You're awfully stuck on that ".9" when the original poster mentioned it in passing and as means to say it doesn't have all the stuff in RDNA2 because it's been traded by custom optimizations.
Now I also think that conversation has run its course, we will just have to agree to disagree that the expression he used was symbolic and not descriptive.
Reality is o'dium is as close a verified insider as we're likely to get, and discrediting his rumours needs better arguments than character assassination.
So I mean I feel I have to respond to this since this has mainly been my direction here I apologize for the long post. I'm not trying to character assassinate Odium, on the contrary as I have said, he can certainly have information but still be wrong. That's not a assassination of his character or whether he is outright trying to lie to people, but people can get things wrong. That being said, I need to caveat where I'm going with this: I have no side in this argument, at the end of the day things will play out regardless of how we discuss speculative specs. There's no going back at this point in time; target discussions are dated, we well into hard end game for this consoles now.
Shifty, you opened up the goal for this discussion to move away from Github and onto any rumours to let in anything to be heard or discussed. I respect that, and in some ways it does open up discussion, but there are downsides to that; broadening the spectrum allows for more noise to come in.
Since with this thread we are dealing mainly in rumours and baseless posts; I need to discuss some things head on about leakers, information we hold tightly, and sources. One of the things that I do today is when I receive information I don't tell people what it is. This is because if I have information no one else has, and that information is leaked by another party, I can qualify the likelihood that the rest of their information is accurate. It's a fairly simple but effective way of determining if something is true or not. And I don't need to say anything about it. That being said, I can evaluate a great deal of journalists or leakers who have a proven track record in this regard, this doesn't mean they are perfect, but they are definitely on signal than they are on noise.
As you both know, there are a great number of individuals on this board that have access to information that would sit inside NDA territory, but these individuals have no credibility, do not want to expose their sources, or do not want to be investigated as leaker. What's interesting is that when the Github data mined information were released, it verified their information in a way that would not expose their sources or declare themselves a leaker. They only needed to continually point at Github data, which to this day has proven extremely accurate.
When you took away Github from the discussion, you took away the chance for these individuals to speak on their intel. Granted the data is now dated and the information that needs to be official is now official. But you can see how what was declared as 'Github Gospel' is really more like, Github is my chance to safely point people in this direction. And that was removed; fairly though, I believe its had it's time in the spotlight and the information is no longer valuable to discuss, but keep aware that there were people genuinely trying to tell you guys something but was brushed off by being called a religious cult.
The interesting thing about leaks is that proven journalists and leakers all start to converge on similar data points which has backings to other data points and so forth. These type of relationships are what provide additional credibility to any data point and moves what would be an anomalous data point into signal.
In my line of work, anomaly detection is a very prominent task that many scientists will take on. We look at sample or population of data, find the lines of best fit and determine what anomalous data points exist outside of that. We build ML algorithms to detect anomalous data points in real time. It's nothing new, credit card companies and everyone has been doing this type of thing forever. The interesting thing with doing anomaly detection is that over time data that is anomalous and then becomes consistent the algorithm will determine that we have shifted the line of best fit. For example; this might be when everyone started saying XSX was 12 TF and that was a far stretch until everyone kept saying it from a variety of sources and it became that.
Having said that; this particular point in which Tott points out "
You're awfully stuck on that ".9"" is just that to me. I cannot ignore how anomalous this data point is. I need to be clear, all of our leakers, journalists, sources have never once declared RDNA 1 or RDNA 2 for XSX or PS5. Not once had it ever come up in discussion in terms of leaks (all declared it as Navi) but aside from expectation there was never a single leak on it. Nothing, not a single peep. Post announcement of XSX official specs, we actually get a ton of tech blogs, journalists, papers of the sort indicating that they were surprised by RDNA 2.0 coming to XSX. And not a single one of them stated in their journals that they have information that suggested it would be, or that they had information suggesting it was RDNA 1.0.
SO when Odium came along _post_ announcement and declared it not 2, but 1.9 a figure of speech I understand this completely. I find this data point to be extremely anomalous for 3 reasons
a) no one else or any other leaker had information on the architecture (he's the only one)
b) no one else has called it less than RDNA 2.0 (he's the only one)
c) he's gone so far to declare how close it is to RDNA 2.0 but quite not (he's the only one)
In my line of work, this is a massive anomaly. There are a lot of people talking specs and leaks and rumours and paste bins and journalist and somehow among all the data that we've seen and come across, this 1 particular point has never come up.
When we look at VFX Veteran's supposed GAF leaks, he brings in a ton of anomaly points as well. A second Playstation, Lockhart, price points in the 800. Once again none of these have ever been discussed.
Normally we ignore anomaly's. Because they're likely not proper signal. And while I respect the idea of opening things up, i have been trying my best to show case what is anomalous and what is not and I'm being countered with vague replies on how things could be. This isn't an attack on someone's character; this is just how I see data points that I know deserve hard scrutiny.
Everything else Odium has said checks out, but everything he has said was also said and known by other leakers MONTHS ago. Jaguar cores and PS5 was discussed by Richard a month ago. 8TF and 9TF PS5 was known in January of 2019. He has provided few data points that I can see are his alone and the one that stuck out is that one. Because post official spec announcement, he posted that, and he's still the only one: and that's deserving of scrutiny.
Maybe the source of the leak was internal...
