It's nothing more than your belief and interpretation. It's your right to ignore holes in this leak and obvious stretch in interpretation to get to conclusion that you like.
But you should agree that it's only you subjective opinion, other people have right to other interpretations of the same facts (their subjective opinions), and all of this just forum rumbling that has nothing with scientific approach.
There are some simple question that have no answers, and having no answers means that conclusion are just "loosy hypothesis" nased on subjective bias and nothing else.
1. It's not proven that the leak is real and actual - not tailored or doctored. We get some vague confirmation that "data is real", but it can mean anything. For example - I get data from 2018 year, change all dates 1 year forward, and post it as leak. Basically all data in it is "real", but all further conclusions have a high chance of being faulty (doctored results). Or Sony decides to create a misleading leak to fool everyone and lower expectation. They ask AMD to do some tests of old stuff, put some garbage into folders and pretend that it is spontaneus leak by some intern. Post it on the internet and watch the forums burn (controlled tailored leak).
There is a reason why all scientific theories cannot be based on just one source and must be checked against independent source.
2. Even if this data is completelly true and reflect current state of chips, there are a lot of other interpretations - BC testing, dual GPU, old revision random testing etc. And they are as good as pushing agenda of "36 CU is a final state" - because it's impossible to hard prove that they are faulty conclusions (not enough data) and most everyone can do is to rely on "I think my explanation is more reliable". And this is not science - science theory is objective, replicable and doesn't allow different interpretations in the same borders of framework.
And personally for me this information have less value than insiders info. Because insiders have educated opinion based on information with presumably with completeness. And everyone who tries to push this leak do a guess with very weak foundation of incomplete information with absence of established framework what this information is about, that is usually a basis to have heavily influenced by bias conclusion.
If you're this willing to put the benchmark leaks under such scrutiny, why aren't you willing to do the same with insider rumors? If you're selectively dismissing certain information in favor of other information, then you should just admit it is being done out of a bias.
I haven't ignored the more questionable parts of the Github leak and benchmark tests; well aware that 100% context isn't present. That's why I say they are probable: basing that off of how stuff just like this played out last gen, and the gen before that. Precedent, history repeating itself, patterns etc. And I haven't 100% dismissed insiders; quite the opposite, actually. With the credible ones, when they post stuff I question, I either ask them for elaboration directly, or speculate fairly how they could be arriving to/hearing those rumors.
From looking at your post, you already assume Sony intentionally had those benchmarks leak out...why would they also mix in Xbox-related tests? Why would they ask AMD to do that? Why would Microsoft agree to that? That's the kind of stuff that could result in court cases, these companies are not petty or nearly at each other's throats enough to jeopardize their business relationships that way.
We have had other sources besides that one Github leak: further GPU benchmarks with Ariel and Oberon (and other chips like Arden, Sparkman etc.) consistently showing similar testing data. Those are AMD's own repositories of benchmark testing data...seems like a pretty good source to me. Unless you're implying people like Komachi have an agenda to smear a company like Sony...all the same, you could speculate some insiders have an agenda to smear Microsoft. Anything's possible in that sense, but I'm going to venture that Komachi, Rogane etc. are not intending to do that, and I don't think most of the insiders are trying to do something like that, either (except perhaps Tommy).
I've speculated on dual-GPU possibility, but I've also questioned why something like, say, dual Oberons, would "only" produce a high-end of 12-13TF knowing that would be both chips clocked well below the sweetspot. So what's left? Dual-GPU Oberon + Ariel (assuming Ariel isn't an earlier version of Oberon)? Same issue. Oberon + Gonzalo? Possibly. But that's as far as dual-GPU speculation can go with what has been uncovered right now.
What's more possible is that there's probably a cluster of CUs on Oberon that were deactivated. That's why I quoted that APU graphic in one of my other posts: it's more likely than a dual-GPU approach given what we know right now. The only thing that graphic doesn't seem to entertain is possibility that a similar disabling of CUs might've been done with the Arden chip. In any case, it's a likely possibility.
Again I'll just say: back in 2012/2013 when benchmarks for PS4 and XBO were coming out, and (a bit later) some of the system features leaks were coming out, did we see similar dismissal of benchmarks and leaks from people (particularly Xbox fans who couldn't believe what they were seeing)? You bet we did. Yet we see how those shook out in the end. I've never once claimed the Github and benchmark datamined leaks are 100% what the next-gen systems will have, but they're somewhat pretty indicative of what is in store for them, absolutely. And we've already considered wiggle room scenarios in terms of those benchmarks that can give credence to some of the more reasonable insider claims, too.
It's not an either/or binary thing, there's room for both to be accurate on some things. But for reasons I've already mentioned plenty of times, I would weigh the benchmarks more on the scale than insider rumors. By how much, I'm still in flux with. But it is what it is.
For me the best concept is still this
First time I've seen this one; looks pretty damn good. Wouldn't mind this being the retail design whatsoever
