3 ms·
I don't think there's that many places in MacOS where they write hand optimised assembly for either architecture. Looking at rust it seems like LLVM is optimisi
by carlhjerpe 6y ago
I don't think there's that many places in MacOS where they write hand optimised assembly for either architecture. Looking at rust it seems like LLVM is optimising a lot of languages (specially that apple uses). This is platform independent.
The M1 is supposedly really fast, but it's not like they optimise userspace for either architecture.
Booting is different on the M1, but once the kernel is alive and well it's just RISC vs CISC.
- saagarjha 6y agoOf course a number of system libraries do get platform-specific optimizations.
- carlhjerpe 6y agoSure, but those are usually written in a compiled language that LLVM will pass over anyways, generating optimised code. My point is that this all "mac is optimised for the hardware" "nonsense" is very non-techy. Sure they don't need to worry about drivers the same way, but it's still x86 or arm, people already target these so it's just about whose code and architecture is the better.