The Still Very Early State Of Vulkan For Blender - No Active Developers Working On It

[phoronix]

Phoronix: The Still Very Early State Of Vulkan For Blender - No Active Developers Working On It

While there has been talk and plans for Vulkan API support within Blender, currently there are no active developers working on it and much work remains before it would be ready for end-users...

https://www.phoronix.com/news/Blender-Vulkan-September-2022

[DMJC]

Vulkan is a very big, very expensive distraction for projects. Yes it has it's benefits, but the cost of rewriting the entire graphics engine is enough to actually kill a lot of projects that would be better off investing the time in features/supporting issues users are already dealing with. Even for VR support Vulkan is often unnecessary and used more as an excuse not to implement VR support e.g: "We'll get around to VR once the Vulkan rewrite is done", 10 years later... no Vulkan rewrite and no VR support, users lose interest and the project momentum dies.

[-MacNuke-]

Vulkan is a very big, very expensive distraction for projects. Yes it has it's benefits, but the cost of rewriting the entire graphics engine is enough to actually kill a lot of projects that would be better off investing the time in features/supporting issues users are already dealing with. Even for VR support Vulkan is often unnecessary and used more as an excuse not to implement VR support e.g: "We'll get around to VR once the Vulkan rewrite is done", 10 years later... no Vulkan rewrite and no VR support, users lose interest and the project momentum dies.

Well. You need to do it even without Vulkan. Apple deprecated OpenGL long time ago and it breaks every other version. Fast forward a couple of years and you will want to use something like Zink for OpenGL stuff because drivers for OpenGL will be absolutely horrible everywhere.

[holunder]

Well. You need to do it even without Vulkan. Apple deprecated OpenGL long time ago and it breaks every other version. Fast forward a couple of years and you will want to use something like Zink for OpenGL stuff because drivers for OpenGL will be absolutely horrible everywhere.

OpenGL support will be around forever, at least on Linux. With or without translation layers.

[c117152]

Well. You need to do it even without Vulkan. Apple deprecated OpenGL long time ago and it breaks every other version. Fast forward a couple of years and you will want to use something like Zink for OpenGL stuff because drivers for OpenGL will be absolutely horrible everywhere.

This is worded too misleadingly. Apple only deprecated native OS-level OpenGL driver-side support since its new APIs are low-level enough to efficiently shim OpenGL around them. People and software vendors will just deploy redistributables like directx and .net that will expose OpenGL APIs and those will continue to work and get new extensions for decades to come.

You'll never NEED to support Vulkan over OpenGL. Worse case scenario you'll need to recompile after changing a few #include's to use helper libraries.

[user1]

Fast forward a couple of years and you will want to use something like Zink for OpenGL stuff because drivers for OpenGL will be absolutely horrible everywhere.

I don't see why would mature and well optimized OpenGL drivers like RadeonSi suddenly become horrible, even in a few years.

People here on Phoronix and elsewhere have this really weird obsessive hype about Zink replacing everything, including mature and well optimized OpenGL drivers. Like why?
I don't understand this bizarre excitement about Zink replacing something that is mature and doesn't even have a reason to become bad and/or obsolete in the future.

I understand that Zink has a very useful purpose of removing the need to write a new OpenGL driver if there's a completely new hardware (like Apple Silicon), but replacing well established and mature OpenGL drivers? This is just dumb.

[Quackdoc]

this is disappointing, I remember quite a few of my friends were disappointed about the CL removal, but were fine as long as vulkan support which was announced was coming and rocm is simply not a viable alternative for many folk considering it's... sad state of affairs.

though at the very least it sounds like they are making progress in the preparatory work. though I find it hard to believe that the blender foundation cannot afford to hire a few devs for it. its not like blender is a small project in fact, if their sponsor pricing and sponsor page is to be believed I would think they would have quite enough to hire even a few high profile vulkan developers .

I understand that Zink has a very useful purpose of removing the need to write a new OpenGL driver if there's a completely new hardware (like Apple Silicon), but replacing well established and mature OpenGL drivers? this is just dumb.

why is it dumb? some of the mesa devs have even pondered the issue, thats not to say that it is the direction they want or are going to go, but at the very least they seem to be giving it more thought than you anyways., I think it was Mr. Ekstrand specifically. nor do I agree with macnuke's sentiment about drivings going outdated that fast, but at the very least it will happen.

[user1]

why is it dumb? some of the mesa devs have even pondered the issue, thats not to say that it is the direction they want or are going to go, but at the very least they seem to be giving it more thought than you anyways., I think it was Mr. Ekstrand specifically. nor do I agree with macnuke's sentiment about drivings going outdated that fast, but at the very least it will happen.

It's dumb because Zink doesn't give a clear advantage in performance over RadeonSi. I think when Zink will mature even more, it will at most slightly outperform RadeonSi in some benchmarks, but in others it will still be slower, just like DXVK, which is already 4 years old. For example, on my RX 580 Radv + DXVK slightly outperform some dx games compared to native dx9/11 drivers, but in others it's a bit slower. So I don't see why would it be any different with Zink.

Btw, I've already said it in this forum some time ago, but if we speak solely about AMD hardware, Zink replacing RadeonSi is an impossible scenario in the current state of affairs. It seems people easily forget that the development of Radv is not officially supported by AMD, while the development of RadeonSi is. For example, Radv devs don't have early access to new hardware documentation and they sometimes rely on the work being done in RadeonSi. Btw, Radv itself actually reuses RadeonSi's winsys component.
That's why unless AMD will officially start supporting the development of Radv, it's impossible to think that Zink will ever replace RadeonSi.

P.S Also, remember that it took RadeonSi like at least 8 years to get to this point of stellar optimization like it is today. And the optimization is still ongoing (currently for workstation apps). So lets say it's about to be replaced by Zink. Then what was the point of all this multi year optimization effort just for it to be replaced when it reached peak optimization? So I say replacing RadeonSi is highly unlikely at least in this decade.

[-MacNuke-]

I don't see why would mature and well optimized OpenGL drivers like RadeonSi suddenly become horrible, even in a few years.

Just experience... There is still someone needed to modify radeonsi for newer GPUs. Same for every other vendor. This work will go the same way as Xorg. Some day noone will care. Most users won't be using OpenGL anymore. Gaming? Wine+DXVK/VKD3D. Apps and DEs? Moving to Vulkan as we speak. Old games? Probably won't work in the native version in a few years for other reasons.

What's not getting used won't be tested and people will... yes... stop caring.

[user1]

Just experience... There is still someone needed to modify radeonsi for newer GPUs. Same for every other vendor. This work will go the same way as Xorg. Some day noone will care. Most users won't be using OpenGL anymore. Gaming? Wine+DXVK/VKD3D. Apps and DEs? Moving to Vulkan as we speak. Old games? Probably won't work in the native version in a few years for other reasons.

What's not getting used won't be tested and people will... yes... stop caring.

Give me one example of a DE that is switching to Vulkan as we speak..
Also, there are still a lot of workstation apps that use OpenGL and that doesn't seem to change at least in the near future.
And btw, this might surprise you, but there are still some very new indie games that use OpenGL exclusively and they're still in early access.