Open-sourcing the Propeller I: an open letter to Parallax
cde
Posts: 37
The following is the email I just sent to Parallax, regarding the future of the Propeller I:
Dear Parallax,
Let me first say that I really like the Propeller I chip. It's easy to use, simple, has a very clean and orthogonal architecture.
I would like to bring a proposition, which you may find surprising at first: to open-source the VHDL code of the Propeller I, and in particular its reference implementation on the target FPGA board you used, as well as the documentation of the architecture.
Of course, this is a lot to ask for. But let me recap the benefits Parallax will gain from this:
Christophe
Let me first say that I really like the Propeller I chip. It's easy to use, simple, has a very clean and orthogonal architecture.
I would like to bring a proposition, which you may find surprising at first: to open-source the VHDL code of the Propeller I, and in particular its reference implementation on the target FPGA board you used, as well as the documentation of the architecture.
Of course, this is a lot to ask for. But let me recap the benefits Parallax will gain from this:
- Parallax will be known and recognized amongst the free software crowd, many of whom don't know about the Propeller and use (proprietary) AVR/PIC/... devices for their robotics projets.
- Following the announcement of the source code publication, we can expect major sites to pick up the news: Slashdot, Ars Technica, and so forth. This will make the Propeller known to a large population. As you offer easy to use Propeller board and robot kits, even casual free software hackers will be able to get on board.
- Many educational institutes rely on open-source cores for their teaching purposes. With an open-source Propeller I, they can explain the details of CPU design to their students, have them try it on an FPGA, and experiment with the DIP version of the chip at the same time!
- Any student to come across the chip will remember it and may reuse it (or the Prop II) at his/her futur work for commercial projects.
- Current and future Propeller users will be able to looks at the internals of their favorite chip and marvel at the beauty and sheer simplicity of it. I hope this will reinforce and "propel"
the user base to new heights. - Finally, Parallax will remain in the history of electronics and robotics for being the first company to open-source their product for the greater good.
Christophe
Comments
Hi Leon,
Right, Parallax used AHDL -- I assume it could be open-sourced as well.
I can see sharing basic algorithms and concepts, but to open source something like the prop itself would completely devalue the intellectual property and basically turn everything into "he with the cheapest FAB wins".
C.W.
Amtel doesn't open source their chip design. Why should Parallax? (Or have I missed something here?)
The Ardunio IS and open source platform, so is the Propeller Platform. What's the difference?
OBC
They didn't use that either:
http://www.parallax.com/Portals/0/Downloads/docs/article/WhythePropellerWorks.pdf
Well, with software that makes a lot of sense, and the reasons are well known and IMHO, both practical and competitive.
Hardware does not work the same way. There is no significant material advantage for Parallax in this move, IMHO.
In the end all open sourcing the Prop design would do is hurt Parallax.
If the open source community is jonesing for a open version of the Prop they can engineer their own just they've done for several other microprocessors.
Rod,
Is there a lot of cutting-edge, innovative IP within the Propeller I? Indeed, it's a beautifully simple and elegant design, but it's not revolutionary either, and Microchip/Atmel/etc. won't suddenly create much better designs thanks to the Propeller source code.
How could you possibly know that without understanding what is actually going on in the Prop? The Prop is a unique processor in the industry and there are certaintly features that would/could be valuable to the other micro companies to Parallax's detriment. Parallax would be absolutely nuts to open it up.
Bill
I, and others, have worked on replicating the prop on an FPGA. When you do this, you begin to understand the sheer brilliance in the design, and also some of the fantastic ways Chip minimised the silicon in the prop.
However, there is another possibility here. Parallax could license the Prop IP and/or the PropII IP (but closed, not open) for inclusion on an FPGA. A reference pcb could be done - I would love to do one, not that I am an expert.
It probably couldn't hurt coming up with a royalty and pricing structure and then do some press released on embedded magazine etc., just get the possibility out there, who knows might get a bite from a small firm that can take the design, add to it, and re-purpose it in another chip or variant.
Andre'
I never seen the old threads. I shall assume their experiences were unpleasant.
Bruce
Not everything of technical value is patented. Many very valuable techniques are kept as trade secrets. With so many countries able to run industries on little else but cheap labor and with no regard for intellectual property whatsoever, it's sometimes best to do your dancing in the dark.
I agree 100%, and that is probably very true concerning semiconductors. I was just surprised to get zero results. I figured I would find at least one or two.
Bruce
Audacious. Have you also sent such a request to Intel, ARM Holdings, MicroChip, Motorola, IBM, Texas Instruments, etc etc? I would be fascinated to see any replies you might get.
Lets' take your proposal one point at a time:
Known ant recognized, yes, but to what purpose? I would imagine that the overwhelming majority of that audience is not interested in making their own processors or getting into the expensive and
complicated world of FPGAs to implement them. Those that want to play in FPGA land have a huge range of other open source processor cores to choose from. http://opencores.org/
Not really. It's just another processor design among hundreds. Most people would skip over it as they do the others.
True, but as we see there are already dozens of processor cores to study out there. There is nothing that special about the Propeller from an academic point of view.
(Of course it is special in that Parallax had the balls to put those elements together like that as a product.)
Perhaps. Not a compelling argument. I would imagine that anyone who has taken the trouble to gain an understanding of processor design has a wide view of the landscape when it comes to selecting
processors for products.
That would be nice, for us. What does it do for Parallax?
That is true. But history might record that they are the first to give away the farm and go bust!
Further:
It seems somewhat pointless to actually use a Propeller architecture on an FPGA. Why would I write a software UART, for example, to run on a Prop COG on an FPGA when I can just use a VHDL UART which
would be smaller and faster anyway? The Prop is a means to replace hardware with software and that's what I can do more effectively in VHDL if I have an FPGA.
How do you counter the possibility that some Chinese garage company could order up thousands of Propeller clones form a fabricator and eat Parallax's profits?
As far as I can tell most "open source" hardware from Parallax or other supplier is no such thing.
Sure they might give you schematics, PCB designs, gerbers, mechanical drawings etc but at the end of the day you are reliant on a very closed source processor and other devices.
As for "getting it", seems some people do and it's big business, see here:
http://singularityhub.com/2010/05/10/13-open-source-hardware-companies-making-1-million-or-more-video/
I understand the arguments for open source, and I know that some companies have made it work in there favor, in many cases because they are just providing consulting services around software or hardware that some other poor schmuck gave away for the "greater good". I guess I see it as being way to Karl Marx, and I'm more of an Ayn Rand kind of guy.
Don't misunderstand, I'm not for being a knowledge hog, I'll answer pretty much any direct question to anyone about a topic if I know they are truly interested and willing to do some of the work themselves, I just can't see giving away the farm.
C.W.
Not wanting to debate the benefits etc., I just don't understand why people feel compelled to request a company to change their business model.
Chris
Questions are not dumb. NOT ASKING questions is dumb. Asking questions is the opposite of dumb. (Although I can see how answering the same question over and over can get annoying, but that tangent is in the opposite direction).
Thanks for presenting this issue cde, I never explored it, and the responses are informative.
The decision is ultimately up to Parallax, but my guess is, they are just laughing at the idea. As it was pointed out earlier, it takes a lot of money to develop a product, and by sharing that information, the product would become an easy target to being copied and marketed at a price which Parallax would be unable to compete with, essentially cutting their own throat.
Bruce
Open code and open data are very important things. With data, having it in open formats means being free to use and own your data, as opposed to somebody else owning it. If you record a movie, and it's contained in a closed software that you and everybody else has to pay for to view it, who owns the movie? Stephen King found this out when he got locked out of one of his own books! Having schematics, code, and other bits available means having what is needed to move your investment from one device to another, or modify it, or anything really. This lowers costs and keeps the dollars spent on actual value. The Prop is real value. What we pay for one either returns more to us, or not, and if it returns more, great! All the docs, code, and things surrounding a prop add value, and keep overall costs low, because we all can benefit from information.
And with software there is a very interesting dynamic in play too. For common, basic computing tasks, open code means being able to do them at a very low cost. That stuff is common knowledge, and where that's true, why not write our own software, work together and all profit from the high use value? That's what open code is about, and it's about freedom too. Want to use YOUR device? Do it! The body of open code presents a lot of use value to anybody seeking to use it. If they contribute something, that value goes to everybody. Physical things don't work that way, but software and information generally does. That's the motivation to build on open code. You get the high use value up front, and if you contribute back, everybody gets more. Sometimes it makes sense to pay somebody for that use value too.
Now there is very expensive closed software available today. People pay that because that software has real value, and it's not common knowledge, and it's closed so that the money can be made to support the making of it. A office program is common knowledge, and one of those is free. That's how open code works to keep the big companies honest. If they are not really adding value, people will just use open code, and the cycle continues. It's worth it to buy expensive software, if it does something of real value.
What gets abused is open equating to free. Two different things.
If we open the Propeller, we don't need Parallax. Well, why do that? It's ok to need Parallax because they are adding a lot of value by producing the Propeller. See how that abuse works? It's very important to not mix up what open code and open data mean, with "let's make it all free" kind of open we see all the time. One is a seriously good value and source of learning and using that is largely free for us to do what we will, and the other is simply a strong desire to get something for nothing.