Mac Studio
Apple (MacRumors, Hacker News):
With the power and efficiency of Apple silicon, Mac Studio completely reimagines what a high-performance desktop looks like. Every element inside Mac Studio was designed to optimize the performance of M1 Max and M1 Ultra, producing an unprecedented amount of power and capability in a form factor that can live right on a desk.
Built from a single aluminum extrusion with a square footprint of just 7.7 inches and a height of only 3.7 inches, Mac Studio takes up very little space and fits perfectly under most displays. Mac Studio also features an innovative thermal design that enables an extraordinary amount of performance. The unique system of double-sided blowers, precisely placed airflow channels, and over 4,000 perforations on the back and bottom of the enclosure guide air through the internal components and help cool the high-performance chips. And because of the efficiency of Apple silicon, Mac Studio remains incredibly quiet, even under the heaviest workloads.
[…]
On the back, Mac Studio includes four Thunderbolt 4 ports to connect displays and high-performance devices, a 10Gb Ethernet port, two USB-A ports, an HDMI port, and a pro audio jack for high-impedance headphones or external amplified speakers.
[…]
And because users frequently connect and disconnect devices, like portable storage, Mac Studio includes ports on the front for more convenient access. There are two USB-C ports, which on M1 Max supports 10Gb/s USB 3, and on M1 Ultra supports 40Gb/s Thunderbolt 4. There is also an SD card slot on the front to easily import photos and video.
Wow. This is pretty much the xMac (but with no internal expansion). It even has the same base price as the old Power Mac G5. With upgrades, it’s basically $900 less than a 14-inch MacBook Pro—no display but likely better thermals. The M1 Ultra (with the smaller 48-core GPU) is a $1,400 upgrade over the M1 Max, given the same storage.
Apple announced today that the Mac Pro is the only Mac yet to complete the transition to Apple Silicon, so it sounds to me like there will not be a new 27-inch iMac, at least not a Pro/Max/Ultra-caliber one. With the Studio Display announced, and no large-screen iMac to wait for, that simplifies my purchasing decision. I’ll likely get a Mac Studio to replace my 2017 iMac, though I’m not sure yet which configuration.
For one thing, I don’t know how well Xcode scales to lots of CPU cores. Overall, the different models of the M1 family are not great fits for my needs. I wish there were an M1 Pro with more RAM or an M1 Ultra with fewer GPU cores. (It also seems like there’d be room for a Mac mini with an M1 Pro.)
The Mac Studio is very much the desktop Mac I’ve wanted for years. A Mac Mini...just more of it. Faster, more expandability, more displays.
Previously:
Update (2022-03-09): Dan Moren:
The only other key difference between M1 Max and M1 Ultra models are that the former has only USB-C ports on the front, as opposed to the latter which features Thunderbolt 4 ports.
According to Apple, the M1 Ultra has a copper thermal module, while the M1 Max has an aluminum heatsink. Copper is heavier than aluminum, hence the weight difference. The M1 Max Mac Studio weighs in at 5.9 pounds, while the M1 Ultra version is 7.9 pounds.
There are two other benefits of large internal SSDs you should consider carefully. Larger SSDs perform better than smaller ones, and if you read the fine print on benchmarks quoted by Apple and others, you’ll see performance figures are invariably obtained on their highest capacity models. The other important benefit is slower wear, particularly if you can keep several hundred GB free at all times.
Apple chose Tuesday to remove the 27-inch iMac from sale. It’s a curious decision since the Mac Studio doesn’t really serve as a one-to-one replacement—especially when you add in the cost of an external monitor.
[…]
I have a hard time believing that Apple thinks that the single 24-inch iMac design is all that’s required.
So, a prediction: I think we’ll see a larger iMac next year. But I don’t think it will be capable of being specced up into the equivalent of a Mac Studio. More likely, it’ll be a larger version of the 24-inch iMac running an M2 chip.
The M1 iMac continues to be branded the “iMac 24″” which seems to leave the door open for a different-sized model. That suggests to me that a hypothetical 27-inch would be viewed more like a size variant and have more in common with the 24-inch model[…]
Kinda fun that Microsoft’s ‘studio’ PC is a beautiful giant drafting table with pen & touch for designers, but Apple’s ‘studio’ PC is a dull little metal box packed with specs for developers, video editors, & workstation-level workloads. We’re definitely in an alternate timeline
Update (2022-03-16): Howard Oakley:
User processes are almost exclusively run on P cores, with E cores being recruited when there’s sufficient demand. Those user processes should therefore be accelerated in proportion to the number of P cores, provided that there are sufficient threads to run. It’s that last requirement which is key: if there are 8 or fewer threads with high active occupancy, then the M1 Ultra’s 16 P cores will be of little or no extra benefit. Only when the number of heavyweight threads exceeds 8 will those extra cores result in improved performance over the M1 Pro/Max.
This is likely to be reinforced by macOS’s management of cores, which are grouped into clusters, typically of four cores (two in the case of E cores in the M1 Pro/Max). When running 4 or fewer heavyweight threads, only the first P cluster (P0) will be active; with 5-8 threads, the second P cluster (P1) will be added; the Ultra’s P2 and P3 clusters will normally remain inactive, at a frequency of 600 MHz and full idle, until 9 or more threads are fully active.
Where the M1 Ultra may prove little advantage is in macOS background tasks, which aren’t just configured to run on E cores alone but also usually have I/O throttling applied. Unless that throttling is eased and E cores are run at higher frequencies, tasks such as Time Machine backups and Spotlight indexing are likely to take as long on an Apple Studio equipped with an M1 Ultra as on an M1 Max, or even on an original M1.
I see the Mac Studio as the spiritual successor to the 2013 Mac Pro. It is meant to be small and compact enough to sit on the desk, not under the desk. It has a lot of IO ports for attaching external storage, additional displays and other peripherals, but it is not a user-expandable machine. The 2013 Mac Pro was compact, if only because Apple gambled on a future of GPU-oriented computation that never really panned out. Fast forward to the present day, and there is no need for trickery; it is the sheer efficiency of Apple Silicon enables the Mac Studio to boast top-tier performance in CPU and GPU benchmarks, all housed in an enclosure even smaller than the 2013 Mac Pro.
John Siracusa joins Jason to talk about the Mythical Mid-Range Mac Minitower, the distortion of the iMac over time, the modular possibilities offered by the Apple Studio Display, and other fallout from last week’s Apple announcements.
Honestly, I have never been so appalled at an Apple strategy. Between the launch of Mac Studio and the simultaneous death of iMac 27, we who have so patiently waited for an Apple Silicon-powered 27-inch iMac are suddenly left with only two options.
We can hang onto our aging computers and simply hope that a new iMac 27 will one day appear. Or we can spend more than double the cost of a typical new iMac 27 for a Mac Studio + Studio Display.
[…]
Never in a million years did I imagine that Apple could leave such a large group of customers twisting in the wind. It’s frustrating.
[…]
Before last week, I would have said it is not even remotely possible that Apple would orphan its iMac 27 customers. What makes our situation extra-debilitating is that we don’t know if this has really happened, or if we just think that this has happened—because Apple isn’t talking.
I still strongly believe a larger iMac with Apple’s pro chips is in development—but I don’t think it, nor the next Mac Pro, are coming anytime soon.
Update (2022-03-17): Jason Snell:
Speaking of chugging, I should mention that the Mac Studio has a fan—about half of its volume is taken up with a cooling system—and that fan seems to run constantly. It’s very quiet, throwing out low-level white noise that I couldn’t hear unless I sat in my office when it was completely quiet. But the sound is very much there, in a way my iMac Pro fan never was, and if you’re ultra-sensitive to fan noise in quiet environments, you will notice it. The good news is, not only is it quiet, the noise also seems fairly consistent. I threw graphics- and CPU-intensive tasks at the Mac Studio, and I couldn’t get the fan to sound any louder, at least to my ears.
[…]
The Mac Studio isn’t for everyone. But for the people who have been dreaming of something in between a Mac mini and a Mac Pro, something that wasn’t an iMac, it’s the fulfillment of a dream.
I guess I’m just going to have to get used to having a computer on my desk again.
I’ve, once again, given this device to a host of professionals on The Verge’s team, and the reactions I saw couldn’t be more different from those we saw in 2020 [with the Mac Pro]. They were impressed. For their workloads, it’s faster than anything they’ve ever used. It’s changed what they can do.
[…]
Oh, and yes: the M1 Ultra really is about twice as powerful in some benchmarks as an M1 Max. On the multi-core benchmarks in Cinebench and Geekbench, the M1 Ultra was consistently getting double the scores of the M1 Max. This did not translate to doubled scores in the real-world tasks, such as the Puget benchmarks, gaming, and the NPBench Python benchmarks that we ran to simulate scientific workloads — there are clearly bottlenecks here that aren’t just based on throughput, and performance will always reflect how well the program you’re in can use all these extra cores and threads.
[…]
It was a different story with graphics performance, however. Apple, in its keynote, claimed that the M1 Ultra would outperform Nvidia’s RTX 3090. I have no idea where Apple’s getting that from. We ran Geekbench Compute, which tests the power of a system’s GPU, on both the Mac Studio and a gaming PC with an RTX 3090, a Core i9-10900, and 64GB of RAM. And the Mac Studio got… destroyed.
[…]
In general, the consensus was that the Ultra is a bit faster. Like, maybe a millisecond or two. But that was pretty consistent across rendering, scrubbing, and everything else they did.
There is a pretty clear throughline from Apple’s reset of the Mac Pro, to its formation of the Pro Workflows team to the development of Mac Studio. This is a port-rich offering that provides users with a lot of what they have been asking for both publicly on social media and in meetings with the workflow team.
[…]
In my testing, to cite one example, the Mac Studio with M1 Max that I am trying out clocked an 8K timeline export from Final Cut Pro that was 3x as fast as the 2020 M1 MacBook and only 3 minutes slower than a near-completely-maxed-out Mac Pro. The Mac Studio as configured will set you back around $3,200. Only around $20,000 cheaper than the Mac Pro I used to perform this test back in 2020.
[…]
The number of ports available on the Mac Studio is something that came out of talking to its pro customers, asking them how many devices that they’re using. The USB-A options were a bit surprising to me, to be honest, but Boger says that their research showed that there was still a legacy need.
From M1 Max to M1 Ultra, the two USB-C ports on the front of Mac Studio do benefit by becoming full-on Thunderbolt. So you can drive a display right up front if you want to. But the SDXC card reader on the front is still UHS-II and HDMI port on the back is still 2.0. No bump up to SDUC and UHS-II, never mind the kind of CFExpress cards I’m using these days, or to HDMI 2.1… which is an absolute joke of a mess of a standard at this point, but would allow for even more bandwidth for even higher resolutions and frame rates, among other amenities.
Update (2022-03-23): Steve Troughton-Smith:
Apple doubling down on Metal does not seem like a strategic win when so many professional tools just aren’t optimizing for it, if they have Metal renderers at all. Instead, most just going through OpenGL/Vulkan compatibility layers, wherein performance is crushed by Windows GPUs
Obviously, the pragmatic approach would be to make macOS a kickass first-class citizen for OpenGL and Vulkan apps. It seems incredibly wasteful to nerf Apple GPU performance so over some misguided pride, or delusion that anybody making cross-platform tools will care about Metal
Not only that, but the mere suggestion that Apple intended to deprecate OpenGL burned a lot of bridges, and there is zero indication that Apple is even starting to repair that reputational damage, years later
As far as I can tell, there is no way to open a Mac Studio. Which isn’t a surprise as far as RAM or GPU upgrades - we knew Apple would never allow that… but the thermals on this machine are super important and we can’t even clean out the fans every few months
The M1 Ultra Mac Studio also produces a constant low but clearly audible fan noise.. which I hate to admit.. is fairly irritating and very unexpected. Especially when it’s literally doing nothing.
Max Tech (via Hacker News):
In this video, we do a complete teardown of Apple’s Mac Studio and we get it down to the core. We show off the internal components, the new copper heatsink and cooling system, the speaker, the power supply and much more!
The Mac Studio’s SSDs are on little PCIe cards, and there are two slots inside for future upgradability
This article explains, in detail, how Apple has chosen to structure its SSDs (the controller is in the M1 system on a chip) and why Apple has chosen to pair chips to individual systems for security reasons.
Apple only rarely introduces an entirely new Mac, but two executives working on the Mac Studio say that its design has its roots in Apple’s work over decades. In a new interview with GQ magazine, vice president of hardware engineering Kate Bergeron traces it back to the first Mac she worked on, the 17-inch PowerBook G4.
ok but can your Mac Studio do this?
Update (2022-04-13): Juli Clover:
iFixit today disassembled Apple’s new M1 Max Mac Studio, giving us a first glimpse at the components inside the machine.
Shortly after the announcement of the Mac Studio and Studio Display at the Peak Performance Apple event, we had the pleasure of talking to three Apple executives about the new machine and how the creative pro would benefit.
So, the Mac Studio definitely isn’t what Apple markets it as (a “modular computer”), but I think I can safely say that it is the computer I’ve been wanting them to make for a decade now: the guts of a MacBook Pro, but with the benefits of a Desktop that I care about.
A lot of my praise for the Mac Studio could perhaps be seen as just finally escaping a host of issues that have plagued the MacBook line forever.
A good example is how the Mac Studio has a ton of ports and ethernet, which while great in their own right, more importantly allow me to get rid of flakey TB hubs. I’ve never had TB or USB-C hub that hasn’t caused weird crashes or boot after sleep issues.
Another example is how my external display finally behaves reasonably. Even if I permanently attached my MBP to a monitor in clamshell mode and never undocked, my windows would always get confused when waking, as if they had readjusted to the MBP screen size during sleep.
Update (2022-04-28): Juli Clover:
Some Mac Studio owners have noticed that their machines are making a high-pitched “whining” sound that appears to be coming from the fan. There are pages of complaints on the MacRumors forums about the issue, and it seems to be affecting a number of users.
Craig Hunter (via John Gruber, Hacker News):
Performance of the Mac Studio is actually off the chart here. Before it went off the chart, the Mac Studio’s 6-core performance caught up to the full 28-core performance of the Mac Pro and then surpassed it.
[…]
Now we see that the Mac Studio shot up to just over 180 GFlops performance on 16 cores (using only performance cores), over twice the performance that the 2019 Mac Pro reached with 28 cores. What’s more, the slope of the performance curve is dramatically different. Whereas the 2019 Mac Pro labored to add additional performance past about 14 cores, the Mac Studio shows very little fall off, indicating that the 16 cores had plenty of bandwidth for memory access and parallel communication without competing with each other very much.
Update (2022-06-02): Maxwell Swadling:
i hate how I have to unplug and replug my usb c displays every couple days into the Mac studio. Still beats it kernal panicing at night with intel, but you would think apple could work out how an external display works by 2022
Update (2022-09-09): Sami Fathi:
Nearly six months after launch, the Mac Studio continues to face significant delays on Apple’s website, with M1 Ultra configurations facing up to a 10-week delay.
14 Comments RSS · Twitter · Mastodon
> I don’t know how well Xcode scales to lots of CPU cores
In all my Swift projects of varying complexity, Xcode compiles do not scale to even a few CPU cores. I never get a full load during compiles, whether i7 or M1 - it usually goes ~1/2 of CPUs if load is high. In my experience Swift builds are completely I/O bound, e.g. I got a ~2x speedup when going from the i7 Mini to the M1 Mini, but I put that solely on the faster SSD.
So the most interesting thing about the Studio to me are the SSD speedz, which might give a very nice bump.
P.S.: I'd have to get a long TB cable, to get that ugly thing out of my view ....
Didn't try that yet (Xcode 13.2 final is unstable enough for me), but my expectations are low. I actually talked to the involved engineer which I happen to know, but don't remember whether it'll actually improve the situation or not. I can report back once 13.3 left RC.
@Helge That’s interesting. I’d just assumed that the reported better compile times with the M1 Max vs. the base M1 Pro were due to it having more CPU cores. But maybe it’s instead due to better memory bandwidth? It is conspicuous that Apple is not highlighting Xcode benchmarks for the Mac Studio.
I was also wondering whether the memory bandwidth would help a lot. We have a small internal speed test compiling Noze.io, and with that the M1 Max isn't faster than the M1 Pro:
Silver Mac Mini M1 2020 2020 M1 8 16 4266 512GB 11.0.1 12.2 5.3 AppKitDebug 0m 20s Silver MacBook Pro 14" 2021 M1 Max 10 32 12.0.1 13.1 5.5.1 AppKitDebug 0m 12.96s Silver MacBook Pro 14" 2021 M1 Pro 10 16 12.0.1 13.1 5.5.1 AppKitDebug 0m 12.94s Space Gray MacBook Pro 14" 2021 M1 Pro 10 32 4 TB 12.1 13.2 AppKitDebug 13.37s
Surprisingly, color doesn't matter either (though there is no red machine yet).
This also shows the SSD speed diff, IMO. I think my Mini does ~3GB/s read, while the Studio (like the MBPs I think) are supposed to do over 7GB/s. (20s on my Mini, 13s on the MBPs)
@Helge This seems to support my initial guess that the Pro/Max difference was due to cores, since your M1 Pros have 10 cores rather than 8 and perform like the 10-core Max.
In all my Swift projects of varying complexity, Xcode compiles do not scale to even a few CPU cores.
It seems to me your table below contradicts this?
the M1 Max isn’t faster than the M1 Pro
Right, it shouldn’t be. The CPU cores are the same. There are more GPU cores, but those don’t matter for compilation. There is more memory bandwidth, but that, too, is mostly relevant for the GPU.
Basically, the M1 has memory bandwidth of about 67 GiB/s; the Pro has 200; the Max has 400; the Ultra has 800. However, a single CPU core “only” continuously about 102 GiB/s of that. Ten cores only use about 243 GiB/s.
Therefore, the CPU only benefits slightly from the Max’s higher ceiling, and they likely wouldn’t benefit from the Ultra’s at all. The high memory bandwidth is mostly there for the GPU.
Still, I would gather from your table that switching from 4 + 4 cores (M1) to 8 + 2 (M1 Pro) does benefit you when compiling Swift. (Part of it could also be the SSD, yes. Hard to say.) The M1 Max, unsurprisingly, is largely pointless for your use case. The M1 Ultra will be as well.
@Sören This is why I thought it would make sense to have a Mac mini (or Studio or iMac) with an M1 Pro, but they don’t make that.
@ Michael: I agree that there's a gap between the ~$1,000 Mac mini and $2,000-and-up Mac Studio. I imagine it will shrink later this year when the Mac mini M2 arrives, especially if that one (which wouldn't surprise me) increases the RAM ceiling to 32 GB. And, who knows, maybe they'll offer that in M1 Pro form as well.
And, yes, I think if you want a desktop, and you want to do software development, the M1 Pro is the sweet spot — and Apple isn't currently offering that.
[ I'm so much happier to live in a "the mini is slightly underpowered and the Studio overpowered" world than the one from just a few years back where it was rather "the mini hasn't been updated in literally a thousand days, neither has the Pro, let's not even talk about the laptop keyboard situation, and who knows if Apple even wants to make any computer like this at all ever again", though. ]
Again, I attribute the speedup to the significantly faster SSD, not the marginal 20% in extra CPUs (which wouldn't reflect a 80?% speedup). In part (the speed notes don't show this), because a clean build does not seem to saturate the CPU (just looking at the activity monitor, none of those qualify as scientific tests).
To verify I'm not crazy, I just clean-built Shrugs (a little bigger app), and while it starts out at full load the first 2 seconds or so, the load is about 50% or less for the remainder (55s in total, pure Swift).
Though maybe that looks like this because the efficiency cores really are not going to be used at all? 🤔
the marginal 20% in extra CPUs (which wouldn’t reflect a 80?% speedup)
It’s not 20%. The M1 has four performance cores and four efficiency cores.
The M1 Pro and Max have eight performance cores and two efficiency cores.
So, an 80% speedup is actually quite plausible.
However, Geekbench seems to show more of a 65% speedup: https://browser.geekbench.com/macs/macbook-pro-16-inch-2021-apple-m1-max vs. https://browser.geekbench.com/macs/mac-mini-late-2020
So, it could be a combination of that, and the SSD.
Though maybe that looks like this because the efficiency cores really are not going to be used at all?
I think there are commands to force processes to not run in the efficiency cores, but my Google-fu is failing me.
@ Julian: nah. The Mac mini and Mac Pro both have USB-A ports, and this is a device in between. A Mac midi, if you will.
Now, if an "iMac Pro" should ever happen again, it's anyone's guess if that will have USB-A. The 24-inch iMac is too thin to make that feasible. So, will they make the iMac Pro much thicker? Will they add more ports to the power brick? Will they add a USB-A port on the side (where there is more depth)?
@Helge: Building a Swift project I get full CPU utilization on my M1 Mac: https://imgur.com/a/7DzqEfP
