning, Rory’s video had gone viral, and I realised that we had accidentally promised the world that we’d make everybody a $25 computer.

While Rory went off to write another blog post on exactly what it is that makes a video go viral, we went off to put our thinking caps on. That original, thumb-drive-sized prototype didn’t fit the bill: with the camera included as standard, it was way too expensive to meet the cost model we’d suggested (the $25 figure came from my statement to the BBC that the Raspberry Pi should cost around the same as a text book, and is a splendid demonstration of the fact that I had no idea how much text books cost these days), and the tiny prototype model didn’t have enough room around its periphery for all the ports we needed to make it as useable as we wanted it to be. So we spent a year working on engineering the board to lower cost as much as possible while retaining all the features we wanted (engineering cost down is a harder job than you might think), and to get the Raspberry Pi as useable as possible for people who might not be able to afford much in the way of peripherals.

We knew we wanted the Raspberry Pi to be used with TVs at home, just like the ZX Spectrum in the 1980s, saving the user the cost of a monitor. But not everybody has access to an HDMI television, so we added a composite port to make the Raspberry Pi work with an old cathode-ray television instead. Since SD cards are cheap and easy to find, we chose them as our storage medium: the original Model A and Model B used full-size cards, while more recent iterations have moved to the now more common microSD standard. And we went through several iterations of power supply, ending up with a micro USB cable. Recently, micro USB became the standard charger cable for mobile telephones across the EU (and it’s becoming the standard everywhere), which means the cables are becoming more and more ubiquitous, and in many cases, people already have them at home.

By the end of 2011, with a projected February release date, it was becoming obvious to us that things were moving faster, and demand was higher, than we were ever going to be able to cope with. The initial launch was always aimed at developers, with the educational launch planned for later in 2012. We had a small number of very dedicated volunteers, but we needed the wider Linux community to help us prepare a software stack and iron out any early-life niggles with the board before releasing into the educational market. We had enough capital in the Foundation to buy the parts for and build 10,000 Raspberry Pis over a period of a month or so, and we thought that the people in the community who would be interested in an early board would come to around that number. Fortunately and unfortunately, we’d been really successful in building a big online community around the device, and interest wasn’t limited to the UK, or to the educational market. Ten thousand was looking less and less realistic.

Our Community

The Raspberry Pi community is one of the things we’re proudest of. We started with a very bare-bones blog at www.raspberrypi.org just after Rory’s May 2011 video, and put up a forum on the same website shortly after that. That forum now has more than 170,000 members – between them they’ve contributed nearly a million posts of wit and wisdom about the Raspberry Pi. If there’s any question, no matter how abstruse, that you want to ask about the Raspberry Pi or about programming in general, someone there will have the answer (if it’s not in this book, you’ll find it in the forums).

Part of my job at Raspberry Pi involves giving talks to hacker groups, computing conferences, teachers, programming collectives, and the like, and there’s always someone in the audience who has talked to me or to my wife Liz (who runs the community) on the Raspberry Pi website – and some of these people have become good friends of ours. The Raspberry Pi website gets more than one request every single second of the day.

There are now hundreds of fan sites out there. For several years, there was a fan magazine called The MagPi, which was produced monthly by community members, with type-in listings, lots of articles, project guides, tutorials, and more. This became so successful that we brought it in-house at the Foundation, which makes it available in print or as a free download from www.raspberrypi.org/magpi. Type-in games in magazines and books provided an easy route into programming for me – my earliest programming experience with the BBC Micro was of modifying a type-in helicopter game to add enemies and pick-ups.

We blog something interesting about the device at www.raspberrypi.org at least once every day. Come and join in the conversation!

There were 100,000 people on our mailing list wanting a Raspberry Pi – and they all put an order in on day one! Not surprisingly, this brought up a few issues.

First off, there are the inevitable paper cuts you’re going to get boxing up 100,000 little computers and mailing them out – and the fact was that we had absolutely no money to hire people to do this for us. We didn’t have a warehouse – we had Jack’s garage. There was no way we could raise the money to build 100,000 units at once – we’d envisaged making them in batches of 2,000 every couple of weeks, which, with this level of interest, was going to take so long that the thing would be obsolete before we managed to fulfil all the orders. Clearly, manufacturing and distribution were something we were going to have to give up on and hand over to somebody else who already had the infrastructure and capital to do that, so we got in touch with element14 and RS Components, both UK microelectronics suppliers with worldwide businesses, and contracted with them to do the actual manufacture and distribution side of things worldwide so we could concentrate on development and the Raspberry Pi Foundation’s charitable goals.

Demand on the first day was still so large that RS and element14’s websites both crashed for most of the day – at one point in the day, element14 were getting seven orders a second, and for a couple of hours on February 29, Google showed more searches were made worldwide for “Raspberry Pi” than were made for “Lady Gaga”. We made and sold more than a million Raspberry Pis in the first year of business, making Raspberry Pi the fastest-growing computer company in the world, ever. Things aren’t slowing down: we make more than 300,000 Pis every month and have sold more than ten million in a little over four, with no hint of a slowdown. If we’d stuck with our original plans, we’d have made 100 or so of these devices for University open days, and that would have been it.

NOTE

The first production Pis were made in Chinese factories, but in 2012 we managed to repatriate all of the production to the UK. Your Raspberry Pi is now made in South Wales, in an area of the country with a proud manufacturing heritage, but few remaining factories. Amazingly, it costs us less to manufacture in Wales as it did in China, and we’re able to do that manufacture without a language or cultural barrier, and with the ability to jump in the car and be on the factory floor in a few hours if necessary.

There is nothing that affects the blood pressure quite like accidentally ending up running a large computer company!