The initial excitement surrounding the Apple M1 chip announced just a few days ago is understandably about how well they will run apps for the Mac. Will they be faster at running FinalCut Pro, for example, compared to Intel Macs. Given what we know at this point, it is already almost certain that the Apple M1 will allow Mac users to enjoy the combination of much improved performance and significantly longer battery life.
One other aspect that is not yet widely discussed, is how this compares to Microsoft’s attempts to run Windows on ARM devices, such as the Surface Pro X. The SQ2 silicon that they designed together with Qualcomm for this device does not perform nearly as well in benchmarks as the Apple A14, let alone the Apple M1. Combine this with the lack of emulation for 64-bit Windows apps that is hopefully coming soon at last, Windows on ARM has been a failure.
Compared to this, Apple is claiming that the vast majority of Intel apps will run on Apple M1 hardware through the Rosetta 2 technology. Given their track record with the initial Rosetta software that allowed a seamless transition from PowerPC to Intel chips, I am inclined to fully believe this. Performance on Rosetta 2 should be good with some emulated apps running faster than when on native Intel hardware. Again, if my memory serves me right, I hardly noticed slowness during the PowerPC to Intel transition so even this claim does not sound outlandish.
There is an additional twist this time. In previous transitions, Apple was playing catch-up with the rest of the PC industry. Previous transitions happened when the CPU architectures that Apple was using were falling behind what was available on Wintel. This was true for the Motorola 68K to PowerPC transition, and it was true for the PowerPC to Intel one as well. This time, Apple is transitioning from the industry standard towards silicon will not be available for the rest of the market.
Given this situation, one thing that I am eagerly looking forward to is the performance when running virtualization software such as Parallels or VMware. Both have announced support for Apple silicon with Parallels demoing a version during the Apple annoucement, and VMware tweeting that they are not too far behind. Given that these are virtualisation software, they would potentially run Windows for ARM on Apple silicon Macs. Even though there will be a performance hit compared to running directly on the hardware, given the performance difference between Microsoft’s SQ2 chip and Apple’s M1, it is very likely that Windows on ARM will perform better on Apple’s M1 even on a virtualisation layer.
Let’s fast forward a few months to when Parallels on Apple M1 is ready. It could possibly be that Parallels on a MacBook Air running ARM Windows (with 64-bit application support) is actually faster than competing ultralight Wintel notebooks and also has significantly improved battery life. If this is the case, then anyone in the market for a good Wintel notebook might actually think that the MacBook Air with M1 might be the best choice — It will run Windows apps just as fast or even faster. The addition benefit is that there will be better integration with an iPhone or an iPad which they might own.
I would say that there is a reasonably high probability that this might turn out to be the case. It would be very interesting to see how this would introduce change the PC landscape. We might see significant efforts from chip designers/manufacturers like Qualcomm, Samsung, MediaTek, etc. to create ARM chips that rival the A1 in performance. We might see Microsoft making significant investments in their SQ chips. One thing is certain, and was even very evident when Android changed course and pivoted to copying the iPhone — once somebody does what was previously unthinkable, then it doesn’t take too much time for the rest to catch up. Now that everybody knows that the ARM architecture can rival and surpass Intel on PCs, other chip vendors will significanly increase their investments and focus on designing their own chips that do the same.
Naofumi Kagami 