for the latest news viz-a-vis "software patents in EU". (Related information on the sites
formatting link
and
formatting link
).
Questions:
How much of this applies to FPGA-design?
Is there any danger that here also this field would be mined with a lots of silly patents? Has that already happened in USA?
Am I right that especially the Open Cores movement would be particularly vulnerable to that kind of "development" with regards to patent (silly or not) infringement lawsuits?
I guess this is not so much problem for design houses using their own cores whose HDL- and other design-files are kept as undisclosed trade secrets.
What has been and what WILL be the view of patent organizations on HDL-designs? (E.g. if the amendments proposed by FFII and related folks get some day through in the EU legislative process).
I.e. is a Verilog or VHDL-source file just "a description for hardware device" or is it yet another (strange) form of software?
Most of the software patent discussion does not apply to FPGA design because fpga designs are clearly covered by existing patent law. However algorithm patents could hurt in some places. (Thin Bresenham, think CORDIC)
Rememeber that a sigma/delta ADC is made from a DAC and a comparator? Austin Lesea has a patent on the idea that you could use the differential inputs of the fpga as the comparator.
Xilinx also has an patent on implementing an 8-input AND in a single CLB. Hopefully your Microblaze clone does not use this feature, they could use that patent to come after you. (Peter Alfke stated in this newgroup that they wouldn't)
Also, Xilinx has illegal european software patents since 1997: Patent EP000000951687B1 [EN] PROGRAMMABLE LOGIC DEVICE PLACEMENT METHOD UTILIZING WEIGHTING FUNCTION TO FACILITATE PIN LOCKING But the European Council still claims that software is not patentable in europe.
Yes.
It is easier to identify patent infringing in open source software, but the damage is usually at least as large for small and medium sized companies. Only the large companies are safe because they achive a balance of power with patent exchange agreements.
Well, they are proposed by the parliamt. It the council that ignored them.
The VHDL source is not patentable. It is covered by copyright. That is something entirely different and completely unrelated.
Patents cover the ideas of the design. Ideas how to build a circuit are clearely patentable in europe, no matter if you use HDL, schematics or a mask layout program to describe the circuit. (Example: 8-input and-gate in a single CLB)
Algorithms as such are currently officially not patentable int the EU, no matter if you use a uC, FPGA or specialized VLSI to implement it (Example:Bresenham Algorithm)
The patent layers try to sneak through this by filing patents of the form "Bresenham algorithm implemented solely by NAND gates". But these patents are very weak because the amount of innovation is so low. (A synthesis tool could come up with the same idea)
I guess here you mixed me with Antti Lukats. However, I come from the other side of the gulf, and I'm just beginning to learn the secrets of EE.
Yes, of course. However, what I was striving after here was that what happens if the public (and EPO) opinion at some point comes to view the writing of Verilog/VHDL-modules as a creation of SOFTWARE (for these "massively parallel reconfigurable processing chips", i.e. FPGAs), instead of viewing it as a description of HARDWARE? This view might come more prevalent if the FPGA's will be getting more and ASIC's less common, and if the data processing/computing applications will prevail over more traditional EE-fields.
For an example, see also John Jakson's view here:
formatting link
"I am pushing the idea that FPGAs are in fact Transputer cousins at a much finer level of granularity ..."
If you can argue that there is a continuum between hardware (wired up TTL chips), through ASICs to FPGAs to compiled C code in RAM, then you would have to say that software can be patentable. Where is the line between what is patentable and what isn't?
If I remember the story from the IBM S/370 and virtual storage, it was that the algorithm (logical structure of virtual memory) wasn't patentable, but the hardware implementing it was. Now, most, if not all, S/370 were microcoded, so wouldn't that be software? The distinction is not easy at all.
Actually that is the trick (the other way around) that is currently used in the EU to file software patents. You do not patent an algorithm, but the "idea" to run the algorithm on a processor, claiming that the combination of algorithm an CPU is a device, the algorithm beeing the configuration of the hardware.
But that really is against the words and the meaning of the law. The idea to run Bresenhams algorithm on a CPU is not a technical invention. If it is, it is so similar to running Quicksort on a CPU that the level of innovation is not high enough for a patent.
But of course you have a big grey area. Because it is hard to distinguish a new multiplier architecture from a new multiplication algorithm. And the barrier of where patents get absurd might indeed move more towards hardware.
If you look at the discussion about software patents, the core argument is, that software is relative easy to develop, and as a result, software projects tend to get huge and incorporate thousands of patentable ideas (or combination of ideas). To check whether a software project violates any patents therefore becomes impossible and software development becomes a gamble.
To check an RF receiver or a new chemical for patent violation on the other hand is relatively easy.
With large FPGA projects with ever improving high level synthesis the situation begins to look very similar to the process of software development.
I believe that the australien view that the patent system in general is not beneficial for society is right.
OK, but say you use Bresenhams algorithm to run the stepper motor on a plotter, implemented in TTL. Would you consider that hardware enough to patent? Consider that it could still be coded in verilog.
I think you have the right idea, though. The level of innovation, either software or hardware, should be the deciding point, not the method of implementation.
Not to mention the problems with patenting genes, DNA or protein sequences, especially when their function is not known.
Unfortunately as of the 1st of January 2005 we have adopted the US system Lock, stock and barrel as part of a trade deal with the US.
Including copyright and patent law.
Fun thing is that some patents that expired here are now back in force (Australian patent expired but the US patent hasn't)
A lot of lawyers here and in the US are rubbing their hands in glee. Quite a lot of small businesses are sweating at the moment on their company names.
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.