Yeah you’re right sorry about that!
Cool, thanks for digging that out.
Just a bit of peanut gallery musing about the seeming reticence to allocate resources towards adding Metal support to Cycles:
Supporting Metal in Cycles isn’t something that just requires a single sprint to add the functionality and then gets to work forever.
The big takeaway from the fact that we’re having this discussion is that Apple tends to make unilateral platform API decisions that force developers who want to use their hardware to scramble to keep up. In theory this is generally to make their hardware as performant as possible, but it’s also often done purely to support other initiatives and corporate food-fights that they might be presently engaged in (ie: nvidia or adobe know all about this).
If BF adds a feature to Cycles, they are sort of obligated to support it. Otherwise, when Apple deprecates Metal for whatever replaces it so that they can support handoff between their computers and their phones or whatever, suddenly this board lights up with people talking shit about how BF is dropping the ball on Mac support.
In light of this, I can certainly understand BF wanting the world’s wealthiest tech company to spend their own money on supporting Cycles development.
I am an Apple user. I do professional video work. I like their hardware for reasons stated above. I would love to believe that users like me can crowd onto Apple’s user engagement platforms and apply pressure for them to fund initiatives like this, but we need to be a little realistic.
Apple doesn’t make most of their money from users like us. They are a lifestyle brand. They make their best money from phones and tablets. They are actively engaged in trying to open other lifestyle brand revenue channels in subscription platforms like Apple Music and TV. This is where their priorities are. It’s unfortunate, but people here who want to cope with this by laying into the BF devs are doing Apple’s dirty work. Please be better.
This sums it up pretty well.
This is message from a year and a half ago and the one big nothing happened in terms of Metal Cycles. This is nonsense.
I am so angry and frustrated with the BF that I am tempted to tear the tags off my pillows and mattress (sorry if only us in the USA understand this).
When the development fund was announced I knew it would come to this. Since that time we’ve heard a lot about about making the fund grow so MacOS will be listened to. This attitude happens in every human organization. The bigger they get the less responsive to the little guy and the more attention is given to the big sponsors; the main goal is to grow the org., Not the support as much of the purpose on which it was founded.
The fund webpage says that now there is 50% more money than when this thread started and Blender Market consistently gives more than US $5,000 each month to the development fund and we are still hearing that the easiest solution is to give more to the fund? Meanwhile not a whisper on any initiatives we are interested in. The BF can’t convince me that equal treatment of all is their concern.
Necessary or not money has gone to administration and support (see above post); and not even a word about any meaningful initiatives for Mac users. The answer to the title of this thread is NO. There is no support for Mac Metal presumably until all the Nvidia and Windows users and large studios have run low on ideas. Donate to the fund, but don’t hold your breath for anything Mac related for quite awhile.
It still seems that an organized Macuser crowd fund is the only way to get Metal on a reasonable timeline. Two (?) years is long enough. Otherwise this thread is not a debate, rather a forum for venting (also a good thing).
Everyone wants some part of the blender development fund to be allocated to some features they want.
Seriously! Probably half or more of the threads regarding a possible improvement to blender use the same argument you do. How do you choose?
Why do the bf keeps telling people to contribute? Because having paid developers is really expensive, you have to pay for the hardware they use, to pay their salary, to pay taxes on their salary, and everything else.
Have you looked at how the BF hires? It is mostly people that contributed already to some part of blender for free. I don’t see many patches coming from a metal specialist willing to help.
So who do you give your money to if you are the bf? Someone who spends most of their free time working hard to improve blender, someone you know well and that understands what blender is about? Or a stranger that hasn’t contributed a single patch.
Obviously I’m exaggerating a little bit in my example but that is a real issue. I’m certain if there was a serious contributer working on a metal implementation of cycles the bf would ay least consider hiring him.
Besides blender shouldn’t have to run after apple every time they make a unilateral decision.
When they drop metal for something else what happens to all the work that has been put in metal support?
I started to follow this thread about a year ago. What I see there is a lot of attention to this topic since the M1 what I’m really happy for!
Unilateral decisions from Apple or not… this is a huge issue. A lot of creative using mac because it’s more reliable for design workflow than anything else. I won’t use windows again and I can’t use Linux because it lack of tools what I need. I’ll follow this topic with interest and hope. That’s all I can do now!
I’ll drink to that! Salute!
Haha, that’s it. Although, I’ve always had a backdoor to Windows. And every time I touch it I am swiftly reminded why I stayed away: too much friction in so many places.
I think the notion of other platforms users getting more from the shared budget is wrong. It’s just that Intel, AMD & Nvidia are interested in having Blender work well on their stuff, which is why they donate developers, money, hardware, libraries and patches, while Apple fundamentally does not care.
People seem to buy their hardware regardless of compatibility, I suppose
I think you’re right. I also think that more people are familiar with the graphics APIs of AMD/Nvidia. There seems to be a small talent pool for Metal graphics programmers in general. Blender is one of a few open source projects that doesn’t have the human-power to port to Metal.
I’m not too sure about this. If I remember correctly the App store is the worlds largest gaming platform.
I might be wrong. Metal isn’t the only graphics API for iPhone and many people use SpriteKit, which is built on top of Metal. I looked at some freelance hiring sites and I wasn’t able to find many Metal experts for hire, but maybe I wasn’t looking in the right place.
iOS Developer here:
Back in 2009 when I created my first app, a 3D strategy game, I wrote all the OpenGL code myself. I.e. load the arrays of x-y-z points from files, put everything into interleaved arrays on the GPU, and then call methods like:
glDrawArrays(GL_TRIANGLES, 0, _vertexCount);
Later (2011), I also built an early AR app using OpenGL and C++ which used similar library calls as above.
Nowadays everyone is writing 3D games using abstraction layers that are above the actual graphics library calls. Apples own SceneKit, RealityKit (SpriteKit is 2D) or 3rd Party game engines like Unity, Unreal, MonkeyX etc. In some cases companies have their own engines (like Blizzard with WoW or Paradox with their Clausewitz engine) that have adaptors for different graphics libraries.
I have used Metal calls here and there, but it’s really a niche that few iOS/macOS developers I know have ever touched. The key people working with it are the graphics engineers at Unity/Unreal who build the adaptors to DirectX, OpenGL, Metal etc.
Metal is more comparable to DirectX, if you critize Apple for making uniform platform decisions you should also critizise MS for making apps use DirectX instead of OpenGL.
The key difference is that on Windows, the GPU vendors decide which drivers they want to provide. They have the option to provide OpenGL and/or Vulkan drivers, even though Microsoft has another API. When it comes to the MacOS, that flexibility never existed.
Everything has it’s ups and downs. I’ve never had to update any driver on macOS, programs from the App Store work without me having to worry about driver version or anything.
I do have a windows partition I use for gaming, but I rarely boot it up anymore as there is just so much time wasted on updating system components, drivers etc.
By tying the GPU support into the core OS Apple has more control over the process and can make it easier for users. Or introduce a completely new chip architecture on fanless MacBook airs thats blowing away desktop computing power with 10x the wattage
PS: I’m happy we can actually still start apps on MacOS that were not downloaded via the Apple AppStore. For a while it was looking like we’ll eventually get to a point where all Apps on macOS have to be approved in a similar process we have on iOS.
That’s nice, but the I still don’t agree with your comparison.
I was just wondering what happened to OP while we’e been in this lively discussion for 2 years.
Hi All,
Yeah, I started this whole thread a while back. I actually got pretty deep into Cycles internals back then and did a bunch of Metal prototyping (outside of Blender) to prove that various techniques would work. But I got completely deflected by work, both parents becoming ill and needing full-time care, and other things.
I still believe that Metal would be a beautiful implementation platform for Cycles, Eevee, and the UI ((certainly easier to develop against than Vulkan based on my experience!). There’s just no way I could tackle this as a solo project without funding or time from Blender developers until my personal circumstances change.
I still love Blender and use it all the time on Mac and elsewhere.
As much as I wish I had advanced GPU support on Mac, I can’t fault the team for focusing their efforts elsewhere. Product management is difficult and rarely makes everyone happy.
Support for Vulkan, Metal, and maybe DirectX offer much more than just making Cycles faster (e.g. real-time dynamic tessellation to support things like displacement). This stuff needs to be figured out and put on the overall Blender roadmap. None of this is easy and the team is rightly taking the time to think it all through.
I’m willing to wait until they get it right on at least Vulkan at which point it will be potentially easier to think through porting to others (or just using MoltenVK).
~chuck