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

[Volker Schmidt]

No big issue here. I hope we will see hdmi free TVs soon. They should be a good deal cheaper.

[LtdJorge]

So stay away from the HDMI junk and use Displayport instead. I'm fine with that.

I'm on LG OLED C3. 4xHDMI, 0xDP 🥲

[duby229]

Clean-room reverse-engineering is the entire reason the "IBM/PC compatible" market exists. IBM used off-the-shelf parts for everything in their PC's except a proprietary BIOS. The BIOS was reverse-engineered separately by Columbia Data Systems and Compaq with no grounds for legal challenge, and here we are.

As long as it can be proven that the developers had no prior influence or relationship, the development was compliant with the definition of clean room reverse engineering and that it was necessary for interoperability then there is no basis for legal challenge. Compaq was clearly able to demonstrate all of these requirements, Although it is true IBM settled out of court with several other companies prior to that, probably because they couldn't prove some combination of those requirements.

[LtdJorge]

I don't get what the problem shoudl be? Code is code, not trademark. If AMD has working code just publish the patch. Also don't name it HDMI, but NotDMI instead, ... what problem is left then? It is obviously implemented in the hardware. Just toggle the right bits. Totally don't see what the problem should me.

The problem is the spec is not open, and is protected by IP laws. Which means, if they make an implementation Open Source, they might be publishing the spec as open, basically. The only way they accept implementations is as closed source, as if reverse engineering wouldn't be possible that way.

[Hibbelharry]

I'm on LG OLED C3. 4xHDMI, 0xDP 🥲

I know: it's big! it's bought recently! it's relatively cheap! yada yada.

But: connecting a TV to a computer also has some drawbacks. You just named one of those.

You could also have bought an OLED computer monitor which would have had DP inputs but you've chosen to cheap out on that. I know you still spent much money but still you cheaped out. We need to name the options you've had. And it's a pitty that the cheap stuff currently doesn't play well with the ideals of linux but thats the way it is currently.

[erniv2]

AMD should stick it to the HDMI Forum by dropping HDMI support for all of their high-end GPUs. Standard DisplayPort and USB-C ports only. Then, sell an official DisplayPort -> HDMI adapter for people who want "legacy" HDMI support..

Thats not how DP > HDMI Adapters work, DP knows how to talk TDMS the HDMI code format, and DP > HDMI crossover cables are non active there is no Compute logic involved, itś just pin compatibility so the HDMI standard still get negotiated between your GPU and the TDMS in you TV no way around.

And even if you where to buy a DP > HDMI cable with 2.1 support with an active logic that would be just a form of upscaling and not the signal you would want to have~~.

Maybe nivida could sell you a 1080p to 4k cable with vsr logic for a few 100 bucks .

[chithanh]

I have an R9 Fury X hooked up to my living room TV, and guess what? I use a DP to HDMI 2.0 dongle (so I can at least get 4K@60). It was I think AU$10 or something ridiculously cheap. I don't see what the issue is.

This is what DP to HDMI 2.0 dongles cost today.
When R9 Fury launched, those were not yet available in the market.
DP to HDMI 2.1 dongles are way more expensive.

only Linux users and specifically, AMD gpu users care about it.

Windows users aren't even contemplating this dilemma

You couldn't be more wrong. Most users who depend Windows know that Microsoft has them by their balls. Only when Linux is a viable alternative it can keep Microsofts ugly business practices in check. This is why even users who would never run Linux on their desktop PC are happy that Valve released the Steam Deck and created a useful Linux system for gaming. It keeps Microsoft on their toes and maintains the credible threat of switching away if the anti-consumer behavior worsens.

It is also unknown to me whether the in-GPU HDMI hardware is HDR- or CEC-capable which would in turn make the converter mandatory.

Source: Fail0verflow's presentation at 33c3 on running Linux on the PS4.

I was there at 33C3 watching Hector Martin hold his presentation, so I am aware.
While we can only presume what was the reason for the discrete HDMI encoder, we do know that:

• No other ATI/AMD graphics chip produced to date supports HDMI CEC
• Sony is on record stating that they planned HDR for PS4 right from the beginning
• Microsoft Xbox One which launched at the same time supports neither HDR nor CEC. (Both Durango and Liverpool APUs are essentially larger versions of Kabini.) HDR support came only with Xbox One S and Xbox One X.

The "Sony needing to buy Panasonic HDMI encoders to maintain a good business relationship" explanation is completely unnecessary and IMO without merit. Sony would have chosen a more high-margin product for such a strategy.

Thats not how DP > HDMI Adapters work, DP knows how to talk TDMS the HDMI code format, and DP > HDMI crossover cables are non active there is no Compute logic involved

This is how passive (DP++) adapters work. Active adapters pose to the host as DisplayPort monitors. HDMI 2.0 and all USB-C adapters are generally active. Most adapters that support only HDMI 1.4 are passive, but some (marketed as Eyefinity capable) are active.

[F.Ultra]

That's not true at all. The legality of it has to do with the definition of "clean room reverse engineering". If the group doing it has never had any contact or relationship or influence with the specifications that they're reverse engineering and the entire process is thoroughly documented such that a lawyer can defend the actions taken as being compliant with the definition of clean room reverse engineering and that it was necessary for the sake of interoperability.... Then yes, in the US it would be legal.

EDIT: AMD couldn't legally do it, but a group of people like us could....

no that is not at all how patents work. Two companies/people can independently come up with a similar solution/invention but if one of them have patented it then the other is in violation. Clean room only solves the copyright problem, the only solution to the patent problem is to either abolish software patents, wait till they are expired or license them.

It wouldn't. The issue isn't patents, just the license agreements of the specifications. The specifications may not be published by any members, but HDMI IF has no legal leverage against people reverse engineering it and then publishing it. Sure they'd most likely sue and maybe one of their lawyers may be able to get some win out of it, but just because the accused party wouldn't have the money to hire a similarly competent lawyer. They have no basis to sue.

HDMI IF also have a patent pool which you gain access to if you license.

[duby229]

no that is not at all how patents work. Two companies/people can independently come up with a similar solution/invention but if one of them have patented it then the other is in violation. Clean room only solves the copyright problem, the only solution to the patent problem is to either abolish software patents, wait till they are expired or license them.



HDMI IF also have a patent pool which you gain access to if you license.

Except this isn't protected by patent law. It's protected by license agreement and that is copyright law.

EDIT: The specification isn't software, the implementation of the specification is. They can patent their proprietary implementation all they want. As long as you don't touch their proprietary implementation and you can prove it with clean room reverse engineering techniques then you can't violate their patent. Same thing goes for hardware patents by the way.

As long as you can prove you have no influence or relationship to it and you can prove that the specifications were reverse engineered using clean room techniques and you can prove it was necessary for interoperability... then it -is- legal.

The only way it could be argued to be a patent violation is if you failed to prove you had no influence or relationship to it or if you failed to prove it was done compliant with clean room reverse engineering techniques or if you failed to prove it was necessary for interoperability. But if you can prove all of this then it is -not- a patent violation.

[Entzilla]

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

One of the limitations of AMD's open-source Linux graphics driver has been the inability to implement HDMI 2.1+ functionality on the basis of legal requirements by the HDMI Forum. AMD engineers had been working to come up with a solution in conjunction with the HDMI Forum for being able to provide HDMI 2.1+ capabilities with their open-source Linux kernel driver, but it looks like those efforts for now have concluded and failed...

https://www.phoronix.com/news/HDMI-2.1-OSS-Rejected

Tank you Larabel for bringing this piece of news to inform the community. Monitormanufacturers ought to drop HDMI as a "standard" or at least demote HDMI to a "well we can add 1 port to support consoles, but that's it" type of approach.

Holy (or HDMI if you will) inferiority complex, batman.