Keep in mind that OpenCL was officially depreciated by Apple in 2018. These benchmarks might thus not be the best thing to look at.
Comparing SOCs to APUs is probably a pointless endeavour anyway, since it’s like comparing apples to oranges, even if all the cool social media hipster like to parade their pointless benchmarks.
The only thing that can be said for certain is that the M1 uses way less power while still performing superbly, other than anything that Intel, AMD, or NVIDIA have put out thus far.
Many people don’t seem to care about their electricity bill though.
The messages are very mixed, when it comes to the performance of Rhino 7 and the M1 chips. I’ve seen people say that it’s normal even on the standard, low-tier one, and others claim that it’s horrible. The truth might lie somewhere between. There’s lots of misinformation out there.
If it were utterly unusable, I doubt McNeel would advertise Apple Silicon support on their site.
would that mean that Rhino 7 could still get a little performance bump?
already hovering over the order button for a while i found claims that the studio is after all not that silent as stated initially. many complaints can be found all over the net. a theory seems to harden that the max are the noisier ones due to having a aluminium heat sink whilst the ultra seem to dissipate the heat better having a copper heat sink instead.
having a no noise mac raised my hopes more than three years ago when i first was talking about rumours claiming that mac will go arm which made me wait and wait and wait patiently. what now, paying 2.300 € more than the max version is not what i had i mind. i am even considering this (which would be my dumbest move ever) i am completely lost. with rumours that some m2 could appear this year still it might be the worst time ever to buy a computer.
That’s the only version where I care about making improvements with respect to OpenGL on Mac. Pretty much the only thing I plan on continuing to work on in Rhino 7 is display performance with OpenGL on M1. If I can find a way to improve the situation, I will try to add it to Rhino 7 (no promises though.) Apple is not improving their OpenGL drivers and if anything they have gotten worse with the release of the M1 computers. Rhino 8 will use metal by default.
i understand that completely, i would not wish you to waist too much time to feed that dying dinosaur anyway. also i think apple would just have to invest a lot of efforts rewriting all their open gl drivers which might be the reason they rather dropped it than dragging it along, just speculating though.
but solely relying on wip 8 feels a little risky… so i am just evaluating what to do. i could live with it but better be safe than sorry.
can you maybe just turn down all kind of gimmicks like aliasing to improve workability?
I have a friend who already got his Max and it’s barely audible. The thing is that the fans always spin at about 1300 RPM, even when under heavy load, which is a plus.
All I could hear was a sound like a very, very slight breeze or air movement when I put my ear right next to the back air outlet. You can also lower the fan speed to 1100 RPM, if you want to make it more quiet.
A misconception is that Apple claimed it to be silent, which is not the case. The website states nearly or close to silent, which seems to be the case. My work computer, a Windows desktop, or my LaCie Rugged external hard-drive at home are way louder.
Much confusion was caused by the first iteration of the power supply board, which seems to have been faulty and thus produced weird hissing or whistling sounds. Some boards had problems with a broken fan and others with coil whine, because relevant components weren’t isolated properly. Later models seem to feature a revised board with a different component layout and proper isolation.
If you want something that is totally silent, get a computer with no fans, like the MacBook Air.
Mac Studio is a workstation computer that is very quiet, which is good enough for me.
With the long wait times for their current models, COVID outbreaks in China, and general supply chain issues, I personally doubt that releasing even more new computers before late summer or autumn would make much sense, if you can’t get them out to the clients anyway.
I can maybe see them showing off something like a revised Mac Pro that would ship later - which I wouldn’t buy either way -, or maybe even teaser the M2 chip itself, but that’s it.
First in line to get an updated chip are also the MacBook Air, 13" MacBook Pro, and Mac Mini, which are the oldest models featuring the M1, and I’m not in the marked for either of those.
if you have a constant (even silent) carpet surrounding you a little swooshing may not add anything at all. also some desks stand in the middle of the room while others are facing hard concrete walls which will whistle back all the high pitch sounds you throw at it with ease. i did not yet read anything regarding coil whine but i gathered that there surprisingly are (were) 2 seemingly different logic board layouts for the same set up. also i assume that these rants were regarding the regular idle sound it emits and not the faulty examples, one can not be certain about that though so i am left wondering.
my macbook pro late 13 is very quiet with an idle rpm of around 2100 also while surfing typing etc. that is the reason i sold a newer model again reverting back to this one because the fans started howling by almost just looking at the screen only.
if 1300 rpm is the idle speed of the studio the noise can just be deeper, but maybe stronger due to a bigger fan maybe causing resonance on a nice table idk, i just have a musically trained ear so i am pretty sensitive.
the macbook air would be a (weak) contester but i wanted a laptop free super compact desk because i had to shrink down my working desk hence aiming for a mac mini initially but waiting for one with just a little more juice.
Apple has promised to end their transition within 2 years. they revealed the big move on the WWDC 2020, now with the WWDC 2022 coming in june and even talking about the macpro in their last event and with a lot of rumours claiming that a lot of new models may be coming including the first m2’s i would assume that at least something is going to come. i am very urged but i might just have to wait those few weeks and then decide. a mac mini with an m2 chip and more ram may just be what the doctor has ordered but i might just go crazy and take macpro, its all about noise for me if any can fill the bill so be it. also the studio is such a cube i might as well just take a pro already… or i might trust your ears then when you get yours.
did you take the 32 core gpu? (seems such a little price difference for such a sign difference. would be dumb not to which makes the 24 gpu core generally a bit of a dumb device thinking about it)
right, probably the only logical choice if you are aiming for a more state of the art performance. please let us know how it performs also in regards of rendering.
I’ve received the Ultra last Friday and haven’t had time to test rigorously.
I’ve installed Rhino 7 and done some minor work in it, mainly in Grasshopper.
Generally, it doesn’t seem to work worse than on my 2016 MacBook Pro, it’s at least somewhat snappier, however I’ve noticed that it doesn’t have a high tolerance for many geometries, before it starts to stutter, which was previously reported by other Apple Silicon users.
The Raytraced view is also nearly as bad as on my old laptop, which means unusable.
Arctic view is super fast, you barely see any noise any more. It instantly refreshes.
I’ve also tested Blender, which works amazingly well, and shows what some optimizations can accomplish. The render speeds are off the charts, compared to my 2016 MacBook Pro.
thanks for the feedback. what would also interest me a lot is how a standard orthogonal 2d view workflow from top, sides feels like with a lot of curves text and/or some 3d geometry. zooming in and out or panning the view and for geometry standards like move, drag, mirror etc. i mean does it pop instantly or do you see any kind of lags?
looking at your comparison i am happy that metal is taking over, yes open gl is just not properly supported anymore obviously. but i sense it could be faster though but then again that is quite some geometry. which monitor are you using, set to which resolution? i am having a huge problem with apples scaling issue which brings the performance down on many computers with users having non apple screens attached not even knowing it and probably complaining about apples performance, just in case you are not aware here i made a topic about it.
also what interests me a lot if the multicore gpu is utilized accordingly. having no in depth computer hardware expertise and knowing that rhino generally does not use any cpu multicore for its main tasks other than some specifically rewritten stuff which is more on the grasshopper component side i believe rather than rhinos commands, i am thinking if that has any effect at all. @stevebaer is multicore gpu something something you need to take care of or is that being handed over by apple?
I haven’t noticed any lags in Rhino 8 with TestMetal active, yet. Here are some tests that I did while transcoding video in the background.
Medium sized project with curves, hatches, 3D geometries, blocks, and text:
23 305 curves baked from Grasshopper:
You can send me a file, if you want me to test something specific.
Apple Thunderbolt Display (2560 × 1440, no scaling).
I am aware of this. On macOS scaling is done on the GPU.
This is currently hard to test, since all my cores are fired up because of the video transcoding, but Grasshopper surely has some components that are multithreaded.
Generally speaking, I’ve noticed that many programs use multithreading on macOS nowadays and you can run many tasks in parallel without having any trouble, like me currently with Rhino, Grasshopper open and ffmpeg transcoding video, and a YouTube stream running.
The Mac Studio Ultra doesn’t even seem to care. It uses about 50W of energy and rocks a current temperature of 52°C (CPU average).
I think this works differently. On the GPU everything usually runs in parallel, there are no sequential tasks. If you program for instance a shader, you have a vertex function that is applied to each and every vertex in parallel, if that makes sense.
