Editorial Note: A previous article about leaked Apple M1 performance proved false. Rather than continue to proliferate misleading performance numbers, we've redirected that story to the results below, which have been submitted to the publicly accessible Geekbench database:
Apple’s new M1 Silicon chips start shipping this week, which means the public is finally starting to put them to the test. So far, we’ve already seen the chip’s performance numbers in benchmarks for native apps and in graphics tests. But now we’re finally starting to see statistics for how well emulated apps run on the M1.
Emulation has been a bit of an albatross around Apple Silicon’s neck since it was first announced. Even if the new chips are faster in theory, would they be worth it if your favorite x86 programs need to be emulated to run on them? What if the emulation makes them slower in practice? However, according to new benchmark results uploaded to the Geekbench website, this doesn’t appear to be a concern.
Over the weekend, a Geekbench user uploaded numbers for an M1-equipped MacBook Air running an emulated version of the x86 Geekbench test through Apple’s Rosetta 2 translation layer. The laptop’s single-core score was 1,313, while its multi-core score was 5,888. That’s about 79% as powerful as the same laptop running the native version of Geekbench, which hit scores of 1,687 on single-core tests and 7,433 on multi-core tests. Still, that gives it higher single-core scores than any current Intel Mac, including the 2020 27-inch iMac with a Core i9-10910 processor. And as for multi-core, it's still leagues ahead of what a Core-i7 2020 MacBook Air can do.
That’s a relief for anyone who was worried that having to process the Rosetta 2 translation layer would make their new M1 MacBook less powerful on some popular programs than older, Intel-based models. And emulated performance is key, since even certain mainstays like Microsoft Office won't have native M1 support at launch.