AMD's Hiring Another Open-Source Driver Developer

[Qaridarium]

i have an solution for the patent issue :

"Remember, remember the fifth of november..."

“We are Anonymous. We are Legion. We do not forgive. We do not forget. Expect us.”

[marek]

The problem is that AMD would probably have to pay extra licensing fees, which doesn't really net them any extra profit in return (because Linux users needing access to those features already have and use the proprietary drivers). It's possible AMD is paying a flat per-year fee or just paid a lump sum or even just cross-licensed some patents, but it's just as possible that they're paying a fee per software unit shipped. If the fees they pay are not tied to software, then yes, it's possible they could include a binary component (much like Intel supposedly did, if I recall) that includes those extra features which plugs into the Mesa drivers. Assuming, of course, that Mesa has the rest of the code in place to actually allow the drivers to implement the feature (Mesa's architecture is a little convoluted).

We do not have any idea what the terms are of the contract between AMD and SGI (and others)

I don't think there is any contract. ATI were sued by SGI for infringing on the FP patent. The court ruled that ATI did not infringe most of the patent claims, and SGI dismissed the rest. ATI wanted to invalidate the patent afterwards but they did not succeed. Even though it was a draw for both sides, it puts ATI in a not-so-bad position.

Here's more info: http://www.patentlit.com/2008/05/02/...ati-is-a-draw/

So from the information above, it looks like AMD/ATI has the rights to distribute software that *may be related* to the FP patent, implied by the court ruling. They already distribute their proprietary drivers, which implement FP rendering, and open source drivers implementing the same functionality would be no different.

Just FYI, there is the 'floating2' branch in my private repository on FDO that implements FP rendering in both core Mesa and r300g (the core Mesa work was mostly done by Luca Barbieri). It's a work-in-progress but there is very little to do to make it really complete. I am just waiting on a mark to merge it to master.

[Qaridarium]

I don't think there is any contract. ATI were sued by SGI for infringing on the FP patent. The court ruled that ATI did not infringe most of the patent claims, and SGI dismissed the rest. ATI wanted to invalidate the patent afterwards but they did not succeed. Even though it was a draw for both sides, it puts ATI in a not-so-bad position.

Here's more info: http://www.patentlit.com/2008/05/02/...ati-is-a-draw/

So from the information above, it looks like AMD/ATI has the rights to distribute software that *may be related* to the FP patent, implied by the court ruling. They already distribute their proprietary drivers, which implement FP rendering, and open source drivers implementing the same functionality would be no different.

Just FYI, there is the 'floating2' branch in my private repository on FDO that implements FP rendering in both core Mesa and r300g (the core Mesa work was mostly done by Luca Barbieri). It's a work-in-progress but there is very little to do to make it really complete. I am just waiting on a mark to merge it to master.

very nice..

[elanthis]

I don't think there is any contract. ATI were sued by SGI for infringing on the FP patent. The court ruled that ATI did not infringe most of the patent claims, and SGI dismissed the rest. ATI wanted to invalidate the patent afterwards but they did not succeed. Even though it was a draw for both sides, it puts ATI in a not-so-bad position.

Nice. Shame they couldn't get the patent invalidated. It really is "dead obvious" to a regular practitioner. The entire rest of the graphics pipeline is implemented in pure floating point on most hardware (and even a lot of software), and it was really just the buffers that kept things in fixed integer formats for space/bandwidth conservation due to limitations on the manufacture of processors, memory, and buses of the time. That patent should be shot in the face.

Unfortunately, there's still over 4,000 other patents that are at least tangentially related to OpenGL (most of which are probably not relevant to an implementation or most users, but it's a lot of shit to sort through to find out): http://www.freepatentsonline.com/res...ery_txt=OpenGL

Gotta love patents like "a method to translate OpenGL to OpenGL/ES". It's a fucking API abstraction. I hate this country's legal system.

[marek]

The info about patents related to OpenGL implementations, according to Khronos, is here:
https://www.khronos.org/files/ip-disclosures/opengl/

[smitty3268]

Even if AMD was willing to distribute a private Mesa version, I'm not sure it would be much help.

I have to think they would insist on doing a thorough code review of anything they put their name on, because I assume that at that point they could then be sued by anyone they hadn't properly licensed IP from. And given how slow they are to review their own internal code for drivers, it would probably be a matter of decades before they would be able to review all of Mesa.

Still, if they made it a priority it could probably happen.

[Prescience500]

I think things are already getting better: http://opensource.com/law/10/12/supr...dating-patents

In the future repealing software patents and what not will hopefully be much easier. Maybe open source/free software legal advocates can begin suing to have patents invalidated. If what I'm hearing is true and I understand correctly, then maybe some of the OpenGL specs might be a good place to start.

[Qaridarium]

I don't think there is any contract. ATI were sued by SGI for infringing on the FP patent. The court ruled that ATI did not infringe most of the patent claims, and SGI dismissed the rest. ATI wanted to invalidate the patent afterwards but they did not succeed. Even though it was a draw for both sides, it puts ATI in a not-so-bad position.

Here's more info: http://www.patentlit.com/2008/05/02/...ati-is-a-draw/

So from the information above, it looks like AMD/ATI has the rights to distribute software that *may be related* to the FP patent, implied by the court ruling. They already distribute their proprietary drivers, which implement FP rendering, and open source drivers implementing the same functionality would be no different.

Just FYI, there is the 'floating2' branch in my private repository on FDO that implements FP rendering in both core Mesa and r300g (the core Mesa work was mostly done by Luca Barbieri). It's a work-in-progress but there is very little to do to make it really complete. I am just waiting on a mark to merge it to master.

i'm not unterstand the article sorry why is amd not in an bad position? and why amd do not pay for that fp patent ?