The brand-new M1 processor available in Apple’s latest MacBook Pro, MacBook Air, and Mac mini offers a groundbreaking blend of performance and efficiency, at least in theory. Lightning-quick app launches, speedy video transcodes, and the potential for as much as 29 hours of battery life? Yes, please.
But Apple’s M1 chip doesn’t live up to its full potential all of the time, at least not yet. Mac software has spent more than 15 years running on Intel x86 processors. And that means that a lot of apps that have been around for a while will need to be redesigned to take full advantage of the entirely new CPU architecture that the M1 uses. In some cases, these apps aren’t much faster on an M1 Mac than they are on an Intel Mac. This is because apps designed for Intel processors run in emulation mode on M1 processors, a situation akin to towing your car instead of driving it. This mode is called "Rosetta 2." (The original Rosetta came into play during Apple's ages-ago transition from PowerPC chips to Intel.)
All of this complicates the decision of whether or not to upgrade to an M1 Mac from an older Intel one. And if you’re open to buying a Windows laptop, choosing between an M1 MacBook and a competitor like a late-model Dell XPS 13 is just as tricky, since Intel-powered Windows 10 PCs have no architecture-compatibility issues between their software and processors.
So let’s take a look at some common computing scenarios to see how the M1 Mac improves (or doesn’t improve) performance.
Basic Productivity Tasks: Apple M1 Aces These
If you’re using a computer to browse the internet, answer email, and organize and edit photos, an M1 Mac will perform just fine. The reason why: On a Mac, many of these tasks are performed using apps that are part of macOS, and thus already run natively on the M1 chip. This everyday-use brigade includes the newly launched "Big Sur" version of the macOS operating system, as well as Apple apps like the Safari web browser, Preview, and the Pages word processing program. (Hit the link for lots more on Big Sur and the changes it brings to the macOS world.)
Indeed, in our tests of the MacBook Pro, the MacBook Air, and the Mac mini, all equipped with the M1 chip, we encountered no significant sluggishness on basic computing tasks using Universal apps, which is Apple's lingo for programs that run natively on the M1 processor. This is encouraging for many prospective M1 Mac owners, since nearly all of them will use their Macs to do these types of tasks at least some of the time.
It’s also not necessarily a foregone conclusion. Previous transitions between processor architectures have not always proceeded smoothly. The current effort to translate Intel-native Windows apps to run on ARM-based processors like Qualcomm's Snapdragons has resulted in PCs that are sometimes sluggish even when running apps that have already been converted. Apple has extensive experience managing silicon transitions, including the migration from PowerPC processors to Intel ones in the mid-2000s. It's nice to see that M1 Macs aren’t at a disadvantage when it comes to basic, essential computing tasks.
It’s difficult to quantify performance on basic computing tasks, but a few proprietary benchmarks out there run natively on the M1 processor and let us attempt to do it. One is Geekbench 5, which tests a processor’s single-core and multi-core power using a bevy of simulated tasks like checking email and playing music. Geekbench also measures how well a processor is able to handle the ever-increasing number of apps that use machine learning to speed up processing.
The Geekbench 5 results suggest that M1-powered Macs are not only faster than their Intel predecessors when it comes to basic computing tasks, but also faster than Apple’s A-series chips for iPhones and iPads, as well as Intel’s latest 11th Generation “Tiger Lake” processors. You can see the Geekbench results on the first tab the chart below.
The chart also includes results from three browser-based benchmarks. These perform many of the same types of workflow simulations that Geekbench does, but rather than testing how well the M1 handles them directly, they instead evaluate how well a web browser can perform them using common software platforms like JavaScript. Of course, web browsers are dependent upon a system’s processor, memory, and other computing resources. So the WebXPRT 3, JetStream 2, and Basemark Web 3.0 tests give us a comprehensive, if proximate, look at basic performance with in-browser tasks as a way to double-check the Geekbench results.
In most cases, the three browser benchmarks still show the M1 Macs outperforming iPhones, iPads, and Windows 10 PCs, but by a much smaller margin. The JetStream 2 and Basemark Web 3.0 tests actually show the iPad Air and Dell XPS 13 outperforming the M1 Macs.
Demanding Workflows: Rosetta 2 Could Be a Hindrance
If you plan to use your new Mac for more than just the basics, assessing the M1’s performance capabilities gets more complicated. In most cases, this complication is thanks to the emulation mentioned earlier, Rosetta 2.
Let's start with video encoding. This is a common workflow for multimedia professionals. Using an older app like Handbrake 1.1 to encode a video on an M1 processor results in decent performance—around 17 minutes to transcode a 4K video to 1080p. That’s actually a bit slower than what the Intel-powered 13-inch MacBook Pro can manage, but not significantly so. But Handbrake 1.1 runs in the Rosetta 2 emulation layer, and when you switch to the native Handbrake 1.4 beta, encoding times improve dramatically. You can see the contrast in the chart below.
The situation is similar with another resource-intensive workflow: 3D rendering. This is the sort of task that game developers and special-effects artists frequently perform, and one that we can approximate using Maxon’s Cinebench benchmarks. The older Cinebench R15 version, running in Rosetta 2, sees speedy performance on the M1 Macs. They clearly outperform their Intel predecessors, and they are slightly faster than similar Intel-powered Windows PCs. The 16-Inch MacBook Pro is faster still, with its beefy Core i9 processor. On the latest Cinebench R23 version, though, the M1 Macs outshine the others we were able to retest by wide margins. R23 is native.
Other multimedia workflows like image editing will likely follow a similar pattern, but some of the most common apps that photographers use aren’t available in M1 native versions yet. They include Adobe Photoshop. We used the current non-M1-native version to apply a series of 10 filters to a standard JPG test image. The M1 Macs were competitive but slightly slower than some Intel Tiger Lake-based laptops in this test.
We don’t yet have access to an Apple Silicon-native version of Photoshop to see how much better it performs, though Adobe says such a version will be available early next year. So if you mainly use a Mac for Adobe Creative Suite apps, it may be worth waiting to upgrade to an M1-powered one until your key applications debut in native flavors.
Gaming on the Apple M1: It’s Complicated
Perhaps the most nuanced part of the Apple M1’s performance is how it handles graphics tasks and complex 3D gaming. It’s also the area in which the M1-powered Macs also offer some of the least groundbreaking results.
AAA gaming isn't a core competency of Macs in general. Pre-M1 Macs, barring high-end models equipped with AMD Radeon graphics for pro content creators, were never known as gaming machines, and the gaming development community has treated the Mac as such. There's no point in porting a critical mass of demanding games to the Mac, when the installed base of robust GPUs on Macs is so small. Hardcore gamers typically like Windows-powered desktops and laptops with beefy GPUs intended for gaming, and the arrival of the M1 probably won’t change this situation.
Like everyday computing performance, gaming prowess can only be approximated with benchmarks, and sometimes the results are quite different than how real-world games actually perform. This is the case with one of the few popularly accepted graphics benchmarks that does run natively on M1 chips, GFXBench 5 Metal. This test’s relatively forgiving Car Chase scene, performed at 1080p (full HD) resolution, does paint the M1 Macs in a positive light, as you can see on the first tab of the chart below...
The Mac mini and MacBook Pro even perform better on this test than does the HP Spectre x360 15, which is equipped with an Intel Core i7 Tiger Lake CPU and a dedicated Nvidia GeForce GTX 1650 Ti GPU. The M1 Macs also perform well when you turn up the resolution to 1440p and run the more demanding Aztec scene in GFXBench.
Testing actual AAA games tells a slightly different story. We selected two demanding, popular titles (Rise of the Tomb Raider, and Total War II: Warhammer) that both use Apple’s latest Metal graphics API, but have not yet been optimized to run natively on the M1 chip. This situation is representative of what most gamers can expect on current Mac-supported games if they switch to an M1 Mac.
The results don't parallel what we saw on GFXBench, but there are signs of optimism. The M1 Macs can certainly handle Rise of the Tomb Raider at the lowest graphics-detail setting, posting results of 58 frames per second (fps) or greater. That handily beats the 37fps that the XPS 13 manages, equipped with Intel’s latest Iris Xe integrated graphics.
But turn the graphics settings up to the high preset, and the M1 performance dives to around 40fps, which might be acceptable to some casual gamers but is well below the 56fps that the Nvidia GeForce GTX-equipped Spectre x360 offers at that same setting.
A similar dynamic is obvious on Total War II’s Skaven benchmark. Emulation accounts for part of the performance discrepancy between these two games and GFXBench—the GFXBench benchmark runs natively, while the games do not. Mainly, the gap highlights the fact that 3D graphics performance in synthetic benchmarks isn't necessarily the same as performance on actual games.
It will be interesting to see how things improve once more titles are re-released to run natively on the M1. Early reports from titles that do run natively, like World of Warcraft, indicate decent performance that's comparable to or slightly better than what Iris Xe integrated graphics offer. But everything we've seen so far suggests that entry-level dedicated gaming GPUs like the GeForce GTX 1650 and GeForce GTX 1650 Ti are still likely to offer more robust performance than the M1 going forward, even once more games are able to run natively.
M1 Performance: What the Future Holds
Overall, the theoretical performance of Apple’s first in-house Mac processor looks remarkable, especially considering this is Apple's first shot at the computing, as opposed to handheld and tablet, platform. But as we’ve seen, not every user will experience that full potential currently, in every situation. If you frequently perform demanding computing workflows using apps that haven’t yet been optimized for Apple Silicon, you should wait for one of two developments on the not-to-distant horizon.
The first and most obvious is the arrival of more M1-native applications. Most third-party developers of apps that haven’t yet been optimized are working to get the job done. Once you update your current Mac to Big Sur, you can use the System Information app to see which of your currently installed programs are Universal apps that will run natively on the M1.
The second development to watch for is the inevitable release of additional and more powerful Apple processors. The M1 name itself suggests that more processors will be coming, and we expect Apple Silicon versions of the iMac and the 16-inch MacBook Pro to show up before too long. These should be able to handle demanding workflows even better.
On the other hand, if you’re a basic computer user who sticks to tasks that don’t require much processing power, and most of what you do is handled by the utilities and programs included with macOS, an M1 Mac fits the bill right now. Just be aware that the performance of late-model Intel Macs is nothing to sneeze at, so you may not notice a big difference unless you’re upgrading from a three-to-five-year-old machine. For more, see our guide to upgrading to an M1 Mac, and check out our full reviews of the M1-powered MacBook Pro, MacBook Air, and Mac mini.
Further Reading
Processor Reviews
Processor Best Picks
The Link LonkNovember 20, 2020 at 01:43AM
https://ift.tt/2UFh8Ot
How Fast Is Apple's M1 Chip? It Depends on the App - PCMag
https://ift.tt/2RGyUAH
Chips
No comments:
Post a Comment