M1 MacBook Air even beats the 10th-generation Intel Core i9 in recent single-core benchmarks, and scores double the Intel MacBook Air on multi-core.
The MacBook Air based on Apple‘s new M1 chip recently appeared in benchmarks revealing jaw-dropping performance for such a small and low-cost laptop. An explanation of what the scores mean can help to understand Apple’s controversial performance claims. While these results might not directly correlate to real-world usage, the relative scores compared with other devices are interesting.
Apple’s M1 chip is based on an ARM architecture, which began as a smartphone processor, though it has recently been used by more than one manufacturer for laptops. Apple, however, is the only one to claim that its heavily-customized mobile processor actually has the world’s fastest processor core for low-power silicon devices. In fact, Apple states that its chip has high enough performance to run many Intel apps, via its Rosetta emulator, faster than they do on the Intel MacBook Air that launched earlier this year. To win while operating with less than optimal speed implies a big margin.
As Apple suggested, the latest Geekbench scores for Apple’s M1 based MacBook Air show multi-core CPU scores that are more than double that of the quad-core Intel Core i7–1060NG7 in the MacBook Air that launched earlier in 2020. The single-core CPU score of 1678 shows an improvement of 48-percent over the Core-i7’s 1136. While not quite the 3.5-times performance improvement Apple claimed, it’s a good enough single-core score to outpace every Intel Mac ever made. As shown on the main Geekbench Mac Chart, the 1251 score of the Intel Core i9–10910 found in the 2020 iMac, suggests a speed just 75-percent as fast as the M1 MacBook Air. That’s single-core only, however, which doesn’t tell the whole story. In multi-core, the 2020 iMac scored 26-percent higher than the M1 MacBook Air. Of course, the iMac is a desktop computer with unlimited energy draw from a wall plug and has a much larger enclosure with fans for cooling. An Intel Core i9 iMac would be expected to win in that comparison and by more than 26-percent.
The Bottom Line & Rosetta Performance
For those looking for an Intel win, there is solace in the crushing multi-core performance of a 28-core Intel Xeon Mac Pro, scoring 18960 (found on the same Geekbench Mac Chart mentioned above), which beats the M1 by 2.6-times. What is expected to be the slowest M1 Mac, the MacBook Air has very impressive performance, as revealed in the Geekbench tests, but real-world experience depends on each individual’s usage. Single-core tests represent performance with day-to-day computing tasks. Multi-core benchmarks show what to expect with more demanding jobs, such as 3D modeling and rendering, video processing and graphically intensive gaming. The benchmarks suggest that the M1 MacBook Air can outperform the CPU used in Apple’s Intel laptops, but a computer is more than just a CPU. The GPU scores matter as well, and they are typically used to enhance performance in the same type of tasks as multi-core CPUs. Apple indicated a six times improvement over integrated graphics and its claims, so far, don’t seem to be overly exaggerated. Perhaps the GPU will prove superior as well?
Another point to consider is the performance when running under Apple’s Rosetta emulator. Apple said performance was good, but didn’t quantify that. Geekbench provides a bit of insight on this point as well. After the Worldwide Developers Conference, some Geekbench scores were posted for a CPU called ‘VirtualApple.’ This didn’t get any notice, since the quad-core CPU scores seemed roughly on par with the Intel MacBooks at that time. However, new ‘VirtualApple’ CPU scores have appeared, this time following the release of the M1 MacBook. The scores jump dramatically, as well as the core count, showing an eight-core CPU. The implication is the first batch of ‘VirtualApple’ scores were related to the quad-core A12Z Mac mini Developer’s Transition Kit running the Rosetta emulator, also known as a virtual machine. The more recent scores may be connected to the new eight-core M1 Mac computers with Rosetta. Based on this, it appears that the M1 chip running Intel code via an emulator is faster than the Intel chip in both single-core and multi-core tests. The performance hit of Rosetta seems to be about 20-percent, if this speculation proves true.
Apple’s claims of achieving best single-core performance for low-power silicon and good performance, even when running legacy apps compiled for an Intel CPU, are seeming less far-fetched. The M1 also features a 16-core Neural Engine, which assists with many of the tasks performed by the GPU when using Intel processors. Perhaps, the claim of several times better GPU performance when compared to integrated graphics may be true as well. The MacBook Air never had an option for a discrete graphics card, so that comparison would be moot. Given the mounting evidence, the M1 MacBook Air may prove to be several times faster than the Intel MacBook Air in benchmark tests, just as Apple claimed.
Next: M1 CPU Performance: Just How Powerful Is Mac’s New Apple Silicon Chip?
PS5 Sells Six Times As Much As Xbox Series X/S In Japan’s Launch Week
About The Author