AMDVLK 2021.Q4.2 Released With Dynamic Rendering, Other New Extensions

[phoronix]

Phoronix: AMDVLK 2021.Q4.2 Released With Dynamic Rendering, Other New Extensions

It's been over one month since the release of AMDVLK 2021.Q4.1 as the latest open-source AMD Vulkan driver update, which is off the wagon compared to the prior weekly/bi-weekly release cadence. But today thankfully it's been succeeded by AMDVLK 2021.Q4.2 as the newest driver release...

https://www.phoronix.com/scan.php?page=news_item&px=AMDVLK-2021.Q4.2

[funforums]

A small question: is the figure still up to date? Isn't the default compiler now ACO, at least on those machine that support it? Or am I misunderstanding the role of ACO in this entire picture?

[-MacNuke-]

A small question: is the figure still up to date? Isn't the default compiler now ACO, at least on those machine that support it? Or am I misunderstanding the role of ACO in this entire picture?

AMDVLK is not RADV

[funforums]

AMDVLK is not RADV

Oh - now I see where I had misunderstood this. Thanks! ;-)

[aufkrawall]

https://github.com/GPUOpen-Drivers/AMDVLK/issues/231

Seems they don't want to support vkd3d-proton at all, just like raytracing. What a silly behavior this is of a multi billion dollar company to just be totally ignorant about valid use cases they don't deem worthy enough for incomprehensible reasons. It's the old AMD all over and over again, some things won't change...

[skeevy420]

https://github.com/GPUOpen-Drivers/AMDVLK/issues/231
Seems they don't want to support vkd3d-proton at all, just like raytracing. What a silly behavior this is of a multi billion dollar company to just be totally ignorant about valid use cases they don't deem worthy enough for incomprehensible reasons. It's the old AMD all over and over again, some things won't change...

I don't want to like that, but I'd like to know AMD's excuse here since Nvidia releases beta drivers damn near the same day that Khronos releases new Vulkan extensions and updates the specs

[user1]

I don't want to like that, but I'd like to know AMD's excuse here since Nvidia releases beta drivers damn near the same day that Khronos releases new Vulkan extensions and updates the specs

Idk. I'm starting to think they realized almost nobody uses this driver on Linux so they just stopped caring about these bugs and features specific to Linux gaming (you probably know Stadia uses AMDVLK, and probably with some other special use cases). On the other hand, lets imagine RADV didn't exist, would this still be their attitude?

[bridgman]

https://github.com/GPUOpen-Drivers/AMDVLK/issues/231
Seems they don't want to support vkd3d-proton at all, just like raytracing. What a silly behavior this is of a multi billion dollar company to just be totally ignorant about valid use cases they don't deem worthy enough for incomprehensible reasons. It's the old AMD all over and over again, some things won't change...

I don't want to like that, but I'd like to know AMD's excuse here since Nvidia releases beta drivers damn near the same day that Khronos releases new Vulkan extensions and updates the specs

I think the extension in question is "VK_VALVE_mutable_descriptor_type" - we are working with Valve on a standardized (VK_EXT...) version to replace the current one.

[bridgman]

Idk. I'm starting to think they realized almost nobody uses this driver on Linux so they just stopped caring about these bugs and features specific to Linux gaming (you probably know Stadia uses AMDVLK, and probably with some other special use cases). On the other hand, lets imagine RADV didn't exist, would this still be their attitude?

If RADV did not exist then I imagine we would have implemented the Valve extension in the short term to provide some support while we work with Khronos and Valve on a standardized replacement.

[aufkrawall]

And once that will have happened (after years of missing support), it will reveal all kinds of other issues all at once that could not have been tested/reported before due ot this. It's just bad practice. And then the next thing will happen, e.g. weird issues with XeSS or unsupported features with RDNA3 etc...