Author Topic: Apple's new M1 Studio  (Read 2719 times)

2022-03-09, 12:01:49

frv

  • Active Users
  • **
  • Posts: 445
  • Cheers Francois
    • View Profile
    • Web
I am about to order a new m1 Ultra.
Just wondering what specs are best.
I think 128Gb ram, 2Tb disk and I wonder if the GPU cores mean anything from 48 to 64. Maybe for Redshift but for Coronarender it probably does not make a difference.
I assume a big harddisk is good for quick access to resources whan starting a render. Biggest bottleneck I experience is starting a render like a 100 times an hour to finetune settings.

Anyhow, I think that Coronarender now is the best render engine to go for in Apple's world. With the latest hardware most likely fast as well. Since my work is about modeling (vectorworks & C4D)  for over 75% of the time and 25% rendering a new M1 ultra is probably faster all together than any PC. I think Apple is one the right track.

I hope the problems with the license server (daily builds v8) are soon fixed.

2022-03-09, 14:28:01
Reply #1

Rutg3r

  • Active Users
  • **
  • Posts: 62
    • View Profile
More GPU cores may result in a better viewport performance. If you need preview animations with complex geometry it can help. It depends from the tasks...

2022-03-09, 20:40:27
Reply #2

BigAl3D

  • Active Users
  • **
  • Posts: 963
    • View Profile
Anyone have any idea how these Apple-made GPUs fit in the big picture? I mean many apps and plugins are optimized for NVIDIA and now adding AMD to the mix. How do these new integrated GPUs work with all the players?

2022-03-10, 07:55:01
Reply #3

Philw

  • Active Users
  • **
  • Posts: 280
    • View Profile
Apart from the GPU renderers like Redshift (who are currently in the lead with proper M1 support - and Blender just dropped a new version with proper Metal rendering backend) - M1 Gpu acceleration for other tasks is sparse right now. X-particles GPU based rework coming late summer allegedly has M1 in the mix.

2022-03-10, 17:11:38
Reply #4

frv

  • Active Users
  • **
  • Posts: 445
  • Cheers Francois
    • View Profile
    • Web
I had a look at the Redshift forum and saw a discussion on time to first pixel. One of the most important bottle necks for editing scenes. The discussion went on to mention that the more gpu's the slower the time to first pixel. Would it be the same for a m1 Studio 20 cpu cores and  64 gpu cores. Slow to the first pixel and fast in the final render.
Anyhow, I wonder if this also applies to more cpu cores and coronarender. Would corona render be slower in time to first pixel with more cpu cores. Otherwise I might be better of with a 10 core m1 Studio and 24 gpu cores. Rendertimes to final render are less of a concern to me. Once I am ready for the final render I can render overnight or on a farm.

2022-03-10, 17:15:28
Reply #5

Philw

  • Active Users
  • **
  • Posts: 280
    • View Profile
I don’t think it’s the same consideration. You’ve basically got 1 GPU with a tonne of VRAM with no PCIE bottlenecks.