Perusing Geekbench makes for interesting reading.

The one thing that astonishes me is the massive leap in performance between my old MacBook Pro 15 inch (2017) and my mew laptop – the single core and multicore performance doubles in a space of 5 years and within that space that SoC is so efficient it no longer requires a fan. I don’t know about you but the more by Apple to their own in-house SoC is the best move they’ve ever made. When I first heard about the change I was expecting maybe some modest improvements but holy heck the improvement is leaps and bounds ahead of the competition. The performance of my Mac Studio, the single core is twice as fast and the multicore performance over triple the speed of my iMac 27 inch (2017). I’ve loaded dBpoweramp onto my Mac Studio to try out ripping and encoding music – very fast, no complaints by me. I’m tempted to see how things work with Handbreak when I have the chance but long story short the fact that it was all done without firing up the fans really goes to show the love of investment Apple has made.

To be frank, I was one of those who were sceptical about whether Apple could scale up an ARM based SoC but I’ve been happily proven wrong. The other benefit of having spent years optimising for ARM has been a rock solid macOS that is optimised for ARM from day one. The reason why I was sceptical was because scaling up GPUs, for example, can be incredibly difficult – scaling isn’t always linear so it isn’t always possible to take a lower power design then scale it up simply by scaling up the number of cores. The other scepticism was regarding Apple’s ARM CPUs from the point of view that the power efficiency may become a limiting factor but once again I’ve been happily proven wrong.

There are rumours that Apple has a big architectural change in the next couple of years so maybe that’ll make the move to ARMv9 which will include SVE2 plus other enhancements but that being said , as Rene Ritchie noted, Apple has already made a lot of those improvements available in their own implementation. The big thing will be where SVE2 fits into it but that being said I don’t think it’ll be the massive improvement given that Neon SIMD extension is good enough for the vast majority of programmers who write software.