Raspberry Pi interview: Eben Upton reveals all (Linux User)

Farnell were talking about 700 hits a second on release day.

Mean while Ubuntu are saying ARMv6 is too hard and busy patting themselves on the back for Unity.

What do I know, maybe this is the right move, maybe Ubuntu will take the world by storm with their proprietary desktop for Android.

Also, Broadcom are subsidising the price of the RPi. I think TI also subsidises the Beagle/Panda boards, so maybe there's just a difference in how much each manufacturer subsidises their board(s).

As for performance, I want to see the benchmark results. The multi-issue super-scalar Cortex-A8 and A9 cores are reported to be able to perform 2.0 and 2.5 DMIPS/MHz, respectively. The aging ARM11 core only does about 1.0 DMIPS/MHz. So the claim that a 700 MHz ARM11 (~700 DMIPS) is only 20% slower than a 600 MHz Cortex-A8 (~1200 DMIPS) just doesn't "add up". Maybe something like a faster memory bus is helping it perform better on a certain benchmark.

http://beagleboard.org/hardware/design

Looking at the beagleboard Bill of materials at a beagleboard is constructed from 67 different components totalling 230 pieces (lots of caps!). That someone needs source, assemble and *test* before shipping out to customers. So even if the cost of material would be $25, you still need to count that not every board manufactured will work, mail will loose products and people will RMA boards that slipped your testing - all adding up to the cost.

Bits from BB manufacturing:

http://jkridner.s3.amazonaws.com/esc/BEAGLE_ESC_4.ppt

Will remain interested to hear the BoM of Rasberry and how they get manufacturing done without pushing the price up.

Firstly, I commend Eben Upton and his co-founder for initiating the RaspberryPi project. Ultimately, it will result in greater educational opportunities for many people who are interested in Linux, in embedded development, and who have a great idea for an interesting project but little resources to get started. A couple of comments though:

1) Price. Sure the BeagleBoard is approximately $100; but the parts list, schematics, etc. have always been available allowing anyone with the know how to duplicate it at a cheaper price if they wanted to. I would argue that RaspberryPi owes a huge, huge thanks to the BeagleBoard project. I doubt there would be RPi without the Beagle, as the Beagle and the community that has grown around it has shown that low cost dev boards are very useful. Saying that "somebody in that BeagleBoard value chain has got to be making a pile of money..." is not very responsible though. Statements as such should be backed up with proof. This is not a competition, we're all in this together.

2) Community. The single most important success factor for the RaspberryPi's success will be the community that grows around it. Having watched the elinux.org stats, I can say that the community is off to a great start, as the number of RPi pages has spiked tremendously over the last couple of months. But the interesting thing to me that fosters true success, is when a core set of members of the community with highly advanced knowledge of the board and Linux embraces the technology. The most evident example of this is the BeagleBoard/BeagleBone community. A community needs these people to figure out the really hard problems, be able to communicate the solutions adequately, and to stick around for the long haul to explain things over, and over, and over to every new user that arrives on the scene. I can tell you that in #beagle, I've seen very dedicated community members help people get their SDCards formatted thousands of times. The thing about these hackers who carry the community forward is that (in my experience) they have a passion for freedom. So I'll be interested to watch (and help) the RPi community grow and see where it ends up.

Now if only my board would arrive ;)

That said, price is interesting, but the RPi is within the same order of magnitude as the BeagleBone. Plus, the RPi doesn't have a serial connector. And even if it did, you'd still need a USB-to-serial connector and, possibly, an RS232 cable. The Bone has this **AND** a JTAG debugger integrated straight out of the box. The only thing you need is the USB cable that comes with the Bone.

Add up the price of purchasing those pieces and/or assembling them to the RPi and the Bone suddenly won't seem too bad.

Personally I'll get an RPi to get a good hands-on with it, but the Beagles are likely what I'll continue recommending for the type of work I do. If nothing else because of the community that's been built around those.

In this case, what got me was the closed MPEG decoder block. It has a driver which downloads a binary blob (actually a DSP executable) for the format that you want to decode, and then shuffles data back and forth in shared memory. But when you get decoded video frames out, they don't include the corresponding timestamp from the input MPEG stream. So the only way to do A/V sync is a hack. If the DSP code were open source, then I could add probably just a few dozen lines to it to propagate the timestamps and everything would be great.

My point is that wanting a free graphics system is not only a moral, philosophical argument, but also a practical issue that gets in the way of making good use of these chips.

If anything, the Raspberry Pi is less open that the alternatives since the closed GPU is "on top". Its proponents also have less excuse for the closed graphics system, since in their case it is their own IP; in most of the other cases the vendor has licensed it from a third party (e.g. TI from PowerVR).

Will someone wake me up when something changes?

He answers a big part of the question himself. If you employ 10 people here it costs you 10 * 100,000 EUR / year. Engineers a bit more, people running your office a bit less, but don't let's make it too scientific now. That's 1 million EUR / year. If you sell 10,000 boards (because they cost 150 EUR many people will hesitate to actually buy one), that's 100 EUR of salaries per board. After that you start to discuss the BOM.

If you are a charity and work (mainly) with unpaid/sponsored resources who have earned their living and a bit more elsewhere the story is a whole lot different.

So one "pile of money" goes to ordinary engineers like me and many other readers here who enjoy the salaries paid in this industry in industrial countries/western world whatever you'd like to call it. (Of course there are some capitalists that make even a lot more money than we do, but I doubt from selling Pandaboards.) Western engineering salaries are not an issue if you sell 10s of millions of devices. But for everything smaller scale they are the killer.

Disclaimer: I have absolute no idea how many boards are realistically sold. But I have the strong feeling the GTA04 people could really be happy if they could plan with 10,000 pieces.