The $25 Raspberry Pi will do for computing what Arduino has done for robotics
Hi, this is Steven Cherry for IEEE Spectrum’s “Techwise Conversations.”
This is an exciting time for do-it-yourselfers.
Last October, Spectrum ran a feature article about Arduino. Here’s what we said.
Arduino has spawned an international do-it-yourself revolution in electronics. You can buy an Arduino board for just about US $30 or build your own from scratch: All hardware schematics and source code are available for free under public licenses. As a result, Arduino has become the most influential open-source hardware movement of its time.
Lightning is about to strike twice in the open-source hardware movement, with the release of Raspberry Pi. For $25, you can get the guts of an entire computer—a single board with a microprocessor,
Wi-Fi, [Wi-Fi can be added through the USB port] and inputs for video, USB, and an SD card. Interest has been intense—the initial launch on February 29th briefly crashed the Raspberry Pi website, and people have been signing up as quickly as once every 6 seconds.
My guest today is the founder of Raspberry Pi, Eben Upton. He’s a chip architect at Broadcom, which also makes the chip that’s at the heart of the board. He joins us by phone from Cambridge, England.
Eben, welcome to the podcast.
Eben Upton: Hi there.
Steven Cherry: So, tell us about the Raspberry Pi and why does the world need it when we already have Arduino?
Eben Upton: OK. So the Raspberry Pi is this little credit-card-size computer; it’s designed for, as you say, low cost and a reasonable amount of performance, and it’s really just a Linux PC. So it’s a Linux PC with an ARM processor instead of the x86 processor that you’d find inside most Linux desktops. We think the reason the world needs this is primarily as a platform for children to learn to program on. We’ve experienced a steady decline in both the number and the range of skills of children programming computers, particularly people applying to study computer science and higher education. And this is really intended to fill that gap—you know, provide a platform, just as I in the 1980s had an 8-bit microcomputer. It’s designed to fill that gap.
Steven Cherry: Yeah. So, I gather from your own work teaching young computer science students, you’ve discovered that in many cases all they can do is program Web pages.
Eben Upton: Absolutely. So, I think we’ve had a reduction from, say, if you think about 1995, which was when I went to college, you could typically rely on an undergraduate having done a substantial amount of real programming, often quite a deep level of technical work on one or more platforms. Many of us could program in one or more assembly languages. And yeah, within 10 years of that point, we were getting to a point where your average applicant was maybe somebody who’d done, as you say, a little bit of Web design, maybe a little bit of Web programming—you know, we saw quite a bit of people who‘d maybe done some PHP but not that kind of deep technical understanding of how machines work.
Steven Cherry: Let’s talk about what you get for 25 bucks, which strikes me as insane: This ARM11 processor is apparently fast enough to play a Blu-ray–quality movie?
Eben Upton: So, what we have is an SoC [system-on-a-chip], a Broadcom SoC, which contains a number of hardware accelerators. So things like decoding high-definition video or doing very high performance 3-D graphics, we actually have dedicated hardware accelerator blocks to help the ARM do that. So the ARM is really is mostly running control-side code.
Steven Cherry: It has an RCA port and HDMI?
Eben Upton: Yeah. So we thought it was important to be able to drive older televisions. You know, one of our visions for this is that it will end up in the bedrooms of children just like, say, my BBC microcomputer did when I was a child. And there we felt although there are going to be a lot of people like me who aren’t particularly price sensitive and have modern flat-panel televisions or monitors, so we need to have an HDMI connection to support them. So we are expecting maybe a lot of younger users to have scrounged up a television from somewhere, an old standard-definition television, so the RCA jack and the analog audio output are important for supporting those people.
Steven Cherry: Eben, I know you can’t disclose the bill of materials, but I gather it’s fairly close to the actual selling price. And I’m told you wanted to do the manufacturing in the U.K., but that didn’t work out, and you ended up in China and Taiwan.
Eben Upton: Yeah, indeed. I mean, this is mostly a supply chain issue. This is a board which largely can be assembled robotically, so there isn’t actually that much of a labor cost issue, but if you look at the global supply chain, you know all of your precursor components are in the Far East, and therefore it’s extremely hard to justify, particularly if you’re then going to re-export the device to, say, some of the BRIC countries [Brazil, Russia, India, and China] or if you’re going to re-export the device to India or to China, it’s very hard to justify bringing those components over to the United Kingdom or over to the United States, doing the assembly here, and then shipping the finished goods to their final destination.
Steven Cherry: Now, the Raspberry Pi is not quite as open as Arduino, right? There’s no data sheet for the Broadcom chip, for example.
Eben Upton: Well, we have actually managed to obtain an edited data sheet, so there are always going to be elements on the chip which are going to remain proprietary. Some of those accelerators that I mentioned are always going to remain proprietary, but the level of where the UART is, or where the timers are, or how you drive the intro controller...those things which are the things you really need if you want to port a new operating system, I mean we have people talking—although we use Linux on it—we have people talking about porting some BSD variants and some more exotic operating systems. We have actually managed to obtain a data sheet, which has enough information to do that piece of work, at least.
Steven Cherry: And I guess that’s not unprecedented. For example, the Android phone is based on an open-source design, but the actual 3-D chip is a bit of a black box to developers.
Eben Upton: Yeah. I think very few if any 3-D core or IP core vendors or SoC vendors who make their own 3-D cores—and Broadcom is one—very few organizations actually publish significant amounts of deep technical data about how their 3-D accelerators work.
Steven Cherry: So, the Raspberry Pi is essentially the system board of a computer. I guess if you wanted to actually use it as, like, a tiny laptop to run a browser or Google Docs or something, you’d still need a bunch of stuff, right?
Eben Upton: Yeah. So you’d need some sort of—I think our guiding philosophy here has been we want to piggyback on top of things that people already have, we really don’t want people to have to have any more expenditure beyond just buying the base unit. So we use a USB mouse and keyboard. Now, those are—there are many of them in the world. Most people can scrounge one up; even if you can’t scrounge one up, they really are only a very few dollars. We use obviously HDMI or RCA to drive a television, so there’s an assumption there that somebody has—I think the assumption that in the developed world somebody can obtain, certainly, a standard-definition television at virtually zero cost is a good assumption. Secondhand standard-definition televisions are effectively worthless. You know, they have negative value, you have to pay someone to take them away, so we’re imagining kids going and getting one of these down out of the attic, maybe going and finding a relative who’s got one, and taking that. So, we use the television for display output, we use an SD card for our mass storage, and there are a lot of low-capacity SD cards out there in the world, right? And then for power we use a mobile phone charger. So, we use a Micro-USB mobile phone charger, which certainly in Europe now is kind of a mandated standard, that all mobile phones have to charge with this format of charger. We expect that that will become the case elsewhere in the world as well.
Steven Cherry: I guess one of the great charms of a project like this is that people are going to do things that you couldn’t possibly anticipate, but you’re already hearing some interesting things that people are doing with them.
Eben Upton: Yeah, absolutely. There’s enormous stuff just in the—you put in the class of “control.” So people that are wanting to use these to control a thing, whether it’s a home automation, whether it’s something in their car...we have a number of people that are into the kind of hot-rodding scene; I’ve talked about using these to control—as sort of replacement engine management computers. People have talked about using them in various aerospace applications: so, balloons, sounding rockets, at least one group of people talking about nanosat. There’s quite an active academic nanosat community, and so these things actually in orbit would be fun. And then there’s a whole range of people who have these kind of media-like applications for them. It has very good media performance, so some people are just looking to have them as a home media center. People want to run Xbox media center on it, maybe, plug it into their television, and then it’ll just sit there serving movies to them.
Steven Cherry: Now, you first launched in the very end of February, and I guess right away demand sort of outstripped supply. Have you caught up yet?
Eben Upton: No, we are a long way from catching up because demand just keeps going up. So the orders that went in in February were all preorders. We are actually—just now it’s the 10th of April—speaking today, we expect the very first devices—no devices have shipped to date—so we expect the very first devices to go out of the door, probably, the start of next week. So, I guess that’s the week commencing the 16th of April, so the very first devices will come out there, and then most of the effort over the last month and a bit since launch has gone to managing the ramp. You know, I think where we were first thinking about this, we were thinking about building 1000 or 10 000 of these. Once you’re thinking about building 100 000 units, then it’s a whole different environment. The class of company you want to deal with if you’re manufacturing is different, and some of the issues which you were prepared to put up with when you were only making a few thousand, in terms of how smooth the factory flow is, particularly the program and test flow, you know, it really pays to optimize that stuff once you go into high volume. So there’s been a lot of effort going into that, and in parallel with that we’ve had a qualification process going on. We’ve been applying for FCC and CE qualification for these devices, which we hadn’t originally intended to do initially. But once it became apparent, both the kind of volume and the demographic, that these things weren’t just going to go to developers, they were going to go to everyone—you know, we have children buying on day one, which we hadn’t really initially imagined—it became apparent we were going to need to get some certification. So we’ve been doing EM—both emissions and immunity—static ESD sensitivity, tests like that, in the lab.
Steven Cherry: It sounds like a lot of work. You mentioned that there is a small profit margin on the pricing, but you set up the whole Raspberry Pi thing as a charity based in the U.K. Don’t you like making money, Eben?
Eben Upton: I very much like making money. One of the nice things is that this is an environment for not making money. I’ve been involved in a number of start-ups, and it’s just nice to have something which feels like a business but where we can focus on just doing the right thing, I guess. My experience of being in business is that you have to worry about two things: You have to worry about creating value, and you have to worry about capturing value, right? If you don’t create value, no one’s going to give you money for it, but you also need to worry about once you’ve created that value, making sure it ends up in your pocket and not somebody else’s. And often you’ll do something which is, I guess, suboptimal from the point of view of value creation in order to make sure that you can capture it.
Steven Cherry: Now, you put this all together while still holding down a day job, which, as I mentioned, you’re currently the technical director at Broadcom, which is the company that makes the chip at the heart of the Raspberry Pi. I guess there are some obvious synergies between the two roles, but have there been any potential conflicts you’ve had to skirt around?
Eben Upton: I guess I mean there are obviously always issues when you’re a supplier and a customer. I guess it’s mostly about I know an enormous amount how this chip works. I was involved in the team that designed the 3-D graphics processor for this chip, so I have in my head this vast amount of information about the chip, and kind of trying to select an appropriate amount of disclosure has obviously needed a bit of finesse, right? If I told people everything I know about this chip, then I would be divulging important stuff which is valuable to Broadcom. So just trying to strike that balance has been...yeah, I’ve needed to think about it.
Steven Cherry: Now, you started this whole effort about six years ago, I guess, and you were still affiliated with a lab at Cambridge University at the time. Is Cambridge becoming something of an incubator of technology in a way that maybe Stanford, at the heart of Silicon Valley, is?
Eben Upton: Right. So we have this thing we call Silicon Fen—I mean, we’re out in the Fenlands here—and we call this place Silicon Fen. Historically, it’s been an enormously important incubator of high tech in the U.K. It’s been one of the small number of, kind of, clusters, sort of like the Stanford cluster or the Boston cluster in North America. So we’ve had a cluster here, there’s a cluster in the southwest of England in Bristol, and there’s been one up in Scotland for a long time, sometimes called Silicon Glen. And one of the issues, I guess, for us is this cluster of high tech around Cambridge, ’cause obviously ARM are here, right? If you want to see the big poster children, ARM is probably the biggest and most successful, certainly the biggest and most successful Cambridge tech stock start-up in a long time. So all of these companies are dependent on a flow of talent, they’re all dependent on a flow of graduate talent, and that’s the real threat here right? That’s the threat which we’re set up to try and mitigate. This idea that if you don’t have kids programming you don’t get students who can program, if you don’t have students who can program you don’t get graduate students who can program, and then that’s the sort of thing that starts to chip away at the underpinnings of these clusters.
Steven Cherry: You really chose a very kid-friendly name for the enterprise. I should mention that it’s Raspberry, like the fruit, and Pi, p-i, as in the transcendental number. What went into that choice?
Eben Upton: It was designed by committee. So the Pi is actually the Py from Python misspelled. When I was first looking at doing this, I was looking at a computer that really was much more limited than the final Raspberry Pi, and really all it could do was run Python, so it couldn’t run Linux. So the current device—you want to run Python on a Linux PC, you boot it up, you type Python, and it runs Python. The original device really could only run Python, so that’s where the Pi element comes from, and the Raspberry thing is fruit-named computer companies, right? There’s been a lot of computer companies named after fruit; there aren’t that many left…names, available names. We had in the U.K. a company called Apricot for a long time, we had a company called Tangerine; in fact, Acorn, which was the company that eventually became ARM, that Acorn was technically a fruit. So there have been a number of fruit-named computer companies, and this is just another one in that line.
Steven Cherry: Very good. Well, you launched on February 29th. I think you missed a huge bet by not waiting two weeks and launching on March 13th, which would have been pi day, but other than that, I think you guys have done pretty much everything right. So I’d like to congratulate you and thank you very much.
Eben Upton: Awesome. Thanks very much for your time.
Steven Cherry: We’ve been speaking with Eben Upton, founder of the Raspberry Pi $25 computer. For IEEE Spectrum’s “Techwise Conversations,” I’m Steven Cherry.
Announcer: “Techwise Conversations” is sponsored by National Instruments.
NOTE: Transcripts are created for the convenience of our readers and listeners and may not perfectly match their associated interviews and narratives. The authoritative record of IEEE Spectrum’s audio programming is the audio version.
A correction to this article was made on 24 April 2012.