Paw Prints: Writings of the maddog

At LinuxCon in New Orleans Intel made a big splash about CircuitCO's new board, the “minnow”, declaring it the “first open hardware board based on x86”.  This, of course, ignored the “Gizmo” board, which came out previously...

Intel then compared the “minnow” against the Raspberry Pi (RPi) declaring that the RPi was not “open” because you could not go down to your corner store and purchase the components one at a time to solder (surface mount, anyone?) those components onto your (multi-layer) printed circuit board.

With the great success of the RPi (which uses an ARM chip) I suppose Intel had to do something, just as Intel had to develop the Classmate PC with Microsoft Windows on it after the OPLC (One Laptop Per Child) was a success using AMD chips (later Via chips) and the GNU/Linux based “Sugar”.

I do actually applaud Intel creating a relatively (199 US Dollar) low-cost board that is relatively “open” (all except for the ancient GMA 600 GPU, which still needs a binary blob after all these years), but if the "minnow" is going to be compared to the Raspberry Pi, please consider the needs of the target audience for the Raspberry Pi, and ask if the comparison is justified.

The RPi was designed to be a tool for middle school and high school students to learn about computers. It was designed to fit (with its case) into a student's pocket. I have tried doing that, and the RPi really does fit, with room to spare.  They could have saved some money by leaving off the analog video out, but they wanted it to work with older monitors and TV sets the students might have at home.

The RPi was designed to sell for 35 US Dollars. If it is destroyed by a student, it is no big deal. If a school wants to buy ten of them, the school pays 350 US Dollars in most places. In certain parts of the world (Brazil for instance), the import duties drive the price up to about 85 dollars, still within the price category where someone might send an RPi up in a balloon, not really expecting to get it back in one piece.

But comparing the 35 dollar RPi to the minnow board that costs 199 US dollars? 199 USD may not seem like a lot of money until you figure out that you can typically buy six RPis for the price of one of the “new” boards. Or when you import the minnow board into Brazil the price now doubles to 400 US dollars. Perhaps Intel/CircuitCO will manufacture the board in Brazil, which would drop the price to 199 US dollars again....that would help.

The Rpi also has a very low power requirement, typically 3 watts. This allows the Rpi to be more readily used for solar-powered experiments and battery-operated experiments than the higher power requirements of the Intel 86 system would not allow, or would be more difficult. I could not find the power requirement for the minnow, but it ships with a 5 volt, 2.5A power supply. Assuming that the minnow board actually draws 2A, that puts the board at 10W, or three times the power requirement of the RPi.

The RPi also does not have or need a heat-sink. The minnow has a large heat-sink on it which increases the weight of the board, and takes up space. At 4.2”x4.2”x the thickness of the heat-sink, the student needs a rather large pocket.

The RPi was not designed to be a part of an embedded system, but what people found was that the RPi could be combined with an Arduino and even relatively low-tech circuit boards to create other "products", and this explains a large part its success.

It is true that the minnow is more powerful than the RPi, but CPU power was not the design center for the RPi, nor was the design center to allow others to manufacture additional RPis in their basement.

What disappointed me most about the minnow was the fact that the board overall was not very “interesting”. Yes, the CPU is 64-bit, and yes the CPU had hyper-threading and supported virtualization, but you can buy more interesting boards (in my humble opinion) with Atom chips on them from Egghead or Amazon today for less money (and sometimes a LOT less money), albeit not as “open” in the hardware design nor as small.

Senior Embedded Systems Engineer David Anders of CircuitCo told ArsTechnica "We use an older processor for several reasons," he said. "The main reason is that internally to Intel there was a lot of concern about doing an open hardware platform because there's not really ever been one. We wanted to choose a processor that we knew was very stable. We used an older one to get our feet wet, so to speak, and understand the design." What? Is Mr. Anders trying to tell us that given the length of time that Atom chips have been around (and shipping), and the number of people inside Intel (and CircuitCo) that have done work with Open Source and Open Hardware that Intel felt they had to use an older CPU chip to “get their feet wet”?  With the number of graphics chips that Intel has at their disposal, they had to use one that still needs a binary blob?

I have known many people at Intel a long time. They have great engineers and great marketing people. Given Intel and CircuitCo's engineering and marketing expertise, I did, and do, expect more from them. And I certainly would not have expected them to compare “minnow” to the Raspberry Pi. Unless of course, it was the only way they could draw attention to the Classmate....er, ah, “minnow”, by utilizing the Raspberry Pi's popularity and success.

To me the name “minnow” is very applicable, because it is like a small, tasteless fish used for bait.

Instead of buying a “minnow”, I have purchased Adapteva's Parallella Board. At 99 USD the Parallela is much more interesting and (for my purposes) at least as open as the "minnow".