Why x86 doesnt need to die

Article note: Generally a really good article to show to folks who imagine ISAs still matter much. I'm disappointed to not see the any direct discussion about the "You can't statically schedule dynamic behavior" issue in it, but it _does_ discuss all the ways in which modern pipes, regardless of the exposed ISA, work around it (they're re-flowing execution activity in a window of instructions, not executing instructions, in-order or otherwise, and they get to change the decomposition properties generation-to-generation without breaking compatibility with the existing software stack). Also interesting that it sets up the arguments for and against the x86-S legacy-free proposal but doesn't name it.