nostr:nprofile1qy2hwumn8ghj7un9d3shjtnyd968gmewwp6kyqpqer26g8v2gdquxkyk43usmlk7gaqhle08p378ld3ldfd6p5yrt8us3qs2sp x86 is on the other of this side spectrum. There is no reason to pretend to support DO...
@nprofile1q... Intel no longer has the talent to implement a sane ISA, not like x86 is a very sane ISA either. They couldn't do it with Itanium, although that thing was doomed from the start with its "make compilers almost sentient" mindset. They couldn't do x86_64 properly even; they wanted it to be incompatible with amd64 after that made it to market.
As dumb as still technically supporting DOS is, it gives you stability for the whole ISA, which is why x86 ultimately became so prevalent to the point that it is still unbeaten except in some niche cases like supercomputing, mobile or embedded. You don't need do patch uboot for every specific ARM SoC you have, you don't need specific SDKs for various ARM SoCs which include modified versions of Linux kernel that will never be upstreamed. You write something once and the thing just works everywhere, because the low-level plane is all the same. Writing a generic x86 kernel is relatively easy even compared to a generic aarch64 one.
As dumb as still technically supporting DOS is, it gives you stability for the whole ISA, which is why x86 ultimately became so prevalent to the point that it is still unbeaten except in some niche cases like supercomputing, mobile or embedded. You don't need do patch uboot for every specific ARM SoC you have, you don't need specific SDKs for various ARM SoCs which include modified versions of Linux kernel that will never be upstreamed. You write something once and the thing just works everywhere, because the low-level plane is all the same. Writing a generic x86 kernel is relatively easy even compared to a generic aarch64 one.
1