3D Rendering on Apple Silicon, CPU&GPU

[terminator-jq]

And remember that the default benchmark is using 128x128 blocks which isn’t optimal for M1.

The developer Panos Zompolas mentions it here.


The scores from the Moana Island scene rendering with Redshift is very promising.

2 x 1080 Ti = 77m
2 x 2080 Ti = 34m:17s
1 x 3090 = 21m:45s
2 x 3090 = 12m:44s
M1 Max (64GB) = 28m:27s

That’s more like it!

I wish these YouTube reviewers would expand to Redshift for their rendering benchmarks since it seems to be giving performance much closer to what the M1 Max is actually capable of. Looks like M1 Max would land somewhere in 3070-3080 territory which is much more reasonable.

 

[sunny5]

 

[ader42]

Could those benchmarks be skewed by poor choice of bucket/tile size? I saw on a technical blender thread that render time could be halved by increasing bucket size…

 

[leman]

Could those benchmarks be skewed by poor choice of bucket/tile size? I saw on a technical blender thread that render time could be halved by increasing bucket size…

These benchmarks are skewed by the fact that the M1 implementation is still barebones. It is a very complex software and the tenderer has been designed for Nvidia GPUs. Maybe Apple can make it faster in M1, maybe they can’t, who knows. But it should be telling that M1 performs much better in some other rendering suites (in relative terms). Should give you an idea of how much untapped potential there is.

 

[ader42]

Does anyone have any input on how OctaneRender and OctaneX are nowadays?

My MBP will be arriving in a couple of weeks and will take advantage of the free 1 year offer from Otoy if it’s a decent option for a render engine?

Am also interested in the local network rendering feature of Octane seeing as I will have at least one M1 sat doing nothing else. Is this for multiple frames or multi-device contribution to a single image frame?

 

[ader42]

Okay, I looked into it, looks like not all Octane versions are natively optimised yet - so Redshift is likely still the best option for us.

 

[Xiao_Xi]

What 3D software shines on Apple hardware and could be used as a demo for the Mac Pro?

 

[terminator-jq]

What 3D software shines on Apple hardware and could be used as a demo for the Mac Pro?

So far Blender, Houdini and Cinema 4D are the only 3D applications natively built to run on Apple Silicon. It sounds like Autodesk is slowly moving over some of their stuff. The Modo developers have showed interest but it doesn’t seem like that’s high in their priorities. Zbrush is now part of Maxon so with C4D being the very first Apple Silicon 3D program, that should be a good sign that the people at Maxon are serious about getting their products moved over.

 

[vladi]

Does anyone have any input on how OctaneRender and OctaneX are nowadays?

My MBP will be arriving in a couple of weeks and will take advantage of the free 1 year offer from Otoy if it’s a decent option for a render engine?

Am also interested in the local network rendering feature of Octane seeing as I will have at least one M1 sat doing nothing else. Is this for multiple frames or multi-device contribution to a single image frame?

Octane is fine. You should get it. I find it be the most user friendly GPU renderer out there. Kind of like Keyshot but on steroids.

 

[jujoje]

What seemed to be a more reasonable review than most of the usual YouTube clickbait bs, with a couple of redshift benchmarks:

Mac Studio and Studio Display — Prolost

prolost.com

The take away point is that:

"My sense is that those of us who make the creative tools that can truly take advantage of the M1 Ultra’s power still have work to do to occupy every corner of performance possible."

And as side note, from the Redshift Forums:

"I saw a guy testing (not the official scene but a scene of his own) on a M1 Max (maxed out with 64gb) laptop and it went from 8:38m to 6m something with the big bucket size, so if that was not just a one of a kind scene, a big performance gain could be in there."

So sounds like there's a pretty easy gain from just changing the bucket size.

I suspect that the GPUs won't be significantly far off Apple's claims, but that most software isn't going to effectively utilise them yet (Redshift and C4D probably being the closest).

Finally, just to add a bit of salt; credit goes to the Verge for using Tombraider and Premier to 'prove' that that the Ultra is slower than a 3090. Good job guys!

 

[jujoje]

Also just looked at the profile of the author of that review:

"Maschwitz designed the Magic Bullet color grading system for Red Giant, and is now Chief Creative Officer of MAXON, where he continues to create powerful, intuitive tools for animators, filmmakers, and motion designers."

I guess he should probably know what he's talking about

 

[mi7chy]

Redshift isn't optimized either. Just wait.

 

[JimmyjamesEU]

What seemed to be a more reasonable review than most of the usual YouTube clickbait bs, with a couple of redshift benchmarks:

Mac Studio and Studio Display — Prolost

prolost.com

The take away point is that:

"My sense is that those of us who make the creative tools that can truly take advantage of the M1 Ultra’s power still have work to do to occupy every corner of performance possible."

And as side note, from the Redshift Forums:

"I saw a guy testing (not the official scene but a scene of his own) on a M1 Max (maxed out with 64gb) laptop and it went from 8:38m to 6m something with the big bucket size, so if that was not just a one of a kind scene, a big performance gain could be in there."

So sounds like there's a pretty easy gain from just changing the bucket size.

I suspect that the GPUs won't be significantly far off Apple's claims, but that most software isn't going to effectively utilise them yet (Redshift and C4D probably being the closest).

Finally, just to add a bit of salt; credit goes to the Verge for using Tombraider and Premier to 'prove' that that the Ultra is slower than a 3090. Good job guys!

Yeah, the verge is....

Some of their quotes about their staff refusing to use anything other than Premiere no matter what are worrying. It's fine to have a preference, but the review is (to my understanding) about the machine and it's capabilities. Are they really thinking its justifiable to produce endless reviews of their workflow. I don't care what their workflow is. I want to know what the machine is capable under a variety of uses.

 

[ader42]

I came across this ZBrush youtube video again, and it got me excited (again) for my MBP which is en-route - and I just had to share it. Apologies in advance to those who have seen it before.

 

[BootLoxes]

I came across this ZBrush youtube video again, and it got me excited (again) for my MBP which is en-route - and I just had to share it. Apologies in advance to those who have seen it before.

I was just watching that earlier today. I wish they would release a native zbrush already

 

[shuto]

I'm still trying to work out if I should get M1 Ultra 48 core or 64 core GPU. Now that we have a 64 core benchmark result from Stu Maschwitz's review, and his M1 Max 32 core GPU result, I think its fair to assume the 48 core would be between the two. I made the below graph to better visualise it...

I'm including 2x980ti's as this is the current setup I have in my PC (must be the most outdated setup here ? ). But tbh I don't have loads of client GPU work so these have suited me fine, and the M1 Ultra rendering as fast as them would be ok for me I think.

I think this graph has been pretty helpful seeing that the 64 core GPU is quite a bit faster than 48 core, and I'm thinking thats the one I should get.

What do other people think - would you spend the extra £1000 on getting the 64 core over the 48 core GPU?

(also posted this on the redshift forum - thought people here would also be able to help thanks!)

 

[iPadified]

It is still amazing that M1 Ultra is compared to dedicated top end desktop graphics cards. Not bad for integrated graphics. I am still wondering how Apple will compete with NVIDIA/AMD in the high end compute regime. It will not be sufficient to glue four M1 Max together even if it was possible.

 

[mi7chy]

Added M1 Ultra results for comparison. Scaling from 48GPU to 64GPU doesn't seem worth +$1000 which is almost the cost of the laptop at top of list or 3080ti FE.

Blender BMW
16.39s - Nvidia 3060 70W mobile (GPU OptiX Blender 3.0)
20.57s - AMD 6900xt (GPU HIP Blender 3.0)
29s - 2070 Super (GPU OptiX)
30s - AMD 6800 (GPU HIP Blender 3.1)
34s - M1 Ultra 20CPU 64GPU (GPU Metal Blender 3.1)
37s - M1 Ultra 20CPU 48GPU (GPU Metal Blender 3.1)
42.79s - M1 Max 32GPU (GPU Metal Blender 3.1 alpha)
48s - M1 Max 24GPU (GPU Metal Blender 3.1 alpha + patch)
51s - Nvidia 2070 Super (GPU CUDA)
1m18.34s - M1 Pro 16GPU (GPU Metal Blender 3.1 alpha + patch)
1m35.21s - AMD 5950X (CPU Blender 3.0)
1m43s - M1 Ultra 20CPU 64GPU (CPU Blender 3.1)
1m50s - M1 Ultra 20CPU 48GPU (CPU Blender 3.1)
2m0.04s - Mac Mini M1 (GPU Metal Blender 3.1 alpha + patch)
2m48.03s - MBA M1 7GPU (GPU Metal Blender 3.1 alpha)
3m55.81s - AMD 5800H base clock no-boost and no-PBO overclock (CPU Blender 3.0)
4m11s - M1 Pro (CPU Blender 3.1 alpha)
5m51.06s - MBA M1 (CPU Blender 3.0)

 

[macacam]

I am happy that Apple takes 3D somewhat seriously. I am surprised they had completely missed the side of creativity and got stuck with video, photography and music. Looking forward to hardware RT in MX at some point.

They had made major in-roads with the intel mac pros. The early Pixar connections to Jobs seemed to create a bunch of waves. The Apple/Nvidia tiff really put a damper on their progress with 3d. Then, imho, the trash can was last straw that really pushed 3d artists away from using Apple products.

 

[rgdigital]

Yep not optimized yet. In other GPU based task and benchmarks, the M1 Max reaches 3070m and in some rare cases even 3080m levels of performance.

GPU rendering has been almost exclusively developed for Nvidia GPUs for the past 5+ years so it’s not surprising that Apple Silicon (and AMD for that matter) are falling behind and need extra development to show their true performance.

If Apple was already getting twice the level of rendering performance before the official Blender 3.1 release then we should be seeing something closer to 3070m performance once the optimizations are built in and especially if the neural network is used for de-noising.



Since version 2.8 onward Blender has quickly moved from a subpar 3D app to one of the top choices for professional 3D users. I personally have spent years paying for the big name 3D modeling applications and last year, I switched to Blender and haven’t looked back. As someone who has been in 3D for 15 years now, I can honestly say Blender has become my preferred software (something I never thought I would say).

Also Blender and Cinema 4D are so far the only 3D software packages that run natively on Apple Silicon. However, for rendering Redshift and Octane are great on Apple Silicon as well (though more optimization needs to be done in those as well).

As somebody who's been using Cinema 4d since version 7, how much of an instinctual change or paradigm shift is it going from C4d to Blender?

 

[terminator-jq]

As somebody who's been using Cinema 4d since version 7, how much of an instinctual change or paradigm shift is it going from C4d to Blender?

Unfortunately I don’t have much experience with C4D but I know for motion graphics work, C4D is still considered the king in the 3D world. Although the nodes additions in the latest Blender versions have enabled a lot of new things.

I personally went from 3DS Max to Maya to Modo and now Blender. Each transition was pretty tough. I’d say the switch to Blender has been surprisingly the easiest though since there’s so many resources out there. The new interface is also much better. I basically started out with the famous doughnut tutorial and from there I was able to get going quickly and apply the skills from the other 3D applications. The Blender market has also been pretty phenomenal for adding new features.

Also viewport performance using the Apple Silicon build on an M1 Max is fantastic! I just completed an architectural rendering project with millions of polygons, a bunch of lights and several hundred objects and it stayed pretty smooth even in Evee mode.

 

[sirio76]

As somebody who's been using Cinema 4d since version 7, how much of an instinctual change or paradigm shift is it going from C4d to Blender?

Blender is free, so any shortcoming will be compensated by that. In general is a capable software with plenty of nice feature, I just feel is always in beta and not as user friendly compared to C4D.

 

[Lone Deranger]

I'll use it for Software Dev & Visual Design.

So a mix of Xcode, Figma, Blender primarily. Then the odd bits of FCP, Logic, Motion, Photoshop, After Effects.

I hope you'll stick around this thread and would be willing to supply it with useful 3D render-tests.

 

[Xiao_Xi]

Apple seems to have started working on the Metal backend for Eevee (Blender). The lead developer in charge of the Eevee rewrite has posted:

Apple is currently providing patches to have a fully working Metal backend. This should provide better performance than of a MoltenVK implementation in some areas.

EEVEE Next

Precision: MacOS openGL is frozen a 3.3 on some Apple hardware we support. So this is the common denominator for now. If the system does not meet the requirement, an error message will be displayed at the top of the viewport. So until there is full Metal support, this will be the state of...

devtalk.blender.org

Hopefully, all of Apple's work on Blender will help other 3D software. ?

 

[-narcan-]

I hope you'll stick around this thread and would be willing to supply it with useful 3D render-tests.

Yeah I can definitely provide some tests / benchmarks. I’ve got the 48 GPU Ultra. Anything of interest that comes to mind ?