Apple is crazy, right? The Mac just had its best year of sales ever, and Cupertino is hitting the platform with a shock like it hasn’t had in nearly 15 years—back in a time when the Mac was not having such a good year. Apple is beginning the process of replacing industry-standard Intel chips with its own, custom-designed silicon.
In a way, we’re not just reviewing the new Mac mini—a Mac mini is always a Mac mini, right? We’re reviewing an ARM-based Mac for the first time. And this is not exactly the same story as all the other ARM machines we’ve looked at before, like Windows 10 on ARM—a respectable option with some serious tradeoffs.
Sure, longer battery life and quick waking from sleep are already out there on other ARM computers. But as you may have seen in our hands-on earlier this week, what we’re encountering here is also a performance leap—and as you’ll also see in this review, a remarkable success at making this new architecture compatible with a large library of what could now, suddenly, be called legacy Mac software.
Not everything is perfect; we’ll talk about iOS apps on the Mac and some other problems, too. But if this Mac mini proves anything, it’s that Apple was not, in fact, crazy. The M1 makes Apple’s strategy seem soberingly sane.
Table of Contents
Apple Mac Mini (2020) with Apple M1 Chip
The big news specs-wise, of course, is the M1 system-on-a-chip. The M1 is an 8-core CPU with four performance and four efficiency cores, as well as an 8-core GPU and a 16-core Neural Engine for on-device machine-learning tasks.
Based on an ARM instruction set, it’s the first Apple-designed CPU and GPU for the Mac. For well over a decade, Apple’s laptops have had Intel CPUs and either Intel, Nvidia, or AMD graphics. The transition away from that status quo starts here. There’s much more to talk about on that front, of course—but let’s get the other specs out of the way first.
By default, the Mac mini ships with 8GB of RAM, but that can be upgraded to 16GB. That’s a lot less than the 64GB cap in the Intel Mac mini, but, for now, that Intel Mac mini still exists in Apple’s lineup.
Currently, Apple has only replaced its very bottom-end machines with Apple Silicon variants. It looks like we’re going to be waiting a while before we get beefier versions with more RAM and more ports (the M1 Mac mini and 13-inch MacBook Pro each only have two Thunderbolt ports, rather than four)—which is a very real bummer for a lot of people.
Likewise, solid-state storage starts at 256GB, but you can go to 512GB, or even 1TB or 2TB. Doubling the RAM adds $ 200 to the purchase price, while going to 2TB of storage from the base 256GB configuration more than doubles the cost of the device. That storage bump is the main reason our review unit’s purchase price is so much higher than the base.
|Specs at a glance: 2020 Mac mini|
|OS||macOS Big Sur 11.0.1|
|Networking||Wi-Fi 6; Bluetooth 5.0|
|Ports||2x Thunderbolt 3, 2x USB-A, 3.5mm headphone, 4x USB-A, HDMI, gigabit Ethernet|
|Warranty||1 year, or 3 years with AppleCare+|
|Price as reviewed||$ 1,699|
The Mac mini has a built-in speaker—which might be a little surprising for a computer like this—but it does not have a built-in microphone. The speaker is bad, frankly; it sounds like an old MacBook Air speaker with the lid closed. There is a 3.5mm headphone jack, though. Other ports include two Thunderbolt 3/USB 4.0, two USB-A, Ethernet, and HDMI.
In terms of wireless, you get Wi-Fi 6 and Bluetooth 5.0.
The box doesn’t include much—just a power cord and the computer itself. You’ll obviously need to buy a display, a keyboard, a mouse, a microphone, and headphones separately. Again, the Apple adage that the cost balloons as you add essentials holds true here as ever.
As the name suggests, the Mac mini has a very low footprint. It measures 1.4×7.7×7.7 inches (3.6×19.7×19.7 centimeters) and weighs 2.6 pounds (1.2 kilograms).
It comes in two default configurations. The first starts at $ 699 and includes 8GB of RAM and 256GB of storage. The second ups the starting price to $ 899 and simply boosts the storage to 512GB.
The base configuration here is $ 100 less than the previous Mac mini, which is nice to see. Again, Apple still sells an Intel-based Mac mini alongside this one, with a 6-core 3.0GHz Intel Core i5, Intel UHD 630 graphics, 8GB of RAM, and 512GB of solid-state storage. Nothing has changed for the Intel version of the device this year, though, so we won’t be going into that.
If you liked how the last Mac mini looked, you’ll like this one, too: nothing has changed. For that reason, we won’t spend too much time on the aesthetics in this review. As before, the Mac mini prioritizes a very low profile. It’s essentially a laptop without a screen inside a 7.7×7.7-inch square.
It looks nice but unassuming. It has that classic Mac silver color, whereas its predecessor was gray. The ports are all on the back, so it should play nice with most cable management solutions.
A few frustrating limitations
Before we get into the good news—basically anything related to M1 performance and software support—let’s go over some really frustrating choices Apple has made regarding ports and peripherals.
As noted above, the RAM and storage configurations don’t reach as far as the Intel Mac mini’s do, and that also extends to ports. Two Thunderbolt ports is just okay, to be honest, even though there are also two USB-A ports on top of that. At least it’s not the new M1 MacBook Air or 13-inch MacBook Pro, which are both limited to two Thunderbolt 3 ports total.
The Mac mini can only drive two displays at a time, and one of those has to be over HDMI. Since that HDMI port is HDMI 2.0, it doesn’t have the throughput to handle 4K at 120Hz, or 8K at 60Hz. There aren’t many monitors like that now, but there increasingly will be over the next few years. Generally, we like our machines to stand the test of time.
Neither the Mac mini nor any of its M1 brethren supports external GPUs. That’s pretty disappointing, and it calls into question Apple’s push for eGPUs over the past few years.
It wasn’t long ago at all that Apple was signaling that eGPUs were the future of Mac graphics performance. Now it’s not clear that they’ll be part of the Mac experience at all.
Additionally, some people have used the Mac mini as a file server. Many of them are probably going to want to stick with Intel for now, as the M1 Mac mini only has a gigabit Ethernet port, whereas the Intel Mac mini is configurable to have a 10Gb port.
Apple sees this initial volley of Apple Silicon devices as the bottom end of its lineup. So we’re in a very odd situation right now where if you want the cutting edge of performance, you have to choose low-end Mac configuration options. If you want lots of ports and RAM, you have to stick with Intel for now. That’s very likely going to change with the introduction of more expensive Macs with a faster, hypothetical M1X chip or something like that—but we have no idea yet when that’s going to happen. We only know that, eventually, it will.
With that out of the way, it’s time for the good stuff.
As noted above, the Mac mini (and its new MacBook Air and 13-inch MacBook Pro siblings) has Apple’s M1 system-on-a-chip, which includes an 8-core GPU, a CPU with four performance and four efficiency cores, a 16-core neural processing unit (NPU) called the Neural Engine, and a whole bunch of other stuff.
Built on the ARM Instruction Set Architecture (ARM ISA), the M1 features 16 billion transistors and was manufactured in a 5nm process. According to Apple, each performance core in the M1 qualifies as the world’s fastest CPU core to date, while the efficiency cores match the performance of some recent Intel Macs.
We read each of the four performance cores as having a clock speed of 3.2GHz, and while the iPhone and iPad’s A14 chip has 8MB of L2 cache, the M1’s performance cores get 12MB. Unlike some previous chip designs, all the performance and efficiency cores can be utilized at once, though there are signs that it gets a little more complicated when it comes to the cache.
Apple claims that the M1 can achieve its strong performance in part because of its unified memory architecture (UMA), which allows the CPU and GPU to both easily access relevant data without having to slow things down by copying it around.
We’ll talk specific performance testing and results soon, but spoiler alert: the M1 is quite fast. That’s especially true for graphics compared to Intel’s graphics solutions (which seem unworthy to even be mentioned in the same category as what the M1 offers). These improvements are thanks to all of the above, plus techniques like tile-based deferred rendering and Apple’s proprietary Metal graphics API, which has been designed to take advantage of this architecture.
This has gotten less attention, but the M1 contains a bunch of other stuff besides the elephants-in-the-die that are the CPU, GPU, and NPU. It has the Secure Enclave, Apple’s encrypted tool for handling sensitive data on device. It has an image signal processor, which isn’t super relevant on the camera-less Mac mini, but it reportedly improves FaceTime camera quality on the laptops. The M1 also includes a storage controller and hardware for driving encryption, among other things.
In 2017, Apple introduced the T2 chip on the iMac Pro, and it went to most other Macs over the next couple of years. The T2 handled security features and various other things like some of what we just listed above, and we speculated when it was first introduced that it might be a predecessor to Apple’s eventual Mac SoC plans.
It turns out we (and everyone else who picked up on that pretty obvious clue) were right. As such, the new M1 Macs don’t have T2 chips. It’s all on the M1 now.
Of course, a change in architecture suggests all sorts of compatibility headaches with older software, to say the least. The M1 can’t natively run apps made for Intel-based Macs. But surprisingly, that ends up not really mattering in most cases. A lot of buyers of the M1 Macs will never even realize anything changed under the hood.
To explore that point, let’s go over the software the M1 does run.
Native Apple Silicon apps
Apple promised that every single currently supported Apple-made app would become a universal app with full native Apple Silicon support, and the company delivered exactly that. From high-use apps like Safari and Mail all the way down to AirPort Utility and Chess, it’s all universal in Big Sur, and it all runs natively on this Mac mini’s M1 chip.
More apps from third parties are coming. Obviously, Adobe is critically important to many Mac users. As we’ll see in the performance section, apps like Premiere are already running just fine in Rosetta. But Apple Silicon variants are on the way; in fact, Adobe has already released a beta version of Photoshop optimized for the new architecture.
Others like Microsoft and Unity have announced similar intentions. Others still, like Autodesk Maya, remain a question mark, though many of those work just fine via Rosetta 2 (which we’re getting to, promise).
Apple’s apps, along with various others, are distributed as universal binaries—meaning you download one file, but it works on either Intel or Apple Silicon Macs. Given that Apple is expected to continue supporting Intel Macs for several years to come and that people will still be using them for a long, long time, this is going to be the status quo for the foreseeable future.
We’ll get into performance benchmarks, but usually, we’ve found there’s not an enormous difference in performance on the M1 between apps optimized for Apple Silicon and legacy software. That said, the Mac has an extensive library that includes software that hasn’t been updated in quite some time. (Though there was a culling when Apple dropped support for 32-bit apps last year.) To make this transition anything close to graceful, Apple has to ensure that legacy x86 apps will run acceptably on the new hardware.
So let’s get into that, because it’s probably the biggest concern on most prospective buyers’ minds.
Running x86 apps
There are really two questions when we talk about legacy apps on the M1. The first and most important is running macOS apps designed for Intel-based Macs, and the second, more niche consideration is running apps designed for the x86-64 architecture on other operating systems like Windows. Let’s start with the Mac, and that means talking about Rosetta.
Every now and then, the Hollywood studio system makes a sequel to a movie that came out a long, long time ago. It hasn’t usually gone well.
Rosetta in macOS Big Sur is such a sequel, but it’s for the most part a phenomenal success. It’s better than the original. The last time the Rosetta name appeared, it was there to help PowerPC applications run on Intel Macs after Apple made the last architecture transition, back in 2006. Now, as Apple leaves Intel behind, Rosetta 2 is here to make apps made for Intel Macs run on the new Apple Silicon ones, like the Mac mini with the M1.
Rosetta 2 translates apps developed for x86 macOS to work on the M1 or other Apple Silicon CPUs, which are not capable of natively running said apps. It does this ahead of time, not on the fly—meaning when you install the app or first launch it (depending on the installation method) it is recompiled for the new architecture. Further, the app is automatically recompiled again as Apple releases relevant software updates.
There are some limitations, though. Ars Technica Editor Emeritus Andrew Cunningham summarized them succinctly in our recently published macOS Big Sur review:
There are a few things that Rosetta 2 won’t do. Rosetta 2 won’t translate kernel extensions, x86 virtualization software like Parallels or VMWare Fusion, or apps that use newer Intel instruction sets like AVX, AVX2, and AVX512. For apps that already know to check for AVX support, since those instruction sets aren’t available on all Macs that run Big Sur, the apps should see that the instructions aren’t supported and run without issue.
Those exceptions aside, the results are near-perfect. You’ll be prompted to install Rosetta when you first launch a legacy app, and that installation just takes a few seconds. After that, and in almost all cases, the user might never know whether an app was originally native to this hardware or not. The only common clue is that the translated Intel apps might take a few seconds longer to launch from the dock for the first time.
We installed and tested 120 apps—a wide variety of types and inclusive of both Mac App Store apps and Web downloads—and almost none of them failed to launch, ran noticeably slowly, or produced errors that impacted usage. The few exceptions were all 3D games downloaded on Steam, which mostly worked but with problems in a couple of titles.
We did run into a few very minor UI irregularities in a couple of apps that already didn’t follow normal macOS design conventions, but that’s the full extent of the problems in mainstream use cases.
Now, things get a bit more complicated when you go deeper than most users will go—specifically, command-line tools. At first, I was frustrated that the Homebrew package manager fails with a note that it only works on Intel machines. But I later realized that it works just fine when Terminal, which is a universal app, is forced to run in Rosetta. Most users don’t even know what this is, though—plus, it might end up working outside of Rosetta in the near future, anyway.
What about Windows?
This should go without saying, but Boot Camp—the utility that allowed users to dual-boot x86 Windows alongside macOS on Intel-based Macs—is no more. That would be OK if virtualization was also an option, but it’s not at the moment.
To me personally, this is the biggest impediment to making the leap to Apple Silicon. I do multi-platform game development, primarily in Unity, but I often execute and test my x86 Windows builds (obviously a hugely important platform for games) in a Windows virtual machine via Parallels. Furthermore, many utilities important to game development are only available on Windows.
This isn’t a problem on my Intel MacBook Pro; these utilities generally run just fine in Parallels. But it’s a big problem on this Mac mini, or any other M1 device.
Parallels and some of its competitors have announced plans to build native versions of their virtual machine software for Apple Silicon and the M1, but the timeline is unclear, and most of the talk so far has been about running ARM-based Linux or Windows.
A couple of people close to this kind of work have told me they expect that x86 Windows emulation will come to Apple Silicon Macs but that it might be a long road ahead. (Not even DOSBox works right now—goodbye, sprawling GOG library.)
Virtual machines have gotten so good in the course of an iterative march of improvements over many years. While some work can be carried over, some can’t, and it may take a while to get back to where we were in some respects. So no, not everyone is a game developer. But there are numerous other people in technology fields who need Windows to run on their Macs, too, and for now, it’s a question mark. If you’re one of those people, I wouldn’t recommend starting the transition until we know more.
Running iOS and iPadOS apps
There’s one marketing message you might hear a lot from Apple about the new Macs besides performance and battery life: for the first time ever, macOS can natively run iOS and iPadOS apps. Further, thousands of these apps are already available in the Mac App Store.
The Mac has long had some very high-quality apps, but that library hasn’t been as robust or vast in breadth as the iOS or Windows software ecosystems. In theory, this change opens some very big floodgates.
Apple laid some groundwork for this not long ago with Project Catalyst, which gave developers tools to fairly easily adapt touch-oriented iPad apps to a mouse-and-keyboard interface, provided those apps were built to Apple’s suggested spec already.
Now, developers’ apps go through an automated screening process to determine if they depend on technologies not available on the Mac, like TrueDepth or multi-touch. If they pass this, they’re put on the App Store. If they don’t, Apple emails the developer to let them know what they need to do to make the app compliant and get it listed. Further, Apple employs some kind of human curation to give apps a special label in the App Store that indicates that they are verified to be usable on the platform.
Unfortunately, I must report that this iOS/iPadOS app capability needs a lot more time in the oven. The iPhone and iPad app experience in macOS is simply not good most of the time.
The simpler the app, the better it’s likely to work. Some Twitter clients are nice enough to use, alongside some shopping apps, but complex apps are hit and miss at best. Take HBO Max: the video plays in a square box that you can’t resize, and it doesn’t match the aspect ratio of the content. There are picture-in-picture and fullscreen buttons, but PIP does nothing, and fullscreen zooms in and crops out part of the video within the window without going into fullscreen at all. That Apple chose to feature this app is baffling. It’s a very bad user experience.
Elsewhere, macOS is known for being a particularly poor OS for games, whereas iOS has one of the most robust game libraries in history—so it should be a win on the gaming front, right? In some cases, sure. Some games that are straightforward tapping experiences, like Disco Zoo or Furistas Cat Cafe, are pretty nice to play. Twitch sensation Among Us (which has an x86 Mac version on Steam) is playable thanks to its straightforward click-to-move interface, but some menus appear cut off on the edge of the screen. (This app was also featured by Apple.)
And then there’s thatgamecompany’s Sky: Children of Light, one of the most critically acclaimed mobile exclusives out there. It uses complex touch gestures for camera and movement. Apple tries to get around this by offering a special interface (“Touch Alternative Mode”) for apps like this that allows you to employ replacements for these gestures, but they’re no substitute.
There are some neat apps in the App Store, but there’s so much junk in there that barely works, I think most users are going to just give up after trying a couple of bad ones—especially when some of those bad ones are apps that Apple has featured for being exemplary, somehow.
Now, having this capability at all is probably better than not having it, at least for some people. Notably, iOS/iPadOS developers could test their mobile apps right on the Mac without compiling a separate binary for the Simulator. Also, most of the apps look nice and are speedy.
But until Apple either figures out a way to better communicate to users what works well and what doesn’t, or developers take the steps necessary to adapt them fully to the Mac, it’s mostly just a mess. There’s so much potential here—I absolutely buy in to the vision. But the execution has a long way to go.
Okay, so iOS apps aren’t the best on this Apple Silicon Mac. But the real pitch here is performance, and it’s time to look at some benchmarks. Before we go over the results, let’s take a look at the specs of the different machines we’re testing here:
|Mac mini (M1)||Apple M1||Apple M1|
|Mac mini (Intel)||3GHz 6-core Intel Core i5||Intel UHD Graphics 630|
|2020 27-inch iMac||3.6GHz 10-core Intel Core i9||AMD Radeon Pro 5700 XT|
|2019 16-inch MacBook Pro||2.4GHz 8-core Intel Core i9||AMD Radeon Pro 5500M|
|2020 13-inch MacBook Pro (Intel)||2GHz 4-core Intel Core i5||Intel Iris Plus Graphics|
|2020 MacBook Air (Intel)||1.1GHz 4-core Intel Core i5||Intel Iris Plus Graphics|
|2020 iPad Pro||Apple A12Z||Apple A12Z|
|2020 iPad Air||Apple A14||Apple A14|
|iPhone 12||Apple A14||Apple A14|
|iPhone 11 Pro||Apple A13||Apple A13|
The important takeaway from that table is that the Mac mini and other M1 devices replace computers at the low end of Apple’s lineup, but we’re here comparing the M1 against the most beefed up (or close to it) configurations of each computer (iMac, MacBook Pro, MacBook Air). In other words, we’re not comparing the M1 against the base spec for any of these Macs, or against older Macs, or against comparably priced Macs—we’re comparing it to the nearly speediest configurations of those Macs available this year, all of which are much more expensive than this Mac mini.
Now, let’s look at some results comparing the M1’s performance with Intel Macs and iOS/iPadOS devices.
The charts tell the story: the M1 Mac mini outperforms the MacBook Air, 13-inch MacBook Pro, and all the iPhones and iPads—in most cases by a substantial margin. It also leaves its own Intel-based predecessor in the dust. A single M1 high-performance core beats all of these CPUs in the single-core test—even the iMac.
The heavily specced-out iMac beats it pretty handily in the multi-core CPU test, but that’s a bit of an apples-and-oranges comparison, given that the iMac draws dramatically more power and is, well, quite a bit larger—not to mention 3.5 times more expensive (albeit with a pricy high-end display built in).
The biggest surprise we see here is the graphics performance. Performance-obsessed computing enthusiasts have long looked at integrated graphics with a certain amount of scorn. Intel’s integrated graphics get the job done for basic productivity tasks, but they fall on their faces pretty quickly when you throw heavy-duty 3D graphics or video tasks at them. This is technically an integrated GPU (though of a very different sort), and it betrays that stereotype.
The two 13-inch Macs we’re comparing against above have some of the fastest integrated graphics Intel offers, and the comparison is… well, you see it. It’s a massacre. In fact, the M1’s GPU is knocking on the door of the 16-inch MacBook Pro’s discrete graphics solution. And again: the 16-inch MacBook Pro (admittedly a laptop, but the Mac mini has mostly laptop-y guts anyway) costs more than twice as much.
And it’s quite solid for gaming, for what it is. It won’t satisfy gamers who want 120fps performance at 1440p resolution and high settings in any modern games, of course. But as a more casual gaming device working with the Mac’s admittedly small library of high-quality games, it works nicely. I loaded up World of Warcraft: Battle for Azeroth and found that this Mac mini provided much more stable performance and higher settings than does my own 15-inch MacBook Pro with discrete graphics—and that was before Blizzard patched the game to support Apple Silicon natively.
Now, everything we looked at above was native on Apple Silicon. At least for the first year or two, you’re likely to regularly be running several apps in Rosetta if this is your daily driver. So let’s compare native and Rosetta performance on the M1 Mac mini.
A Rosetta app definitely isn’t always as performant as a native app—and as far as the browser tests go, well, you know how Chrome is to begin with. 3D apps that use Metal exhibit very similar GPU performance regardless of whether we’re talking native or Rosetta, though. But these benchmarks still show Rosetta apps on the new Mac mini’s CPU beating Intel apps running natively on an Intel Mac mini, MacBook Air, or 13-inch MacBook Pro, per the previous Apple Silicon vs. Intel results.
Ars writer Jim Salter already came to similar conclusions in his hands-on (and while testing the M1 against non-Mac machines, to boot), so this won’t be a surprise to anyone following our coverage thus far. But just because we saw this already two days ago doesn’t make it any less amazing.
When I wrote about the iPad Pro’s A12X back in 2018, I called it astonishing and wondered aloud what something like that would be like in an actively cooled machine. Well, here you have it. It’s even more astonishing—though it bears mentioning that Geekbench and its ilk don’t reflect sustained performance (which active cooling would most benefit), only burst.
Now I’m left wondering something else aloud: what might an M1X designed to compete with discrete graphics in a 16-inch MacBook Pro or iMac look like? I don’t know, but the Mac platform has a fascinating road ahead of it, to say the least.
The Mac’s future is looking bright
If you thought Apple was crazy for moving away from Intel in the Mac, you weren’t alone. It sure seemed crazy. The Mac just had its best year of sales ever, and yet Apple couldn’t possibly rock that boat harder than this. Going into this, I was ready to see some strong performance, but I was skeptical that Apple Silicon would work for me.
I’m still not sure the Mac mini is an ideal machine in every respect because of its limited port and peripheral options, but amazingly, Apple didn’t rock the boat. The boat is faster, it’s longer lasting, it’s better—and it hardly took on any water at all.
Between the move to universal apps across Apple’s entire catalog, to promised native updates from various software-as-a-service companies like Adobe, to the performance and seamlessness of Rosetta 2, the majority of Mac users may never even have an inkling that a change of architecture has occurred.
They’ll just notice that their Macs are faster and cooler, get better battery life, gain machine-learning features over time—and I suppose that they can run iPhone and iPad apps, if said users aren’t picky about those apps being nice to use.
Tomorrow, we can (and should) go back to debating about walled gardens, App Store cuts, repairability, marketing slogans, China labor practices, and everything else. But today, technology enthusiasts have something really cool to celebrate. It’s nice to see that once in a while—and it will be fascinating to see how the industry moves forward from this point.
As the first of Apple’s Mac chips, and as one designed for its lowest-end devices, this is probably the slowest Mac silicon Apple will have ever made. But already, it is the beating heart of some of the fastest Macs produced to date.
In other words, the future of the Mac looks quite bright—at least in terms of hardware and performance. Apple has moved to a new architecture, free of the shackles of compatibility. And yet it did that without actually impacting compatibility in terms of the user experience, in almost all cases.
That’s the part that surprised us here at Ars. We figured the Apple Silicon move would bring performance benefits. We didn’t expect it all to go quite this smoothly right out of the gate.
For this machine, “think different” takes on a new and entirely literal meaning. Here’s to the crazy ones, indeed.
- Great performance across the board
- Legacy x86 macOS app support and performance are generally rock solid
- Low footprint in every respect: size, power, thermals
- Everything that was good about the Mac mini before is still good
- RAM and port options are limited
- The iOS and iPadOS app experience is a mess
- The path for emulating x86 Windows is still unclear, and that’s an essential part of a lot of people’s workflows
Listing image by Samuel Axon