I get that, and I don't disagree with your assessment. I'm just throwing in my two cents towards the holistic perspective that AMD's perf/watt has been poor for so long so that even these improvements which might have been good a few years ago are now barely meeting the minimum of what was promised. What I want to know is if it's just the continuing symptom of GCN's perf/watt issues or a product of the difficulty AMD is facing with designs on the 14nm node, and more specifically, porting the inefficient GCN design to 14nm. Curious cat.
I really think there's just something wrong with 14nm LPP. I mean if Fury (non X) has 92% the efficiency of P10 on 28nm, then surely this must be something to do with 14nm LPP. Even discounting for HBM Fury should still end up with at least 80% of P10's efficiency, and that's just ridiculous for a node shrink + transition to FinFET.