HDMI Forum Rejects Open-Source HDMI 2.1 Driver Support Sought By AMD

[pete910]

Look at your picture.

You're running 8bpc with YCbCr 4:2:0, which takes half the bandwidth of 4:4:4, so 12.91Gbps, which fits inside HDMI 2.0 speeds (FRL2).

Also, is your display really DCI 4k, i.e. 4096x2160? Most are 3840x2160. If you are, then it's 13.71Gbps, which still fits inside FRL2.

Now it makes sense!

[F.Ultra]

Feel free to see it that way, if it helps you sleep better at night.


Yes but look at Ubuntu, you can check a box on install to get patent encumbered codecs installed and this doesn't seem to be a problem even in the US. Why would it be different with HDMI?

That are for software for which Canonical/Ubuntu have licensed the patent for use by it's users. HDMI IF will not give Ubuntu such a license to the HDMI patents to carry software that violates the HDMI trademark and copyright.

[duby229]

That are for software for which Canonical/Ubuntu have licensed the patent for use by it's users. HDMI IF will not give Ubuntu such a license to the HDMI patents to carry software that violates the HDMI trademark and copyright.

This guy here^^^

They can't patent software that they didn't write. They can only patent their own proprietary software. If we wrote an open source implementation based on specifications that were discovered using clean room reverse engineering, they could -not- patent it. This is -not- a patent issue. It's a license issue.

EDIT: Once an open source implementation exists then they can take their patents and their logos and their licenses and go fuck off with them. It won't matter anymore.

[F.Ultra]

AMD can't legally do it. They have access to the spec and proprietary implementations of it. They are tainted already. The developers of a clean room implementation would have to be able to prove to a court that they've never had any contact or influence or any kind of relationship with it. AMD already does.

A group of people like us definitely could because we can prove that.

It's not a leap of logic, it's actual law. That has already been tested in court. There already exists plenty of case law.

EDIT: Recently the US supreme Court presided over a case concerning googles implementation of java. Ideologically it was similar to this circumstance. In the end they didn't even consider the technical argument and cited fair use. Like I said, if you can prove you had no influence or relationship to it AND you can prove it was done compliant with clean room techniques for reverse engineering AND you can prove it was necessary for interoperability, then it -is- legal. This scenario clearly falls under fair use, not dmca.

It's a leap of logic because you used the presence of an AMD created clean-room implementation as proof that HDMI IF have no patents covering the HDMI 2.1 specs but since there are no such AMD created clean-room implementation that whole idea falls by default.

Any potential clean-room implementation will not matter at all _IF_ HDMI2.1 is covered by patents because a clean-room implementation can only avoid copyright issues, not patent issues. Patents covers ideas while copyright covers implementations and you cannot clean room the idea, only the implementation.

[F.Ultra]

This guy here^^^

They can't patent software that they didn't write. They can only patent their own proprietary software. If we wrote an open source implementation based on specifications that were discovered using clean room reverse engineering, they could -not- patent it. This is -not- a patent issue. It's a license issue.

for crying out loud. Please educate yourself on what patents are first before you make these silly comments. You are thinking, talking and arguing about Copyright but write Patent. They are two completely different things.

And yes for AMD this is a license issue since they would loose the HDMI license if they made their patch public. We don't know if that patch implements ideas that HDMI IF have issued patents for (and the reason we don't know is because HDMI keeps this a secret), if they have done that then whatever implementation or reimplementation you do to solve the same problem will be covered by their patents, it doesn't matter who wrote the software, what matters is who filed the patent for the solution to the problem.

[kiffmet]

How do SBC manufacturers like Rockchip and Amlogic solve this issue? The RK3588, Amlogic A311D2 and S905X2/Y2 are said to have working HDMI 2.1 support on Linux.

[hartree]

On the bright side, I can confirm the current DP to HDMI adapters work for 4k 120Hz RGB with VRR and HDR on linux.
Tested with a Cable Matters 102101 adapter with VMM7100 chip and firmware 7.01.124, a Radeon 6900XT, Nobara 39 and a LG C1.

[duby229]

It's a leap of logic because you used the presence of an AMD created clean-room implementation as proof that HDMI IF have no patents covering the HDMI 2.1 specs but since there are no such AMD created clean-room implementation that whole idea falls by default.

Any potential clean-room implementation will not matter at all _IF_ HDMI2.1 is covered by patents because a clean-room implementation can only avoid copyright issues, not patent issues. Patents covers ideas while copyright covers implementations and you cannot clean room the idea, only the implementation.

Wtf dude? That is not what I said at all... Actually as I've said repeatedly AMD can't make an implementation, they're already tainted, but people like us -can- I'll let my words speak for itself. You seem to have a comprehension problem, which is ok, but now that I'm recognizing it I just have to let your opinion not bother me.

[duby229]

And by the way, just because it seems some people believe you can, you can't patent ideas. That's NOT what patents are. Your patent has to be sufficient to prove that it would work in reality exactly as described.

EDIT: Prototypes aren't required to prove your patent, but it's how most patents are derived. You make something and then you recognize that you've got some things that are patentable

[Anux]

That are for software for which Canonical/Ubuntu have licensed the patent for use by it's users.

Do you have a link that says so? Canonical would certainly have made this public but I cant find anything.

If you take a look at VLC which basically any Linux distro (even debian which is super careful with patented stuff) ships: https://www.videolan.org/legal.html
I don't see why this wouldn't be possible with an HDMI driver.