SS on macs - parallel windows

Has anyone tried SS on Mac with parallel windows?

Very curious to know if the hardware capabilities of M1 ultra can be truly utilised?

@team SS, will you make some test runs and let know how it goes?


We, being curious too, made some small tests involving parallels on Mac M1 Pro 16 2021 (base version) vs. Dell G5-5590 2019 (notebook)

Demo scene - example room.

Draft settings (default)
Samples - 100
Bounces - 8
Lightmap Resolution: 75
Max lightmaps: 2
AI denoiser: off

Mac M1 -
Parallels desktop 17
Windows 11
Using 8 cores, 12GB RAM
Virtual CPU 3.20GHz

Time to bake: 20 m 58 s
Dell G5 (notebook) -
Windows 10
Intel i7-8750H CPU 6 cores 12 threads 2.20 GHz
GPU: Geforce RTX 2060, 6GB RAM

Time to bake (using GPU - CUDA): 7 m 08 s
Time to bake (using CPU): 13 m 15 s

Is it expected that running Shapespark within VM will be much slower, however it is not beyond any usability. It is possible that M1 Ultra would perform better, but we don’t know how well Parallels manages M1 GPU cores.

Subsequent updates to Parallels software may bring some major improvements in this regard.

As for the base functionality - We have not tested all features of Shapespark with parallels extensively, but it seems that everything is working as expected. Import / upload, baking options, previews, denoiser, editing possibilities, etc should work correctly.

To sum up:

Shapespark should run properly with Parallels, but with very limited baking performance.
We would advise to use Windows system for best possible experience.

1 Like

I used to run SS on our my MacBook Pro 2018 which has a much lower performance than the M1. It did work and got us out of a jam, but its a much smoother process running it on a dedicated windows computer. Was good for draft renders on though go!

My MacBook was running 3.1 GHz Dual-Core Intel Core i5, 8GB Ram and Intel Plus Graphics Card 650 1536MB - realise this isnt a powerhouse but anything over a medium bake would take a about 18+ hours to render. Parallels tends to be quite strenuous to run so just from experience, I dont think its the most efficient method.

1 Like