One thing to note is that 16-cores GPU is the M1 Pro which has ‘half’ of many things: half the memory bandwidth, half the hardware media engines, half the supported screen-monitors/total-resolution, etc.
I think if it were me and already at 32GB of RAM I would definitely pay the price for the 24-core GPU bump.
Just in case, I’m not buying anything yet so I got no skin in the game. Still thinking if I just continue on with an iMac and wait for the iMac M1 Max whatever or go fully portable but powerhouse dockable. Since work from home I have used a laptop a total of 5mins maybe… loving the iMac big screen and experience as a whole.
True this, maybe it’s time to complain a bit to apple about what’s up with their mice… the concept is ‘nice’, the touch top, gestures based, etc but man it is so ergonomically horrible, the click feels nasty (for me at least) too, it slides weirdly over surfaces, etc.
Is that like the After Effects track thing that starts flooding 3d points on the footage? or also like the recently added FCP features where can you drop an effect ‘on the face of a person’ and the thing just pins it to that?
I think it’s nice as a post thing but having it directly inside a 3D software has the benefit of attaching 3D objects directly, I think it’s useful yes… Ian Hubert has some amazing examples of him tracking stuff for his short.
More than rendering performance, I’m curious how it would handle geometry and textures calculations.
Like: how responsive would it be with shader and geonodes, how would it be at handling a scene with >10 million polys across a dozen objects, while maybe having to texture a few, etc.
My 2015 Xeon with 32GB is struggling hard now…
4 Likes
Renzatic
(Professor Emeritus Billy H. Wafflesmith XIV Esq. Certified Legomaniac)
3234
Based off what we’ve seen previously, transforming vertices, geonodes, and simulations seem to be the areas where the M1 truly excels. When it comes to anything that requires the CPU to calculate tons of raw numbers, the thing is an absolute beast.
One of the developers of Redshift renderer says that Apple GPU in M1 Max will be able to access 40gb+ of memory of 64Gb total. Apple obviously put a cap on memory being used by GPU to not let the system struggle.
Same person say current M1 Macbook Pro Redshift renderer can use 11Gb of 16Gb.
I’ll keep posting updates, 2 of Redshift forum members say they bought a M1 Max 64Gb that are expected to arrive next week.
I wonder if that’s a Redshift thing only? Cause Jules from Otoy say he’s using 64GB. And from my personal experience with the M1 I go over the 16GB limit in most cases with the M1 then switching to swap.
True, but the Math doesn’t add up. If Redshift devs are saying they are limited to 40GB and you say cause of the OS needing the remaining, that means the OS needs 24GB?
Mac OS runs off a few GB. I mean Apple still produces 8GB only computers.
Since the M1 these systems can utilize almost twice what the available RAM is because of the way they do swap memory with he SSDs. And now with the 7.4GB read/write SSDs in the new machines, Swap is going to be even more beneficial.
I’ve rended scenes up to 30GB VRAM on my 16GB M1. And only had it say out of memory at 33GB.
Blender tells you in the bottom right corner during and after rendering. If it’s not showing for you, right click at the bottom bar and select VRAM statistics.
Renzatic
(Professor Emeritus Billy H. Wafflesmith XIV Esq. Certified Legomaniac)
3248
Out of curiosity, what were you rendering that ate up over 30GB of RAM? I thought I went overly crazy with my last render, yet it didn’t even top out at more than 2GB.
