@ChuckMcManis @f4grx when excessive pipelining failed to soak up the transistor budget, pentium M added 4 megabytes of cpu cache, they added more execution units and went hyper-threaded then full SMP, they integrated graphics on die with framebuffer and 3D engine texture memory, all the branch prediction and speculative execution that gave us meltdown...

All this was _after_ photolithographic die size shrinks stopped giving them free doublings.