It s a great video that. There’s one from Apple Developer site too that’s worth a watch to see how it’s doing it in software / hardware. I think it may have been posted already.
1 Like
SoI just read new OTOY newsletter and octane X 2024 has 2x to 12x speedup for M3! This is similar to what we saw with optix over basic cuda!
They also have a new geometry pipeline that is faster and uses less memory for both cuda and metal hardware that has HWRT(m3).
Network rendering between cuda and metal so windows machines can be used as nodes.
And this is before meshlets(nanite) which is supposed to come later.
5 Likes
Nice… and pretty incredible! I wonder if those improvements will make it to the Octane Blender version?
It surely will. They show some nice speedups even for Nvidia in their forums. For now it’s an alpha of standalone app but the Mac OS version isn’t yet downloadable. I hope I can test soon.
These speedups make the m3 pro actually pretty useful for rendering (previews and stills) and having windows machine as a node is fantastic.
1 Like
You can either export/unpack orbx file from standalone or load it in blender as an asset or material. It’s not very intuitive or straightforward but works.
That said, blender octane is the same as standalone no? Just the free tier versions are usually a version behind the latest AFAIK.
Anyways, I’m used to render using RNDR so I always build scene and animation in blender and then export orbx and then upload that to farm for rendering. Cheap and fast… That said, if unreal can do the job I prefer that.
All these 3D apps starting to optimize for Apple Silicon has me getting visions of grandeur for the doubling of the MAX chip to the Ultra. 
The M2 Ultra scaled up from the M2 Max considerable better than from the M1 generation. I’m hoping scaling hits close to 100% percent for the M3 generation. If so the Studio is going to be a little monster that fits nicely under my monitor.
32 CPU cores
80 GPU cores
32 Neural Engine cores
256 GB Unified Memory
And the 6 Thunderbolt 4 ports will keep me happy with only 2TB of SSD space.
Price: Just under $7000 with tax included, if the top tier RAM and CPU prices stay the same as the M2
3 Likes
I’m curious: is this indeed the case? In other words: is the Octane free tier for macOS reasonably up to date, and will it only be a matter of time before the M3-optimized Octane is available for use in Blender?
And are there also two Octane versions for macOS, like the Windows versions: a custom Blender build and an add-on version?
Octane for mac os is called octane X. The free tier/app was just few days ago updated to latest stable version.
As for blender version I haven’t tried the free one as I have licensed one installed.
I don’t know anything about the add-on version, I haven’t tried that even on windows.
The M3 optimized octane is in alpha so I think for a while just as a standalone and later blender version and later the free tier. For me it doesn’t matter much as I would build scene in blender and then export to standalone always for a bit better memory usage and it’s also necessary for rendering on farm (rndr).
1 Like
We are back on track! Thank you Blender team! 
2 Likes
Thanks for the info.
Since Blender introduced the improved Principled BSDF shader and the AgX color profile in 4.x, I generally like the look of Cycles renderings a bit more than Octane results. Previously it was the other way round.
Of course this is a matter of personal preference (not aiming for one of those endless “this renderer versus that one” discussions 
), but Cycles is more convenient than an external / add-on renderer, and is constantly being optimized for macOS.
4 Likes
So far I find octane much better with memory management and also faster at least on mac.
1 Like
Uhm, somehow I completely missed BCON 2023 …
Nice contributions :
https://www.youtube.com/playlist?list=PLa1F2ddGya_84ye1JAvIGh0Yp7utEKZI_
1 Like
Dumb question: With all the jazz about unified memory usage, why does Blender’s memory consumption double when rendering on GPU ? Rendering on CPU uses nearly half the memory, shouldn’t it be the same ?
2 Likes
“Whisky: Wine Supercharged with the Power of Apple’s Game Porting Toolkit”
Maybe this could be a doorway to running certain Windows-only apps on macOS, like MagicaCSG, without having to go through the hassle of setting up Wine, virtualization or a dual-boot setup.
4 Likes
I think it is relevant for this topic: here is Cycles on an iPad
it makes one wish the Blender Foundation and Apple would talk about those AppStore rules and somehow allow full Blender there.
5 Likes
It’s not so much that you’re filtering but more being relative to the version. Every system took a hit in points for 4.0 bench, Apple, Nvidia, AMD. Even if the M3 chips saw gains for HWRT it is still lower than if HWRT would have happened in 3.6. i.e. M2 Ultra 3.6 score is higher than the M2 Ultra 4.0 and same goes for Nvidia’s cards.
And it’s not like anyone PC or Mac will be using all the advantages that come with 4.0/4.1 and then render in 3.6, cause those two version aren’t compatible with each other. So may as well filter the bench to 4.0 and go from there.
2 Likes
HWRT iPads are right around the corner! What a crazy time to be alive.
I too would love an iPad Blender version. As you say, I hope some day they can get it on the App Store.
5 Likes
Still No: I want EVERYBODY to see Apple on the front page without filtering. It’s not about me, it’s about Apples perception to the 3D audience
1 Like
I get what you’re going for but do you compare Geekbench 4 numbers against Geekbench 6 numbers when the two have different results? No, the industry will reset the chart and only list CPUs for the current version.
You’re not “filtering” when you choose the proper test bench.
With the new Opendata numbers for 4.0 even Nvidia isn’t as fast as Nvidia. 
Hi, Phoronix publish a benchmark test Windows 11:Linux Distributions but Apple is missing.
They test some render engines, VRay, Luxrender, Appleseed and Blender 4.0.
How does the M2/3 systens compare to the AMD Pro CPU?
Cheers, mib
1 Like