Comments by "Scott Franco" (@scottfranco1962) on "why can’t computers have thousands of cores?" video.

From the quality of code I see generally, I believe a lot of performance improvements come from giving each program its own core. This is because the programs are looping or polling continuously waiting for conditions to end, such as a read call, a network transfer, etc. Programmers have been taught to be lazy and use so called "non-blocking" calls to get things done. The result of this is that giving such a program a dedicated core to just sit in a loop will increase overall performance because it is not getting in the way of other programs executing. I call this "burning a core" (like dropping a useless card in a card game). This mode of programming automatically limits the usefulness of adding cores. Once you have enough cores to waste on programs that parallel task inefficiently, then further cores will not help your speed.
2