Mesa Vulkan Drivers Reach An Inflection Point: Idea Raised To Be More Like Gallium3D

[madinside]

Obviously, given the naming history of Vulkan, Rift or Subduction or something other tectonics related will be the name of that. Bonus points if they get something metal related in there.

[Sethox]

Great, but please, don't call that Vulkium3D.
Enough with the terrible names.

Eruption3D

[r1348]

Vulkanium3D
Vulkulum3D
Vulkanogaster3D

Don't care really, as long as it has 3D in the name, because the '90s are never really over.

[Party0445]

Vulkanium3D
Don't care really, as long as it has 3D in the name, because the '90s are never really over.

Long live the 90s.

[thxcv]

Love all the name suggestions I'll add Vulkrum3d to the list.

OT I hoped someone makes DukeVulkan3D with RT

[Daktyl198]

When Vulkan started out drivers were much more thin and low-level but with time the Vulkan API feature set has built up.

Vulkan is already ruined, and is only going to get worse with time. There's a reason developers preferred DX11 to OpenGL, and it's the same reason they prefer DX12 to Vulkan. Extensions ruin the entire point of a stable, developer-friendly API. It's so bad that apparently, Vulkan is now considered high-level enough to have a RUNTIME beneath it. Like wtf? The whole point of Vulkan was that it was at or below the Gallium runtime level.

[kiffmet]

Daktyl198 as if D3D12 having different "Shader Models" and feature tiers is any different. There, developers also have to probe for the availability of features and then act according to what's actually available.

Vulkan extensions may be a bit more tedious, but that gets relativized by the most commonly used stuff getting promoted to new Vulkan core versions.

And of course there'll always be some runtime steps involved in both APIs, i.e. for common, device agnostic data handling/formatting/converting and housekeeping tasks.

[ehansin]

I was thinking Vallium3D - of course to some that might feel like a downer!

[mbriar]

Vulkan is already ruined, and is only going to get worse with time. There's a reason developers preferred DX11 to OpenGL, and it's the same reason they prefer DX12 to Vulkan. Extensions ruin the entire point of a stable, developer-friendly API. It's so bad that apparently, Vulkan is now considered high-level enough to have a RUNTIME beneath it. Like wtf? The whole point of Vulkan was that it was at or below the Gallium runtime level.

Only that D3D12 has a runtime too, which games even ship their own version of to make use of new D3D12 features on older versions of Windows 10 (D3D12Core.dll), and has tons of optional features, I think we are at version 19 of the D3D12_FEATURE_DATA_D3D12_OPTIONS struct now. The shader model also got a bunch of extensions and is at version 6.8 now. It's pretty much the same thing as vulkan extensions.

[Developer12]

Vulkan is already ruined, and is only going to get worse with time. There's a reason developers preferred DX11 to OpenGL, and it's the same reason they prefer DX12 to Vulkan. Extensions ruin the entire point of a stable, developer-friendly API. It's so bad that apparently, Vulkan is now considered high-level enough to have a RUNTIME beneath it. Like wtf? The whole point of Vulkan was that it was at or below the Gallium runtime level.

You do realize the "runtime" in question is the entire userspace component of MESA, right? that's always been there and hasn't become appreciably larger.

Having a common vulkan frontend would solve the biggest problem of extensions: having a uniform feature set. If extensions were implemented in a MESA vulkan frontent then every driver would automatically support them. Potentially even proprietary drivers on windows. (MESA is MIT-licenced after all)