>Right, except in the one case where they did choose ARM?
You mean the M1?
I'm fairly sure this was designed even before the RISC-V base spec, the unprivileged one (no supervisor vs user, or MMU), was ratified. It was ratified in 2019, by the way.
The batch of extensions ratified by the end of 2021 contains important featuresets like vector or hypervisor support.
So, of course they didn't pick RISC-V back then. Yet, nobody in their right mind would pick ARM today.
Leaving aside the merits of the two ISAs, for Apple using ARM for the M1 had the huge advantage unifying all their devices to a single ISA.
Based on just that (and the fact that they own a perpetual license to all things ARM) they would likely make the same choice now and even in a couple of years.
>for Apple using ARM for the M1 had the huge advantage unifying all their devices to a single ISA.
Wait, let me read that again. Oh, so you're saying that Apple wants to unify their ISA... by switching their already ARM smaller/hidden cores to RISC-V?
No I am saying that even if RISC-V was strictly better than ARM Apple would have likely still chosen ARM for the M1 generation as it would be the simplest way to unify their ISAs across their devices.
You mean the M1?
I'm fairly sure this was designed even before the RISC-V base spec, the unprivileged one (no supervisor vs user, or MMU), was ratified. It was ratified in 2019, by the way.
The batch of extensions ratified by the end of 2021 contains important featuresets like vector or hypervisor support.
So, of course they didn't pick RISC-V back then. Yet, nobody in their right mind would pick ARM today.