iMacs Running Apple Silicon In Emulation Mode May Beat Your Intel MacBook Pro

Apple‘s June 22nd announcement that it was dumping Intel CPU’s and moving the Macintosh to internally developed ‘Apple Silicon’ raised a lot of red flags for some readers.

While there are many valid concerns about issues like compatibility, speed bumps in the transition process and the ultimate loss of legacy applications, many readers also fundamentally questioned whether the chip designs that Apple has developed for over a decade to run iOS and its variants are powerful enough to run a ‘real OS’.

The first prototypes of Macs running Apple Silicon are now out in the wild, and speed tests are starting to appear that suggest that performance worries may be overblown.

Geekbench metrics of the prototype Apple Silicon Mac in Apple’s Developer Transition Kit show that a Mac with a repurposed iPad chip, running Geekbench in emulation mode via Rosetta, has better Multi-Core Geekbench performance than a 2020 Macbook Air running on an Intel Core i3 processor.

‘Apple Silicon’ Geekbench:

2020 Macbook Air Geekbench:

Geekbench is a cross-platform tool that’s designed to measure your device’s CPU and GPU Compute performance.

The Macbook Air is an entry-level machine, so, as a point of comparison, we checked benchmarks on one of Synthtopia’s Mac laptops, a 15-inch Late 2016 MacBook Pro. Geekbench stats show the Developer Kit Mac Mini beating our MacBook Pro in single-core performance and offering comparable Multi-Core performance:

MacBook Pro (15-inch Late 2016) Geekbench:

Note that these are preliminary results. The Mac Mini computers included in Apple’s Developer Transition Kit are not intended to be production computers, and are arguably built with half-baked hardware.

For musicians, it remains to be seen how well Apple’s in-house chips scales up for higher-performance systems. High-end MacOS and Windows PCs blow away the Geekbench scores for typical laptops. For example, a current top-of-the-line MacBook Pro, with an Intel Core i9-9980HK running 8 cores, offers about twice the performance of the DTK Mac Mini.

But the fact that the first ‘Apple Silicon’ Mac performs as well as it does – running in emulation mode, on a chip repurposed from an iPad – bodes well for the ‘Apple Silicon’ switch.

23 thoughts on “iMacs Running Apple Silicon In Emulation Mode May Beat Your Intel MacBook Pro

  1. No worries about performance? There are several Mac Mini models from 2011 with higher Geekbench scores than that…

    The Macbook Air is an incredibly slow computer not made for cpu intensive workloads.

    I’m just saying that not even beating Apple’s budget model from nearly a decade ago is certainly worrying.

    1. The DTK is running an underclocked version of the A12X processor with the 4 efficiency cores disabled. The rumor mill suggests that the upcoming silicon will be a 12 core chip that should run faster.

      I wouldn’t be worried.

    2. It’s not even slightly worrying – remember, that was running x86 software through an emulator. Never mind that it was prototype hardware AND software. Running native software, the DTK mac mini probably already comes close to (or surpasses) the performance of the latest macbook pro if the scores of the macbook pro were double. By the time they are actually selling hardware, running silicon intended for general purpose computing instead of mobile, and with optimized software, it will surely be able to easily outperform and likely while delivering far better battery life, too.

      If building ARM-native code really is as simple as recompiling on the DTK mac, expect native benchmark results sometime tomorrow, and don’t be surprised if the numbers aren’t 50% or more higher.

      1. Sorry iMan – 2011 and 2012 were confused here – anyway –

        Intel Core i7-3720QM 4 core 2012 mac mini – there was a six year wait for apple to go to four cores in the mac mini again

        https://browser.geekbench.com/macs/325

        724
        Single-Core Score

        2794
        Multi-Core Score

        model 500 – ‘welcome to the future’

  2. lol
    these chips will never be on sale like that 😉
    its just a test kit for ARM code
    who cares how this benchmarks?
    everybody knows a chips are a powerhouse.
    used a new iPad lately?

    1. look at single core performance of latest a chips in iPads (no fan) vs. Single core performance of Mac book pro(with fan to get rid of the heat)
      and you get why its time to say good bye to intel and x86

  3. 1. This is cherry-picking. You can find a range of Geekbench scores here for comparison: https://browser.geekbench.com/processor-benchmarks

    These result show that the arm cpu can barely hit the baseline score of an i3-8100, which is a quite a cheap cpu.

    But I don’t think it’s fair to make comparisons with products that are not yet finished, so let’s wait until a that.

    2. This is also besides the point. The real problem is in thinking that an ARM CPU, or any kind of CPU will be significantly faster, or cooler or better in the next couple of years. They won’t, because we’re closing on the the 3nm barrier and the rate of development is getting exponentially slower and slower.

    This means that changing the main architecture of Apple computers won’t make them significantly faster, so if that was their goal, it was a totally unnecessary step.

    1. +1 – and likely that wasn’t their goal. Not being reliant on Intel and also streamlining development efforts are more likely to be their goals.

  4. i do not understand the point of these tests. they have not much to do with the machines to come. while recent mac books are not the fastest, ipads are getting faster and faster – so let’s wait and see. i think, a company the size of apple who successfully made the step forward to their own chips in the mobile world would definitely do this for the rest of their range too. no surprise.

  5. These benchmarks on a dev kit don’t tell much other than the final product will probably be different than those scores. Also the NDA claims public benchmarking isn’t allowed probably for that reason.

    While I like the idea of Apple creating its own chips for Macs and other products it does lock us into the Apple ecosystem even more and Apple is extremely slow in adding new features or even listening to its user base.

    I’m also worried that products will get more expensive as Apple is one of the greediest companies on the planet. (yes I’m an Apple user).

    Apple loves to sell overpriced products that can’t be upgraded or repaired. The bigger the company gets the worse it becomes as we see in what they sell today. I wish they got slapped for things like that as it’s only profitable for Apple but not for the end users in any sort of way. They’re taking away resources that belong to all the people on this planet. Heck Apple even tries to not pay taxes unless the rules are made in their liking. They simply have way too much power over us.

  6. It’s clear that Geekbench is running on x86 and not ARM based instruction sets, so these tests are meaningless.
    Performance will be greatly improved with these new chips, software just has to be rewritten to take full advantage of that.

  7. I’d like to clear up some confusion: This is virtualization, not emulation which makes a huge difference for what’s otherwise implied in the article. Here’s a rundown of the differences: http://techgenix.com/what-difference-between-emulation-vs-virtualization/

    This isn’t something new or cutting edge, but the de facto standard for cross-platform compilation (between different architectures) in *nix-environments. This works well, and when running geekbench command-line, that ought to work fairly well also. But that’s far removed from real world results from an actual rosetta-product with complex workloads in OSX, so much so that it has no bearing on what the end performance actually will be.

  8. perhaps this will bring their ridiculous prices down and will introduce a new method of measuring performance,remember giga-flops?

  9. Meanwhile, REAL MacMini with i5 performance numbers like 1000/4800. And another thing – what about VST x86 compatiblity? Is Mac finished in audio department?

  10. If you want real performance relationships, you need to compare the ARM chips not to current mac models, but instead to Intel/AMD based PCs running in systems that are not performance/heat throttled. Thinking a new Macbook is “a fast computer” is the crucial mistake in all of these comparisons. An “Intel i9 chip” will not give the same performance in a low clock cycle variant inside a heat limited laptop as it will in a desktop enclosure running at double the clock rate on a fast backbone motherboard with m.2 drives.

    1. People also need to realize that price/performance on apple computers are horrendous. And with a switch like this Apple will not lower prices even if it is cheaper for them to make. Which is why I really hope that windows(laptops in this case) start moving to the new ryzen mobile chips as they are mighty good and maybe can offer some really good computers compared to apple… Even though they are relatively new and being a new adopter comes with many risks… As far as just a good ol’ self built… it will always be hands down more powerful than a Mac at the same price range definitely while still competing with higher prices Macs. Though from what I’ve heard the only 3 good things about apple computers are — screens,battery, and OS. However those things don’t factor into price/performance so it’s hard measure the worth… At the end of the day… I’m more interested in the video game developer space and how apple plans to tackle us as an industry because video game developers have always been hesitant to code for a totally new OS that doesn’t offer any marketplace at all and with the switch to new processors… Studios have to be really incentivized to code for first off a new OS and then having to optimize on now a new 3rd cpu… Apple probably has the money to pay for a studio to do this but are others gonna follow suit if the game just flops because that hints the marketplace isn’t there…. Idk tbh but just food for thought

  11. > a Mac with a repurposed iPad chip, running Geekbench in emulation
    > mode via Rosetta, has better Multi-Core Geekbench performance
    > than a 2020 Macbook Air running on an Intel Core i3 processor.

    LMFAO, sure it has! How can that be news??? I write this on a Vaio i3 from 2011, and it´s shitty slow as hell!

Leave a Reply