Thoughts
I cannot optimize my computer programs for efficiency because if you're trying to be efficient there is only one correct solution and it is
a custom integrated circuit. Every thing else is inefficient. As soon as you introduce a CPU it's inefficient. If you can't afford a custom IC you can try to get away with an FPGA but it's not as good. It's just fricking not.
This is why systems programmers drive me insane. If you're worrying about efficiency at the level of individual assembly instructions you just use an FPGA. Leave me the hell alone and let me write my bad slow Zig code.