Thanks for your thoughts.

We are in agreement mostly. Note that for now I am looking at/discussing the problem, not evaluating solutions. Once we (well, I) understand the problem better, and from the perspective of the cores vendors themselves, perhaps we (I?) will be able to come up with solutions that do not “F@#)(* with the cad tools” that much and are worth while?

It should have read:

“the results will be the same as now; honest players will license your _IP_, and dishonest ones won’t.”

Just as with music DRM, you’re trying to make bits uncopyable. It may seem that this is practical inside specialist hardware, but those bits will end up being just bits again, once they’re inside the FPGA. It only takes one cracker with an electron microscope and a bit of time and ingenuity to crack the DRM system once, rendering your DRM’d IP back into logic equations, and the whole process is rendered futile. Expect this to be available as an illicit commercial service somewhere in China within three months of your system going into production.

The only way to fix this would be to cripple the FPGA industry through vastly complex mandatory DRM measures designed to prevent FPGAs loading arbitrary sets of logic equations, and a huge infrastructure for somehow tracing the provenance of every subexpression of every system of FPGA equations in the world, no matter how transformed or obfuscated, back to its rightful owner. Since this is (a) practically infeasible (b) quite likely mathematically impossible, (c) defeats the whole point of FPGAs, it just isn’t going to happen.

The results will be the same as now; honest players will license your DRM, and dishonest ones won’t. (The countermeasures will also be the same; the only players worth going after are the big ones, and they have so much to lose that it’s far, far, cheaper for them to licence your IP than to risk your reverse engineers catching them pirating it.)

The music industry has just spent a decade learning this lesson, painfully and at great expense.

There’s already some precedent for open-source hardware and it’s not obvious why this shouldn’t be taken further.

DRM tends to mess things up alot, as it requires changing FPGA architectures, F@#)(* with the cad tools, etc…