Initial Impressions – Apple M1 Mac mini
The all-new Apple Mac mini was delivered yesterday and testing has been initiated. While there have been plenty of reviews, unboxing videos, and benchmarks of the synthetic or real-world performance characteristics already performed, our formal review is going to take some time based on errata experienced within the first twenty four hours of ownership. For standard productivity tasks, the responsiveness exceeds expectations. Granted, the all-in cost at the entry level of Apple’s revised ecosystem is significantly more expensive than an 8GB Raspberry Pi 4 with sufficient external storage. However, the rapid uptake of developer support for a number of applications at launch does bode well to supporting this wholesale change of computing architectures. Andrei Frumusanu’s review over at Anandtech provides an excellent deep dive into the technical aspects of the benefits of Apple Silicon.
Immediate benefits that stem from a power-optimized SoC include a near-silent compute experience and a chassis that is comfortable to touch. Apple’s design philosophy and mantra with the outgoing Intel-based Mac computers has historically allows the outgoing processors to run at ninety degrees or higher during normal operation. In the Space Grey Mac mini, flogging the system with CPU-intensive transcodes would make the Mac mini uncomfortable to touch for extended periods. The 2018 and newer Intel-based Mac minis also wouldn’t ramp the fan speed up to a highly audible level in these circumstances without the use of a third-party tool. Under full load, there’s minimal heat to be detected by touching the surface of the M1 Mac mini.
The quirks we’ve experienced in our time with this unit include the following events:
- Brief yet random dropouts of Bluetooth connectivity to an Apple Magic Keyboard and Magic Mouse 2.
- Reduction in wired network transfer speeds when the processor is heavily utilized. The beta release of the M1-native HandBrake triggered this scenario while transcoding one 1080p file to an Apple TV 1080p 30fps Surround format.
- Blizzard Agent, running through Rosetta 2, went unresponsive and required logging out and back in to cure system pauses/hiccups.
We’re expecting improvements to manifest in Big Sur 11.1 and beyond to address Rosetta 2 performance and capabilities. Thus far, most of the bugs that have been experienced have been minor in nature and don’t detract from how responsive the overall desktop experience has been on Apple’s first non-Intel outing in the traditional compute arena.
There are also some excellent examples of balanced and steady performance being achieved in our initial look and testing. Highlights include the following metrics:
- Native arm64 version of World of Warcraft (9.0.2) maintained steady 60fps+ performance at detail level 5 and 1080p resolution while running around cities and exploring the open world. For comparison, this level of performance on a Ryzen 5 3400G required dropping down 2-3 levels of detail while consuming more power.
- Multiple, back-to-back runs of CineBench R23 Multi-Core benchmarks remain over 7700 points with a minimum of variance between runs.
- Initial tests in Final Cut Pro X, using available media and projects, feel equally as performant as they do on a 2020 iMac with the Core i9-10910 with 64GB RAM and the 16GB Radeon RX5700 XT Pro GPU.