Is it easily accessible/useable for devs? Don’t know, but I guess Yes. Feel free to fill me in here.
Are there developers who know Blender and Metal? I know of two, and I am not a coder. There might be many more.
Does the Blender Foundation have the money to fund development on all three platforms? Yes, more than ever.
Are there artists who use Macs? Yes.
Would those artists benefit from Metal optimised apps? Yes, very much.
Is Blender a commercial company, forced to prioritise OS segments purely by user numbers? No. Never was, and I hope it stays this way.
Also, I don’t see how Windows, or CUDA, is more open, or less proprietary, than Apple/Metal. Or how Nvidia is less interested in selling their hardware and having software run on their hardware only. But again, maybe I am missing something here.
Blender is relying on proprietary tech anyway. It is running great on Windows, and using CUDA extensively, and that is totally fine, because artists benefit from it. But so they would, too, from proper support for Apple hardware and Metal.
Nvidia is interested enough in Blender that they are a Patron of the development fund. I assume they understand that by having really good Blender support for their GPUs, they may sell more or better GPUs.
To the Foundation, it shouldn’t matter why anyone is a Blender patron, or not. The app has been around long enough without big tech patrons, and support was equally great for all major platforms. It would be lame if they now made their commitment dependent on how much a particular company is paying, or not.
Cycles started out with CUDA and CPU support initially. CUDA was pretty much the only choice at that time. Both additional backends which were added later, meaning OpenCL and OptiX have literally been sponsored by AMD or Nvidia to support their hardware. Those are very specialized solutions, just like Metal for Cycles would be. From my point of view there is not really a difference, except that it seems less likely Apple would sponsor it. This kind of sponsorship would be the easiest way to get Metal for Cycles.
I see. Can’t blame them for taking things for free. Also, Blender is still working pretty great with (new) Macs, courtesy by M1. It’s just that platform support is off balance, and I’d rather see the BF restore it than wait for someone else to sponsor it, or not.
I’d rather have support and optimisation for Metal and Apple Silicone M1 chips than Blender Video Sequence Editor updates. Final Cut Pro is already extremely powerful and not even that expensive.
Also, we need to update the icon for MacOS to fit in with the new design language.
I’m currently using the Blender Alt 3 one from https://macosicons.com but it would be cool if it was updated officially for macOS build.
I understand the situation you and many other Apple enthusiasts are in and I don’t like the situation either! Metal support for Cycles would be fantastic, but it requires a significant investment which would need to be justified!
To give you a little bit more context, Apple has been holding back Blender for quite a while! Contrary to Windows and Linux, the hardware vendors can not just release new device drivers. This led to a situation where the OpenGL version on MacOS got quite outdated. The Blender developers decided to stick to that version, even though it was clearly holding back Windows and Linux.
The situation has changed a lot and keeping up with Apple became more difficult. When you have such a diverse audience, it is better to take a conservative development approach in my opinion. It should be preferable to have as many people on board as possible, while risking to not always support the newest and shiniest features.
The shiny features tend to be very expensive, but as they are a selling point for hardware vendors, they quite often sponsor them (not just for Blender, but often in video games too). As Apple is different in so many ways, they don’t seem to be doing that.
While not terribly important, I am actually curious what kind of investment it would be. It would be great if someone involved chimed in into our discussion here for some clarification.
Taking one step back I realise that as long as Blender works, and as long as the website isn’t saying “we are ending support for macOS” there is no real reason to worry. Like You said, the situation is complicated and there appears a lot to be going on. But we can just assume that they will figure it out.
We should all write to apple asking them to fund blender development. Maybe if enough of us write to them they will give blender a grant or a dedicated developer like many major tech companies have.
Otherwise we can always set up a community fund for Metal / Apple Silicone optimisation.
How much do you think it would cost to do this?
It’s been clear for several years that OpenGL has no future - it’s a zombie, not just in macOS, but in Windows/Linux as well. The situation is somewhat similar to Flash. Why the hell would Apple invest in OpenGL software?
The OpenGL situation I described dates back quite a while. If I remember correctly, it was like that even before they introduced their first Metal version (could be wrong about it though!).
The hardware vendors are constantly updating their drivers, but on MacOS that isn’t possible as Apple doesn’t allow it. Even if AMD or Nvidia likely were ready to updated the drivers, Apple simply blocked them.
This is what the Blender Development Fund is for. I joined it, and have been a supporter for over a year now. I am counting on the fact that everybody will have an advantage out of my funding, and everybody else’s. This includes Metal support.
While I understand the urge to lobby for his own agenda, this IMO contradicts Blenders mantra for creative freedom, and openness in general. This sounds like just going where the money and the loudest voices are.
Nothing wrong with this per se, it is what most commercially oriented companies would do. But I’d rather have the BF, which is non-commercial, stay somehow independent and distribute their resources at their own discretion, providing equally good support for all platforms.
One of the great things about open source software is how it adds value to older computers. The 2013 Mac Pro could be a great machine for Blender with Metal support. Even if it took a few years, it would still be worth it for Cycles to add Metal to breathe life into the older Macs.
I took a pass at learning Metal last year. There is definitely a high learning curve for graphics programming and frameworks. I’m hoping to return to to it at some point in the next few months. If anyone would be interested in working through a book on Metal, message me.
I am a MacBlender enthusiast and I also hate this situation. I’ve read all the posts in this thread and agree/support all, or in part every one of them. I believe that after all these years (2? 3?) There seems to be a growing desire to drop the expectation of Apple’s participation, Don’t wait for Blender to do it (clearly neither has the desire/capability to do this) and do what is right by creating funding or at least an organized effort to get what we want for Cycles. If there is a need and capability I want to contribute processing power by internet to accomplish the task, and any $ I that can also.
Judging by the conversations here, it might be acceptable to all to:
Develop Vulkan/MoltenVK for the near term to use while
a Metal version of cycles is developed.
This way we all get to see what GPU rendering looks like with cycles. In the meanwhile, Luxcore and Prorender are my renderers of choice.
How do I donate computational power to the cause? iMac Pro 10 core Vega 64 with 64GB of ram.
If I understand correctly, Apple provided some code to create M1 native Blender.
A version prepared by Stefan Werner has been available for almost two months:
It is incomprehensible why there is still no official version released by Blender Institute.
a version that may be incomplete or buggy??? That might work? With serious bugs?
this is in no way ready to be implemented .
Getting a huge software like blender to work well under another architecture is a huge task and you can’t expect it to be done in a few months.
Especially when the hardware you are talking about isn’t widespread. I don’t think professional workstations using apple silicon are out yet and i don’t even think Apple has given the blender developers any hardware to work with.
There is a download section called ‘Experimental’ and there were published Blender versions less stable than Stefan Werner’s build.
It may sounds weird for you but I use M1 Mac as my main ‘workstation’ now (despite I have much more powerful machine side by side) for ‘professional’ jobs and it works quite nice with non-native (final, alpha, beta) and native (SW experimental) Blender version.
My point still stands. You can’t take a piece of code that “just works” and integrate it like that in a big project. Doing that too quickly is the best way to introduce more bugs and to create impossible situations for the devs.
Development takes time and money. Especially when it’s related to something as low level as a new architecture.
Stefan’s AMR64 build is actually pretty great. There are some stupid bugs (like, opening preferences from a full screen Blender crashes it), but mostly it is stable and actually a joy to use on an M1 MacBook Air. Which is the reason why the HP Z tower underneath the desk has been collecting some dust over the last two months.