Hahaha. I’m currently rediscovering 3D Coat, which has its ups and downs. Lots of features are brilliant, but the UI / UX has its quirks and it’s buggy here and there.
Sorry for the off-topic deviation by the way. 
2 Likes
Rumor mill 
Curious if there will be blender at that event
2 Likes
More rumors. 
1 Like
Yes me too or if they will wait till WWDC.
Also Apple registered 3 new Mac model numbers , so I am really curious what spring will bring.
If I had to guess the line with the 2 numbers you s the mini M1 Pro and Max (if they make different model numbers based on SOC but not sure).
The other new Air?
Or perhaps it is the Mini and the new M2 or they are all M2s as someone pointed out as the model numbers are quite different from all the M1 machies. 
I’ve never seen these kind of results replicated in any real life scenario where the gpu is really critical. It’s enough to see redshift, current cycles x metal performance or even metal supported games.
When you have a 3080 (with proper wattage) the m1 max is nowhere near. It’s more on par with 1660ti/2060 performance.
Synthetic benchmarks are not so useful IMO
1 Like
But Sofar we had m1 mobile devices
The mac mini has the brain as a a laptop
interesting will be dedicated desktop devices where watt and battery usage is not ghat much of a focus
Then we can see
yes, of course desktop performance will be better, but again, m1 max equals my 1660ti laptop videocard, so we’re comparing m1 max with a mobile chip. The m1 is much more efficient, which is great, and the laptop is much nicer and portable than my legion 5. However it also costs 4x times more and it’s 3 years newer than the legion 5 hardware.
I am looking forward to see what they come up with next, but as I said, IMO there’s still a huge performance gap with GPU performance, and the competition is also moving in the meantime.
2 Likes
Apple machines need this.
Sorry could not resist and thought it was to funny that Nvidia is giving them away.
3 Likes
This whole GPU discussion is kinda funny considering that people even bought a BestBuy 200$ membership price to be able to buy a GPU and then the stock was sold out instantly …
2 Likes
Synthetic benchmarks are the most useful in this case because many games aren’t written for Apple Silicon and/or Metal. But for games written for Apple Silicon and Metal, the M1 Max is on par with a laptop 3080.
Of course rendering that uses RT, the nVidia cards are superior.
3 Likes
This will go well with the new RTX 3090 Ti. Nvidia’s specs for it are so high-powered they are reportedly having trouble finding dies that even work. The company is allowing the field to be wide open for AMD’s upcoming MCM cards if they need stunts to paper over the availability issues (along with Jenson Haung being so consumed by pride now it might be time for a CEO change).
Could you show me some real life scenario (3d rendering or video game) where the m1 max performs on par with a fully watted mobile 3080?
My 1660ti mobile doesn’t have RT cores, still performs as the m1 max in blender.
Games with native metal often have very similar perfomance when compared to the same game emulated on rosetta. They did a great job with Rosetta at Apple.
1 Like
I think it is more about OpenGL vs Metal and app/game port quality than Rosetta vs Native. If I am not wrong OpenGL/Metal code runs natively even in Rosetta mode.
There aren’t many decent Mac games, but here’s an example – M1 MAX vs RTX 3070 (140 W) in Baldurs Gate 3:
There are also World Of Warcraft tests somewhere that show that the M1 MAX actually comes close to the RTX 3080 crap in rasterisation.
Fordee linked to GFXBench results that confirm this:
There are Apple Motion tests from 2019 on Barefeats which shows that Metal app can be 2-3x faster than OpenGL:
It is all about Metal usage and optimisation for a Mac. Back to Blender topic - all this means that Metal Eevee would scream on a Mac. OpenGL old fart is a tragic waste of Mac resources.
1 Like
Interesting that you posted a YouTube video with the original M1. If you want to talk about the M1 Max, please use that as the comparison, not the M1.
vey interesting! Well, I am looking forward to see that in blender then! Maxon hasn’t done a great job with redshift if the m1 max is capable of this performance then, since there performance is quite far from even a 2060 from what I’ve seen
sorry, my mistake, I didn’t realize that and it wasn’t my intention. That said, it’s just a comparison between the native metal and rosetta emulated performance, I am not sure if with other chips it would change that much
No optimizations have been done for Metal Cycles yet, so difficult to compare, but how long does your 1660ti mobile take to render BMW scene?
We are talking about two different things - rasterisation (e.g. viewport, Eevee, games) and raytracing (e.g. Cycles X). M1 MAX is very strong in rasterisation but not so good in raytracing (at least at the moment). M1 MAX is nowhere near RTX 3080 (full powered) in raytracing. Therefore, I would not expect miracles in final Metal Cycles X version. Overall it is just OK.
2 Likes
40 secs in 3.0
ok that makes sense, thanks. It’s unfortunate, since ray tracing is what most interests me. Let’s what happens with the next chips
1 Like