I am not assuming anything. I don't see Bonaire and naturally go thats a Durango derivative or Durango is a Bonaire derivative. But I don't naturally discount it either.
Given that I asked him whether he thinks Bonaire would make a good placeholder in the alpha/beta kits, timing is irrelevant (Im guessing sometime last year). The point isn't to convince him he is wrong, the point is to find out why he so against Bonaire being anywhere related Durango, when there are alot of coincidences in place while AMD is employing an unusual strategy for product release.
Coincidence is when two things that have nothing to do with each other happen to match up.
Bonaire and the rumored design for Durango come from the same technology pool from the same design company with apparently very similar performance targets.
Coming from the same pool of options and some of the same criteria means it's not coincidence.
However, coming from the same pool doesn't mean one begat the other. For one thing, that's sort of hard to do because they probably had periods of concurrent development, where neither design was set in stone and didn't actually exist at the time claimed.
A chip isn't a good predecessor or placeholder if it doesn't exist.
Their similarites almost certainly can't be coincidental, however, that doesn't mean either one controlled the development of the other. It's not necessary, and potentially problematic if a confidential contracted product leaks into another.
Let's note that the Sea Islands ISA document, which Bonaire is alleged to belong to, has features consistent with Orbis. I think Microsoft would be very interested in knowing if Durango wound up in Orbis. On the other hand, it's not clear that Bonaire has existed as long as Durango.
edit:
By way of analogy, the two designs are like siblings or cousins from the same set of ancestors. Assuming there isn't some kinkiness going on, a sibling shouldn't have undue influence on the genesis of a new member of the family, but there's still going to be a family resemblance.
Fixating on a few numbers such as FLOP counts or memory bus totals is not strong evidence because the common ancestry only has so many sane ways to combine.