My original article you are talking about explicitly stated heterogenous compute is the future. It talks about transistor budgets going to LLC, GPU, media, and

My original article you are talking about explicitly stated heterogenous compute is the future. It talks about transistor budgets going to LLC, GPU, media, and ISP. Much of the CPU efficiency gain comes from that doubled LLC size.