Read this quick article on CPU branch prediction.  I feel like a decent understanding of CPU hardware design, but in reality I only know the low level details  till about 92.  Since older CPUs are simpler, they are easier to understand and were developed but much smaller teams.  Modern CPUs are designed by massive teams and each are specializing in certain areas.  It’s interesting to see how topics like branch prediction evolve to use more transistors to get better results.