How are you so quick to dismiss that Glofo is not to blame. AMD is basically pumping a lot more voltage than necessary as to qualify a lot more dies as working Rx 480. The fact that you can undervolt and get better performance due to reduced clock throttling shows that the process has significant variance which is forcing AMD to use more voltage to qualify the maximum dies possible.
The WHOLE semi industry is struggling to scale to lower process geometry. Naturally, those with less R&D resources are going to struggle more than others. Of course, I can't help but notice the "Robin Hood mentality" for AMD. Root for the little guy, to rise up against all odds.
Few years ago, Intel was touting about being an IDM, that their process team and architecture team can closely collaborate together because they are all under one roof. That was about the 90-65nm generation.
Nowadays, its a requirement. Design for process. Because that's how hard things are. Reaching the limit. So you can't blame on GloFo only, neither AMD only. GloFo is just providing the recipe, AMD is the one that has to gather the recipes to make something useful out of it.
We are always hoping for that "killer app", "breakthrough material"*. Maybe its time to realize that there's a hard limit. We didn't just pick the low hanging fruits anymore, we've almost exhausted all of the tree, exhausting all energy at getting the every little bit that's left.
*Here's something sobering to think about.
Golden age of Dennard(traditional) scaling
Silver age of device architecture(FinFET, GAA FET)
Bronze age of advanced materials
Golden, because the big gains came easy and at low cost
Silver, because we had medium gains with some effort and moderate cost
Bronze, because we have low gains with lot of effort and is very expensive