fbpx

Rounding out 2020, Apple made the first move in its transition from Intel to Arm-based Apple silicon processors. The MacBook Air, the 13” MacBook Pro, and the Mac mini are the first Macs to feature the new chip.

Apple’s M1 processor is an integrated SoC (system on chip) design, which combines CPU, GPU, Neural Engine, cache, and RAM into a single integrated component. The M1 uses an 8-core CPU divided into four high-performance and four high-efficiency cores designed to balance performance, power demand, and heat load. It also includes an 8-core integrated GPU. The Neural Engine is a separate 16-core unit within the chip that’s optimized for machine learning tasks.

Initial versions of the M1 are offered with 8GB or 16GB of unified RAM. While this may appear a first-generation product, it in fact builds upon Apple’s years of hardware design, development, and implementation in iPhones and iPads. The M1 is already based on a chip architecture with an established track record.

FCP2020016 3macs

First impressions

My review unit included the base model M1-equipped Mac mini coupled with the LG UltraFine 4K 24” Display. These items are each $699 USD from the online Apple Store. At first I was a bit concerned about the mini having only 8GB of RAM and a 256GB SSD, but that really wasn’t much of an issue.

FCP2020016 front

The Mac mini comes configured with two Thunderbolt 3/USB-C/USB 4.0 ports, two USB-A ports, HDMI, ethernet, Bluetooth, and Wi-Fi 6. The Mac mini can support a Pro Display XDR at full 6K resolution, plus an additional 4K display over HDMI. The LG display I was using connected on one of the Thunderbolt ports and includes a Thunderbolt loop-through along with additional USB-C ports.

The M1 Mac mini with this LG 24” display form a very nice combo. The LG display comes the closest to the iMac Retina display of any of the third-party monitors that I’ve looked at. If you were buying a system for general computing use, then it would be hard to go wrong with these.

Real-world comparisons

I just bought a 2020 8-core 27” iMac and had weighed that decision against an M1 Mac mini purchase. The test comparisons that follow in this article are relative to that machine. I was curious whether or not these tests would validate that choice. Of course, this is a base model M1 Mac mini and I’m comparing it against an iMac that costs five times as much, which might seem a bit unfair. Last year I tested the decked out Intel Mac mini and the M1 blows it out of the water. So comparing “like” products doesn’t really seem valid either. I was also interested to see whether more RAM or a beefier GPU would make that much of a difference.

FCP2020016 perf

FCP2020016 chipThe M1 Mac mini is one of the best-performing small computers in this size or price range. To call it the Bruce Lee of computers is no hyperbole. It definitely outperforms its class and rivals “more powerful” and more expensive units. To compare purely based on numbers is misleading, which is why Apple has been silent about CPU clock speeds.

I have seen it written that this is a 3.2GHz chip, yet one of the benchmarks I used identified it as 2.4GHz. I presume that’s either an error or the way the CPU operates is different enough from the x86 architecture to make it a meaningless parameter.

Apple gains a lot of efficiency from the SoC design, so 8GB of unified RAM outperforms RAM on a conventional computer. Some have suggested that RAM on the M1 chip compares with double that amount due to efficiency. If that’s true, then 8GB and 16GB would equate to 16GB and 32GB on an Intel-based Mac. I’m no expert on the engineering design, so take that assessment with a grain of salt; however, based on my experience with the M1, it certainly feels close to being true.

FCP2020016 PxlML

The 16-core Neural Engine should not be ignored, because it handles machine learning functions. Modern software is increasingly adding machine learning features, so the Neural Engine takes that load off of the CPU and GPU. One example is Pixelmator Pro’s Super Resolution algorithm to resample images during enlargement, which is accelerated by the Neural Engine. This ML processing is significantly faster on the M1 mini than my older MacBook Pro.

The caveat is that this type of performance becomes evident with optimized software and media. To get the most out of the M1 chip, you need native applications that have been compiled for both Big Sur and the Apple silicon architecture. Any non-native application runs under Rosetta2 emulation. The first time you install a non-native application, Big Sur goes online, downloads and installs Rosetta2, becomes it doesn’t already come in the OS. This isn’t just with applications. For example, RED’s Apple workflow installation requires Rosetta2. Although that may only be the installer and not the component needed for REDCODE media itself.

Optimized media

Media is a critical factor. The M1 Mac mini flies with ProRes codecs in Final Cut Pro, even up to 8K resolution. Moving this media around is like a hot knife through butter. But the same is true on my iMac. I also installed Resolve 17.1 Beta (native) and Premiere Pro (not native). Both did well with this same test media. Likewise, RED 6K files played well, with smooth skimming in Final Cut Pro.

FCP2020016 resolve

Then I loaded up a recent commercial project of mine into Final Cut Pro, Resolve, and Premiere Pro. It’s a :60 grocery commercial, which was filmed with the Panasonic EVA1. That media is 4096 x 2160 encoded with a 10-bit 4:2:2 H.264 codec. While this media isn’t handled perfectly on an Intel Mac, it is generally smooth in all of the standard NLEs. These files were log-encoded, so the V-Log LUT was applied on all clips, plus some minor color correction.

Both Final Cut and Resolve on the M1 Mac Mini handled this media reasonably well, though not as fluidly as ProRes. However, the Premiere Pro timeline was very sluggish with it. Worse than on the iMac. Obviously the combination of a non-native application with less-than-optimized media can be a challenge. (Adobe launched native support for the M1 chipset on Dec. 21 with public beta versions of Premiere Pro, Premiere Rush, and Audition. This announcement was made after my M1 tested was completed.)

Performance testing

I tested a variety of project types, both for playback and export times. Media was played from an external Glyph Thunderbolt 3 NVMe drive, which delivers comparable read/write times to the internal SSD. My tests included Final Cut Pro, Premiere Pro, Resolve, After Effects, and Motion in side-by-side comparisons between the M1 Mac mini and my 8-core iMac (with 96GB RAM and the 16GB 5700XT GPU). The Glyph drive provided a testing constant.

You can review the chart for the render times. No surprise, the iMac had faster render times than the M1. Sometimes by a little and other times by a lot. My gut feeling is that the difference is largely due to the AMD GPU in the iMac, rather than CPU processing or the amount of RAM.

FCP2020016 M1 render chart

The first test was my Holiday Cabaret project, which I described in my previous post. The final show is 80 minutes and includes up to 44 layers of video. The mini would play smoothly up to about 36 layers in Final Cut Pro using Best Performance, which actually outperformed the iMac Pro that I originally used to edit this project. The iMac handled all 44 layers well.

FCP2020016 cabaret

The biggest difference in render times was with the commercial project in Resolve 17. Granted, the software is still in public beta and you may see differences when the final version is released. But Resolve loves GPU power, so clearly the iMac showed a greater proportional difference in this test than the others. I can guarantee you that if this test were comparing an M1 mini versus an Intel mini, that the M1 would have smoked the competition.

FCP2020016 speedOne test that was really interesting was a comparison between After Effects and Motion. While After Effects has never been praised as a fast nor efficient application, it is RAM-hungry, so I expected to see a large difference between the iMac and the Mac mini, due to the much larger amount of RAM in the iMac. 

The iMac was indeed faster, but only by a few minutes. However, the Motion export was exponentially faster on both Macs. This may be due to the Cycore plug-ins built into After Effects, which are rather long-in-the-tooth. Render times for After Effects were around 20 minutes, but Motion clocked in at a minute or less for a very similar composition!

If you set aside render times and only judge by real-time playback performance using optimized media, then clearly the M1 Mac mini easily holds its own against a more expensive and theoretically more powerful Mac.

Plus, it was whisper-quiet during all of my testing and never felt hot. There’s no reason that any FCP editor working in HD, 4K, or even higher resolutions wouldn’t be happy editing with the M1 Mac mini. Of course, editors should opt for the 16GB RAM option and a larger SSD, because you can’t upgrade those later.

Some considerations

If you purchase an M1-equipped Mac today, understand that many applications, utilities, plug-ins, and peripherals are still being optimized and tested by developers. Apps like Pixelmator Pro, Affinity Photo, and Resolve 17.1 Beta are ready now. Adobe will soon release native versions of Lightroom and Photoshop.

I tested Accusonus ERA4 audio plug-ins, which were fine. Video plug-ins are a different issue. Old Motion templates that I personally created continued to work, but Color Finale 2 and FilmConvert Nitrate were flagged as incompatible within Final Cut. For now, Color Finale 2 does not work at all. FilmConvert does appear to work, but playback performance is affected.

FCP2020016 m1 filmconvert

There’s been chatter about whether or not RAID arrays work with the M1 machines. Like other hardware peripherals, such as video and audio i/o hardware, you need to check with the manufacturer. The software drivers required to use that device may or may not be compatible with the M1 architecture or macOS Big Sur. OWC and Promise are working on and testing drivers and utilities like SoftRAID to be compatible with both Big Sur and the M1 Macs. I haven’t tested RAIDs thoroughly, but I did plug in a four-drive OWC RAID via USB 3 and it mounted and performed as expected. I haven’t tested a Thunderbolt RAID, though.

FCP2020016 back

One minor design point on the Mac mini itself. It currently features two USB-A ports that are rather tightly spaced. Two charging cables will fit side-by-side without issue. However, two USB thumb drives or license keys (dongles) with fatter casings won’t. A simple solution is to buy an inexpensive Anker 4-port, USB hub.

Conclusion

Who is the M1 Mac mini for and what can you do with it? In some cases, it’s obvious - such as school media classes. But also editors needing to replace their aging, pre-2013 Mac Pro towers. Or the editor working from home and needing something better to edit with than that knockabout laptop.

You get amazing performance at a superb price point. Apple’s design towards efficient processing is certainly playing out in real-world situations. Today’s M1 Mac mini is a definite contender for general computing, design, photography, and audio. If you can stay within the Final Cut Pro and Pro Apps ecosystem - then you’ll get stunning results, even in this base model. Forget that it’s only $699, because it’s ready today if you want to do pro-level work.

 

Written by
Top BloggerThought Leader

Oliver Peters is an experienced film and commercial editor/colorist. In addition, his tech writings appear in numerous industry magazines and websites.

He may be contacted through his website at oliverpeters.com

Log in to comment


markspencer's Avatar
markspencer replied the topic: #111844 01 Jan 2021 18:36
Great review, Oliver, thank you!
Oliver Peters's Avatar
Oliver Peters replied the topic: #111845 01 Jan 2021 20:11
Thanks. I've been enjoying your videos from Belize!
Whohauser's Avatar
Whohauser replied the topic: #111846 01 Jan 2021 21:03
As I have been using an M1 MacMini with 512gb drive to replace my 2013 8-core MacPro for the past 3 weeks, I can say that owners of that model of MacPro and lesser will find themselves generally experiencing a faster FCP workflow. Of course third party plug-ins don’t all work right now. The one place where the 2013 MacPro seems to still work faster is with CPU heavy programs like Handbrake or DCP-O-Matic. I have been using a beta of Handbrake that uses the unusual M1 CPU layout better (roughly twice as fast compared to the official version) but I haven’t made a direct comparison to the MacPro yet. My occasionally malfunctioning 2013 MacPro will now be used as a render machine for CPU/GPU heavy tasks such as DCPs (if I ever get a request for those again) and Topaz AI upres projects.
pszilard's Avatar
pszilard replied the topic: #111852 02 Jan 2021 07:43
One aspect of the new M1 devices, which is often overlooked, IMHO, is the 8GB ram versus 16GB ram. The primary reason why the performance difference is not huge, is that swapping to the SoC SSD is so fast, HOWEVER - and this is really important - SSD has a limited lifespan, unlike ram. In the case of the M1 boxes, it is obviously impossible to replace the SSD when it wears out, so you can toss your device in the bin!

So my advice is, if you are actually planning to use your M1 box for "real" work (or play), for a few years, then ALWAYS go for the 16GB ram model.
Oliver Peters's Avatar
Oliver Peters replied the topic: #111854 02 Jan 2021 15:32
I'm not sure where your lifespan info comes from regarding SSDs. The M1 Macs use the NVMe versions and not standard SSDs to start with. I have been running various Macs - all with SSDs - and at this point, no issue with lifespan. Witness the many 2013 Mac Pros still in service with the original internal drives. You can also assume 3-5 years with spinning disks, yet many last far longer.
Oliver Peters's Avatar
Oliver Peters replied the topic: #111855 02 Jan 2021 17:06
Just to add some info on SSDs. For grins I did a quick search of current SSD lifespan info. The general estimate is that SSDs have a rated lifespan of 10 years. Since this is based on read/write cycles, you may get more or less time. Google ran a study that showed 25% less replacement with SSDs than HDDs. However, they also reported higher error rates as SSDs aged. The bottom line is that you will likely get better life from an SSD or NVMe than an HDD under regular use. In any case, it's moot, since modern Macs don't come with user-replaceable main, internal drives. I certainly concur with the 16GB RAM recommendation. Hopefully even more in the next M-series Macs. And I also suggest running external drives for media and cache files. Leave the internal for apps and basic document files.
KB's Avatar
KB replied the topic: #111868 02 Jan 2021 19:25
I'm wondering why the 8GB RAM version - I understand the price point but I can't see trying to work with video with less that 16GB.

I used to think even 16 was insane but I met some folks that cut feature films on a MacBook Pro with 16GB and was convinced.

I'd like to see this done with 16GB.
Oliver Peters's Avatar
Oliver Peters replied the topic: #111870 02 Jan 2021 19:37
I understand what you are saying. The review mini was the base model and that's what I was supplied to work with. That being said, I didn't find a huge detriment to having 8GB. The general belief - and my testing seems to confirm this - is that SoC RAM roughly equates to double that of an x86 machine. So that 8GB is comparable to 16GB in an Intel Mac. If you ignore render times (which weren't bad, all things considered) then performance was excellent with optimized apps and codecs. For example, 8K ProResHQ was smooth as silk in FCP. With external storage, you could easily cut a feature film on this machine. The only app that I tested that is considered RAM-hungry was After Effects. It's render times were poor on both Macs - especially when compared with Motion. So having 96GB of RAM on the iMac didn't seem to matter all that much.
JoeEditor's Avatar
JoeEditor replied the topic: #111874 03 Jan 2021 06:41
I'm waiting for someone to load up a 2.5 hour long documentary with 8 GB RAM on an M1 and see what actually happens.

And RAM isn't so much speed, but how much data can be buffered into it, so I don't see how any 8 GB RAM on any machine can equal 16. You'd need to do a text of a project that eats up a ton of RAM, really. Then see what happens.
Oliver Peters's Avatar
Oliver Peters replied the topic: #111876 03 Jan 2021 14:46
It wasn't a 2.5 hr doc, but I did load up my Holiday project which had an 80 minute timeline - lots of layers plus text for subtitles. Mainly 1080p/29.97 ProRes. No issues. In terms of real-time layers, it did better than the iMac Pro that I'd edited it on. One big factor not to be overlooked is drive speed. My media was all on a fast TB SSD. But, yes, different project types will tax a machine in different ways.