The new M1 Max and M1 Ultra Mac Studio look insane! Been waiting for something like this for a long time — the Mac Cube reborn, but with the most powerful / efficient CPU/GPU available.
Hoping to be able to afford the Ultra version, but even the base Max is a well-spec’d beast. Planning on shifting my workflow from laptop to desktop + iPad. (Basically I can get the Studio Max + iPad Air 5 for about the same price as the MacBook Pro that I was previously considering)
Really excited to see how this new Mac will run SU / LO and TwinMotion.
(also noted that the new M1 iPad Air 5 presentation included a reference to SketchUp - can’t wait for SU for iPad!)
Now that looks promising. I haven’t owned a desktop Mac since the mid '90’s because they would never make a mid-market, sweet spot mini-tower type machine. The Mini was too wimpy, the Pro to ridiculously priced, and I didn’t want an iMac.
So they actually showed SketchUp for iPad? I didn’t watch the event.
At one point I saw about 750k people watching so not a bad plug for SU+iPad and to the OP’s question…they sure are making a good case for power users to upgrade desktops. And here my iMac Pro has been serving me so well.
Does anyone know about M1 chip rendering (ie V-Ray) performance compared to say iMac Pro?
You can browse the v-ray benchmarks on the chaos site and there is a around a 5% increase in performance for an M1 Pro Max be the 2019 models with the intel chips on the same benchmark.
Most of the Power of the M1 thingy comes from Multicore and GPU stuff. Single-core is good but there is no difference between what the Pro and the Ultra has to offer. Which is - unfortunately - not helping us Sketchup-Users all too much. As far as I can tell only one core can be used for Sketchup. I am using Sketchup on my old iMac 4Ghz QuadCore i7 from 2014 AND on my MacBook Pro M1 Max. The difference in performance is - not that noticeable. And Vray is not native M1 yet so no real render-beast either.
So if Sketchup is your only work-horse you can go for a “low-end” Mac Studio - just lot’s of RAM that always important.
I am very curious to see if the Sketchup-Dev-Team will come up with ways to use more then one core at one point or the other. EG: rendering multiple viewports in Layout (one viewport per Core) or stuff like that.
I see that Blender 3.1 has just been released with full Metal support:
Really interested to see someone test it with a fully-loaded Studio Ultra (128GB RAM / 64 core GPU). Curious as to what that GPU access to untold amounts of (unified) memory will yield in real-use terms.
Since we’re still waiting for M1 / Metal support for other renderers, could make a SU to Blender model/render pipeline the fastest option currently available…
Right On napperkt!
Looking at Mac Benchmarks - Geekbench, it appears that ALL the M1 individual CPU cores run at 3.2 GHz clock speed (Please correct me if I’m wrong!).
Similar to you, I am currently running a iMac 4Ghz QuadCore i7. I’m interested in the new Studio of course (especially as it looks like Apple may kill any future high end 27" iMac?), but will any of the M1 line of chips actually improve the speed and performance of SU compared to my current (but old) 4.0 GHz speed chip? On the face of it, it appears to be a 0.8 GHz DECREASE in CPU clock speed!?
Does anyone out there know if this is correct?
And … do the M1 GPU cores also run at 3.2 GHz clock speed?
Hey @Colin, you got an M1 a while ago, any thoughts?
Any insights would be most appreciated!
Yes, thank you Colin …
But do you have any thoughts about the individual core GHz clock speed questions that napperkt and I raise?
Since we are only using ONE core (regardless of how many we have), isn’t the clock speed of individual cores key to SU speed/performance?
Have you been able to compare your SU performance on the M1 (at 3.2 GHz) against anything with a faster clock speed?
Anyone else out there with an M1 have thoughts/comments?
Unfortunately GHz number means nothing if you are comparing two different chip architectures. Even between two different Intel generations of chips it’s not a comparison metric. Deep dive into single core Geekbench results > it’s all about comparing apples with apples
The M1 (Max) is definitely a great system. Even if Sketchup isn’t radically different in performance in itself I can have a lot of different apps open at the same time and have many processes running. Haven’t been able to bring it to its limits yet.
Thanks jellway, “apples with apples” … I assume pun intended!
Anyway, you say GHz is not a comparison metric … it’s puzzling then that the GHz clock speed has been used (at least in the past) so widely to tout the virtues of the various chip iterations? A higher GHz was always “better”.
I appreciate and acknowledge your point about the great multitasking improvements that the M1 chips offer, but I’m trying to focus solely on SU/LO performance.
So, with this new Apple chip architecture, how DO we evaluate the potential SU (and especially LO) performance boost of these various M1 variants? We need to “evaluate” this potential so that we can make informed decisions on how much of this new silicone horsepower we should buy to get a given improvement in performance. @Colin, thanks for reporting that your M1 is working well for you, but how much better than a given older Intel chip? Isn’t Trimble/SketchUp interested in quantifying and publishing the measurable performance boosts that these new Apple chips offer? If not exact numbers, then at least a general range of potential improvements. Aren’t there some industry accepted “benchmark” tests that can be applied to put a yardstick on what we can expect with these new chips?
Anyone?
I think that generally we don’t recommend specific hardware, because if what we suggested ended up not working well, a customer could make a case for not paying us. We have some big customers where them not paying us would hurt!
That aside, I don’t mind posting test results, those are factual, and if you go on to make a buying decision influenced by the figures, it’s your own decision.
To answer part of your question, I found that my M1 Max running SketchUp 2021 under Rosetta was very nearly as good as my work Mac that is a 2019 Intel CPU MacBook Pro. The SSD is faster too, so even ahead of 2022 being released it was feeling like a good experience.
With 2022, some of what I can measure on M1 native compared to Rosetta suggests that native is often 2x Rosetta.
Funnily, Apple did feature SketchUp as being a good use of M1 on iPad. I don’t mind if other people brag for us!
So if we dig a bit into Geekbench results, it’s a pretty good way to get in indication of the performance comparison between processors. Apologies if I am over explaining this, but will do so for people reading along that might not know all the detail! When thinking about SketchUp of course, only look at the single core results given its not a multi threaded app (larger number being better). I have three machines here. One Intel based Mac and 2x M1 based Macs.
Clock speeds (MHz number) don’t necessarily increase with performance.
Your 27" iMac is in the same league as my old Intel MacBook Pro 16" (1085 vs 1105)
My M1 MacBook Pro is approx 60% faster than my old Intel MacBook Pro and your 27" iMac (1105 vs 1782). So not double the speed > more like 1.5x.
My M1 Mac mini isn’t that far off my M1 MacBook Pro. This is a pretty good indication that most of the M1 Mac models are performing similar in terms of single core performance. So buying an M1 Max or M1 Ultra chip is a waste of money unless you use other applicants that either take advantage of multiple processing cores or the graphics architecture. You can see this on the Geekbench results that you posted earlier > all hover around the 1700 mark.
Thank you jellway!
This is exactly the data/feedback I was looking for! (I added qualifying/clarification text in italic parenthesis for clarity)
This is the conclusion I was digging for, and hoping to have confirmed. When it comes to SU/LO performance, the single core “speed” is key, so the boost we will see from the M1 family of chips will be limited by that. In addition, since all the single core speeds are the same through the M1 family, the bigger versions won’t effect SU/LO performance any more beyond the base M1 chip.
The only questions I have left now are:
What about all those M1 GPU cores? Do they run at the same speed as well?
And, do more of them have ANY effect on LayOut performance in particular?
What about the future M2 chip? Will THEY feature an even faster single core speed? (One can only hope …)
So, while I would love to have a good reason to buy one of these souped up Apple Studio machines over a base M1, I guess an even greater SU/LO performance boost isn’t one of them!
Thanks again for helping clarify this and confirming my suspicions!
