AMD Radeon Navi 2 / VCN 3.0 Supports AV1 Video Decoding

[Danny3]

It was a sarcastic question related to somebody questioning you for using window$.

Well, I'm not loyal to any tool that I use.
I just use the best tool for the job.
For some Linux is better, for others Windows is better.
I think I will continue to do that until one wins clearly and can do more than 95% of what the other can do, hopefully this will be Linux.
But this I think it will still need to take at least 5-10 years, time in which I might still need to use both
Hopefully, one day AMD will enable SR-IOV on consumer GPUs and Virtualbox will enable GPU passthrough.

Whatever the case, good times will come eventually.

[cjcox]

'

Dude, it's not hopeless. Nobody's actually getting sued for using AV1 and the world keeps spinning. Get therapy, you are depressed.

Please don't confuse my optimism with depression. Why??

[rene]

Yes and I stand by every word of it. You kind of ran away after I mentioned that I had worked on software decoders, didn't you?

VVC and AV1 also share a number of pixel-banging routines, so again it isn't completely different transistor sections on the die. It'll be the one programmable video decode engine, different microcode for each codec they want to support.

I did not run away, I have simply more important things to do and a real life. Your original post even mentioned MPEG, and while the codecs certainly share some conceptional similarities, they can not share a lot of details due to the patent encumbered nature of some industry pushed standards. Also AV1 implements several new coding techniques, and I not only have doubts this can "just" be _efficiently_ implemented in microcode, as seen in the news hardware vendors also spin new, dedicated hardware support for this.

[chithanh]

That rather depends on what you think is "the common definition of that term."

I mean both the historical definition, and the 2012 common definition of IEEE, ISOC, W3C, IETF and IAB.
https://en.wikipedia.org/wiki/Open_standard.
Also what is important to hardware vendors when implementing the standard, namely that any changes to the standard be approved by a standards body with fair and non-discriminatory participation rules. Hardware vendors will readily pay a license fee (even if they'd rather not), but throwing away your ASIC design and starting new on the whim of a single entity isn't gonna fly. Unless the stakeholders largely decided to move in that direction.

This is also why Intel's announcement of Thunderbolt going royalty-free was met with limited enthusiasm among hardware vendors. Only after it was submitted to USB-IF and subsequently standardized as part of USB 4.0 those vendors became confident enough to jump aboard.

Most academics, the European Union official definition,

To my knowledge there is no official definition that applies to all of the European Union. Per the Wikipedia article above, the scope of that definition is limited to the European Interoperability Framework for Pan-European eGovernment Services.

as well as that of the governments of several individual countries do not consider that an "open standard." According to their definition, that would be a "standard," but not "open." According to that definition, use has to be unrestricted (and therefore free of charge) for the "standard" to be "open."

Indeed that is the area where there is most disagreement between definitions, as can also be seen in this table: https://en.wikipedia.org/wiki/Open_s...of_definitions.
However the those who require open standards to be royalty-free are mostly definitions from individual countries and NGOs. The IGOs and international organizations (ie. when many come together and decide on a common definition) drop that requirement in favor of RAND licensing.