I think that will be there for some time to come with blender when one uses Cuda.
But I have to say the Mac mini moonlight access to my win10 machine is fantastic.
That’s a nice way to work when you do your edits on a silent machine and then for rendering fire up the pc.
Can’t speak about anything MacOS specific, but Lumen does software tracing in case there’s no RTX present. Works on non RTX hardware like a GTX 1060.
Probably not the best place to bring this up, but since we’re somewhat on the subject…
Has anyone checked out Godot’s approach to GI? It’s fairly impressive, and doesn’t require any specialized APIs to run well.
Do you know people who work at these companies? Because I do, TD’s at that, and I have had extensive conversations with them about this specific subject. Do they know about GPU renderers? Of course! Do they do some stuff with them here and there? Sure! Are those renderers the ones that end up being used on what you see on the big screen? Nope!
P.S.
I didn’t even mention the fact that Blender is still considered a joke to those guys. Sorry but that’s the truth.
It unfortunate that at the moment neither Lumen nor Nanite are enabled for MacOS. I suspect if/when they eventually are, they will require Navi2 at minimum.
…k?
Does this matter ? It is not only a software that is defining a studio production pipeline.
It wants to be.
Look, I’m here, I’m using Blender so I’m a fan. Just wanted to point out to the other guy that GPU rendering has received a crazy amount of hype, and people who simply don’t know any better see the current lack of Metal support as really bad when it really isn’t such a big deal. I’ll take a faster CPU Cycles X and an OIDN 1.4 over a Metal port any time! (although both would be cool)
I guess it all depends.
From my experience I will always rather use GPU over CPU since for my work it is faster and RAM limits are not an issue. However obviously I do not work on films but product / architecture work - so requirements are different.
Ok, but even that depends on your hardware. I am seeing people spending several thousand dollars to pick up a couple of 3080’s, whereas the same amount would give them a powerful 32-core Threadripper completely unencumbered by memory limitations, and quite possibly just as fast if not faster for certain things (GPU’s are notoriously very poor at rendering VDB’s for instance).
AMD Ryzen Threadripper 3970X 32-Core Processor CPU Windows bmw27 2.82 48.6618 (render time)
GeForce RTX 3080 OPTIX Windows bmw27 2.90 11.862(render time)
Nah, FIVE times faster. GPU acceleration is just insane.
You might argue that there are some things optic can’t render well. Then Cuda is 2.5X faster (find it in open data)
Do some research before you talk.
You sure you work in studio or whatever you just said?
Can we talk with facts here?
On the majority who use an R7 R9 or i7 i9 processor user, GPU acceleration render is only way to render fast unless you use cloud renderer.
If you don’t know how much metal improve an application, go do some research.
All those ppl who crown themselves as pro are well as noob. Really, pro won’t just go Hi I’m a pro bc they know they will never be a pro and there is always things to learn.
Maybe learn how to read before you comment…huh?
Guys, please keep the discussion friendly, thanks. 
Sure, and also let’s stop this idiotic bickering over Windows/MacOS etc. and get back to the whole ARM Blender thing.
Ive noticed when rendering larger scenes with eevee that ive never had any memory problems, even with memory-intensive scenes. I looked into it, blender was using 10gb of vram, the concept where the SOC has full access to the memory (instead of a small part of it like it is on intel cpus) is pretty usefull.
It comes at the cost of slower gpu memory, but I think that tradeoff is worth it. Ik some 3d programs can eat into the users cpu memory as well, but why isn’t it used more often
This is all about hardware differences and optimizing it
One of the main reason why apples soc is so powerful even with such a lot wattage is how it can deal with data internally
A dedicated GPU needs to send data via the pci bus
With apples soc those bus connections which also can be bottle necks are not present
This also enables Apple to use ram compression very effectively
Apples cameras are not the best form a hard ware stand point but their trick is what they do with signal processing
Do you have the 16gb version? I have tested some scenes in my studio partner’s 8gb MBpro and while the viewport performance is astonishing, I’ve suffered several crashes when rendering medium to large scenes. I’m thinking to grab one as secondary computer, but I need to know that I will be able to render on it when needed.
I do have the 16gb version! If you have a scene in mind I can test it. I haven’t suffered a ram-related crash yet.
I can’t share this scenes ATM, but eventually will PM you if needed. Anyways I’ll try to get a 16gb one if I decide to buy one. Thank you!