Sigh . some click bait links are not really "your best friend". Googling with no reading comprehension or critical thinking skills is not "your friend".
No clue what that expreview.com thing does for me it takes way to long to load. ( indicative of tons of disco hocus pocus ).
The WCCF post is weak as water. First claims that Navi will be the 6th GCN iteration when it isn't. That Vega 7nm is the 6th. ( not the 5th. There may be some AMD PR or leak that is claiming it is '5" but if add instructions, then that is a next iteration. )
So if 6th was the 'last' and Navi comes after the 6th then ...... ta da ... new foundation iteration.
Was "Vega 20" 100% crystal clear at that point. Perhaps not 100% but there was enough to be somewhat reckless to have promote notion that "Vega 7nm" was purely just a shrink and not another GCN iteration.
Second major flaw
"... We were told that work had already started on the brand new major-architecture before even left, ..."
Right according to the 2016, 2017 , and 2018 roadmaps is Navi. Go and look at the patent filed on dates for the three patents linked in earlier in the thread. (the date on which is published is only relevant in being able to see the completed application. The date filed is more important if this is a patent explicitly being developed for a real product. ) late 2016 , late 2016 , mid 2017. The content of patents is typically worked on many months before they are filed. That puts that research work back in 2015-2016. So was a new arch being worked on before Koduri left in late 2017? Yes. Do these patents' context date about a year earlier than his departure? Yes. Was Navi being worked on a couple of years before his departure? Yes, at least the planning ground work stage for years mark and certainly well in flight by the leaving time. ( it wouldn't be an explicitly named item on a 2016 roadmap if that were not the case).
Getting a significantly different micro-architecture out the door can 3-5 year effort. AMD started in 2014-2015 then 2019 is about on target. We have substantive patents being put together in 2015 and a new named arch coming in 2019. It isn't 100% conclusive but it fits but the timing and the little details that AMD attached to the Navi release on these timelines.
"...AMD has already hinted in its slide deck about the existence of this major architecture.
Remember the “Next-Gen” micro-architecture that was listed in their roadmap? Well, its the one and the same. ..."
That's mostly hand waving. I covered that in previous posts. "Next Gen" is far more likely just a placeholder for a name they'll come up with later. It isn't necessarily descriptive of implementation differences from GCN. It comes after Navi is probalby mostly what 'Next' means there.
In fact this comes a relatively shorter after ...
"... and we don’t even have some tantalizing codenames to offer but one thing is for sure, it ".
Right! There is no codename for the next generation so AMD is using the label "Next Gen" . If you supposedly don't have a code name and all AMD wrote down was something else then that something else
is pragmatically the code name. Just because it isn't a 'Tantalizing' code name doesn't mean that it is not a code name. "Next Gen" isn't sexy/tantalizing enough so it can't be a code name is 100% rubbish. Just garbage.
Thrid major flaw (hand waving)
"... t is known that this brand new architecture will result in a leap that is at-least as great as the TeraScale to GCN shift. Since the process will be the 7nm+ optimized node ... "
if the new arch is so much better of an implementation why does it require a new process shrink. If have grips on a significantly better arch implementation then should be able to put it onto the a 7nm fab process and still get a major boost. If it is required that get to a shrink to get a boost then the shrink is just as likely playing a major role there; not the arch.
