>>27
Intel and AMD spend billions on register renaming, out-of-order execution, flag renaming, micro-ops, and floating point stack optimization to work around the limitations of a 70's subscalar accumulator-based register-starved ISA in hardware, and compiler writers and assembly programmers spend time and money on instruction reordering, floating point stack optimization, and spilling registers to stack to work around the limitations of a 70's subscalar accumulator-based register-starved ISA in software. It's the worst of both worlds. It's not high level and orthogonal enough to make life easier for software programmers and it's not simple enough to make life easier for hardware engineers.