Hardware Freedom Day is April 20th!

Alas, the Hardware Freedom I've wanted for decades is to be able to program FPGAs using my own tools, but the information needed to do this is locked away by the FPGA vendors. 

Oh well, maybe someone will come up with a 3D printer for microelectronics some day

Alas, the Hardware Freedom I've wanted for decades is to be able to program FPGAs using my own tools, but the information needed to do this is locked away by the FPGA vendors. 

Oh well, maybe someone will come up with a 3D printer for microelectronics some day

Hate to mention this at this late date .. but 4/20 is already a symbolic date in some cultures ..

Are there any FPGA vendors that are more open than others?  I believe Xilinx and Altera are pretty much the whole market but it would nice to highlight any smaller players that embrace openness.

Drew Fustini wrote:

 

Are there any FPGA vendors that are more open than others?  I believe Xilinx and Altera are pretty much the whole market but it would nice to highlight any smaller players that embrace openness.

The only one I ever heard of was Atmel [bet you didn't even know they even made FPGAs!] as part of the FPSLIC family IIRC.  I don't know if this is still the case or even if the FPSLIC family still exists.

The excuse constantly given is that customers fear that if the architecture is open, then competitors will be able to reverse-engineer and copy products.  That excuse is pretty feeble in devices that have internal storage and security bits that prevent reading back the contents, or in devices that include bitstream encryption.  But given that only a handful of highly-idealistic people are asking for this, and we don't represent any direct sales, the chances of any vendors changing its mind on this is a long shot.  Still, I make cranky comments whenever possible and maybe one day the right person will see the light.  As the bumper sticker reads, "People are deceived en masse and enlightened one at a time."

There was a lot of talk over the decades about FPGAs being used as reconfigurable compute engines.  This use has never really had any traction, and my hypothesis is that this hasn't occurred because FPGAs are too hard to design with using the current tools.  The tools are fine if you're designing a custom chip for a product and your other choice is ASIC.  But if your other choice is to wire up an array of high-performance microprocessors, the FPGA solution is too difficult.

If FPGAs were open like many microprocessors, people who could solve this would work on it.  But given that any results will be "purely academic", the kind of people who want to see something actually working (e.g., all engineers) have no incentive even to start.  It's IMO truly a missed opportunity, but could be rectified at any time by an FPGA vendor if it chose to do so.

John Beetem wrote:

 

If FPGAs were open like many microprocessors, people who could solve this would work on it.  But given that any results will be "purely academic", the kind of people who want to see something actually working (e.g., all engineers) have no incentive even to start.  It's IMO truly a missed opportunity, but could be rectified at any time by an FPGA vendor if it chose to do so.

Perhaps what might work is to appeal not to our interest but to the self-interest of manufacturers:

• The big boys have the sector all sewn up, but none of them offer a product with full open documentation.  If you are a small company that wants to challenge the incumbents, this gives you a wide open market niche with no competition.  And even after some of the incumbents have joined you, you will still be a de factor leader in this new open sector.

• The news of appearance of a fully documented FPGA or combo PLD+microcontroller would spread like wildfire among the open hardware and open source software communities --- half the job of marketing would be done for you, and for free.  The synergy between "cheap", "open", and "community" is incredibly high.

• The dedicated open source EDA communities would be able to provide tooling configured to your devices right out of the box and with no restrictions on distribution.  That's a direct path to your company door, for free.

• Many open source fans would be more than happy to contribute to software projects directed by you, as long as the resulting software is fully open source.  This can reduce or even totally eliminate the high cost of in-house software development.  What's more, the never-ending burden of software maintenance practically disappears.

• You will be the dearest friend of educators in EE and related sectors, for whom open toolsets and complete documentation is extremely important and a matter of principle.  Using proprietary tools in class is really just company-specific vocational training, not eduction at all (except possibly as a side effect),  And once exposed to your devices as undergrads, engineers will naturally gravitate towards them later as well.

• Plan for the unexpected.  Arduino capitalized on the openess of Atmel AVR microcontrollers and created a whole new and extremely active community meme space, and I'm sure that Atmel are not complaining.  They're well deserved benefactors of their policy on open documentation.  One cannot forsee exactly what will happen when a fully open FPGA appears, but the opportunities it opens are so immense that it is quite reasonable to be "ready for the unexpected", and then to embrace the community's excitement and fuel it further.  It all leads to sales.

I'm really quite surprised that it hasn't happened already.  The prize is there for the taking.

Morgaine.

Morgaine Dinova wrote:

 

...

 

I'm really quite surprised that it hasn't happened already.  The prize is there for the taking.

I'm not surprised it hasn't happened.  First, it couldn't have started happening until about 5 years ago when the basic FPGA patents began to expire.  IMO Xilinx's 4-input LUT is great granularity, particularly if you can also use them as distributed RAM cells.  And the Xilinx approach to routing is very good.  Now that the patents have expired on most of the core ideas others can run with them without getting sued.

Second, we have the classic chicken/egg problem.  Nobody is going to develop tools for a part that doesn't exist, and nobody's going to buy a part that doesn't have decent tools from day one.  So you have to both produce a chip and tools, and have them working as reliably as Xilinx (WLOG*) on the day of release, and have a chip price that's not embarassingly higher than Xilinx.  This is pretty hard and requires a team with a lot of diverse skills.  You can do a kick-starter for a small hardware project and make PC boards for a few US$10K.  But try raising the US$10M you'd need to get a new chip plus tools off the ground?

I was hoping that an Asian semiconductor manufacturer might take advantage of the patents' expiration and produce something, but they'd have to do something about the tools.  So not very likely.

Anyway, here's my approach:

1.  Write cranky but non-abusive comments on the subject at every available opportunity.  Maybe someone with both imagination and authority will see them and become enlightened.  It only takes one.

2.  Work on my own tools so an infrastructure is in place if "miracles do happen".  If it's clear that my tools are 90% of the way there, it gives a prospective vendor the idea that all they need to do is make the chip, eliminating 90% of their risk.

3.  Wait for 3-D printers to get down to the micron level so I can design my own FPGA   I'll still need my tools, so I'll keep blasting ahead with #2.

*WLOG = Without Loss Of Generality, a wonderful phrase you sometimes see in mathematical proofs.  In this case, I mean that my comments apply equally to Altera, Microsemi (who now sells Actel chips), Lattice (who recently acquired Silicon Blue), Cypress, and anyone I forgot.

John Beetem wrote:

 

I'm not surprised it hasn't happened.  First, it couldn't have started happening until about 5 years ago when the basic FPGA patents began to expire.  IMO Xilinx's 4-input LUT is great granularity, particularly if you can also use them as distributed RAM cells.  And the Xilinx approach to routing is very good.  Now that the patents have expired on most of the core ideas others can run with them without getting sued.

That's very relevant and insightful, well spotted,  Patents are such an almighty disaster for all things technical, and this is yet another example of them holding back progress in technology under the totally fake excuse of "encouraging innovation".  That hasn't been true since the end of the Industrial Revolution, and it's debateable if it was true even then.

Second, we have the classic chicken/egg problem.  Nobody is going to develop tools for a part that doesn't exist, and nobody's going to buy a part that doesn't have decent tools from day one.

Although this is true, there are many different ways to swing this particular cat.

One approach which I like on principle is open and collaborative, and even has a widely-known existing example:  Parallax's forthcoming "Propellor 2" is being designed in very open collaboration with their community, to the extent of Parallax providing their bitstreams for emulation of their forthcoming device on Terasic's DE0 NANO and DE2-115 FPGA boards (Altera Cyclone IV).  It's working out extremely well for them, with the community even identifying design bugs that would have been much more costly to fix further down the line.

The above is an example of collaborative device development, but I think it applies equally or even more so to software tools.  In other words, the chicken and egg problem can be solved by making the two developments proceed concurrently.

Parallax's forthcoming "Propellor 2" is being designed in very open collaboration with their community, to the extent of Parallax providing their bitstreams for emulation of their forthcoming device onTerasic's DE0 NANO and DE2-115 FPGA boards (Altera Cyclone IV).

Wow! I hadn't heard of that. Thanks for pointing it out.  What an excellent example of a true Open Source Hardware development cycle.

Apparently it was Culture Freedom Day yesterday.  Pity that this very noble intention is only a joke in practice, as popular culture has been locked away behind corporate firewalls and "suing as a business" for some decades now.  Prof Boyle's extremely insightful articles and public domain book about it are available here .

Future historians may well refer to the current era as the Dark Ages of Culture, as it's nearly all locked away.  The current generation has benefitted by everything from the first half of the 1900's being totally free and in the public domain (just try to run a music course without it), but not even the Library of Congress is legally able to open its archives of the last half century to the public. Humanity's legal parasites have won, bigtime.  They've imprisoned this generation's cultural wonders, and there is no visible end to the sentence.  Greed, greed, greed.

A few cultural repositories like Archive.org and Jamendo.com are bucking the trend, but it's a drop in the ocean,  These are grim days for cultural freedom.  In 2013, unless you have a well-padded wallet or purse, you are legally required to walk in the cultural darkness.