Oh absolutely I agree completely just wanted to say Usman had been one of the more reliable sources despite it being wccftech which report on every rumour under the sun.
I think what you said also extends to official info when it's vague like the "4x pure processing power over Xbox one X" which many people entertained to meaning only GPU ~= 24TFs. That one was quite an eye opener.
Oh yeah, but I'd still argue claims like that which are intentionally vague as to obfuscate the real factual details should be dismissed out of hand by everyone from the outset.
Frankly its pretty disappointing that MS would resort to such BS claims. If they intended to state the CPU increase, be specific and deal with that. Or talk about how devkits perform on a realworld XB1S game demo (i.e. like Sony did with Spiderman) if the TFLOPs numbers aren't flattering because a direct comparison with GCN hardware is misleading.
My issue with MS often is they treat gamers like idiots with details like this. Either be honest and specific enough to convey your message in a truthful way or hold info back until you've figured out how to do so.
Anyway, that's by the by.
Oh for sure - on a general issue of being cautious with info, I completely agree. Re. Navi 10/12 though, what was the difference there? The parts this year aren't Navi 12?
Usman claimed the new 5700 cards were Navi 12, when in fact they are confirmed by AMD to be Navi 10.
I can corroborate it using logic - if they literally use the GPUs AMD presented yesterday as AMD presented them, they will not have HW RT (whatever on earth AMDs version of HW RT is). If they use GPUs that are not those presented yesterday then HW RT (whatever on earth AMDs version of HW RT is, which we have no understanding of its capabilities at all beyond that it is "select in application" on their planned high end cards to replace Radeon VII) becomes of course more possible.
I would be completely shocked if both MS and Sony managed to come up with two competing and different hardware designs here regarding the GPU and an implementation of "HW RT". I am already dreading this "HW RT" wording at the moment as it is very intransparent - it could literally be anything.
So essentially your statement is a truism and not really saying anything?
I mean, of course MS or Sony won't take the RX5700 or RX5700xt chips "as is" because they are custom chips ans not discrete desktop GPU parts. So your entire premise is flawed which undermines the point of what you're trying to convey. In which, case i'd argue your statement is only logically consistent, not "100% valid" or accurate as you state.
As to the hardware RT, there really isn't anything to suggest MS or Sony can't jointly develop their own custom GFX IP blocks for accelerating some part of the RT algorithm.
Tbh, "RT" is used so broadly in the industry (covering things like Voxel Cone Tracing and SDF Raytracing) that Sony's/MS's approaches to HW acceleration could look very different to RTX.
I believe
chris 1515 has discussed previously about Sony's employee's historical photon mapping patents and work with RT using K-D Trees as opposed to BVH.
There are many ways to skin a cat, and Sony and MS only need to come up with the high level conceptual ideas of what they want to accelerate in HW while AMD works to implement that in silicon - that's why it's called a collaboration.
Yeah. It's most likely the 2020 Navis have something, and the console chips, being based on that same 2020 design, have whatever that is.
We should stop talking about 'hardware RT' as if it's some binary checkbox. It could come in at any level. I'll accept that it is going to produce some acceleration - and that's very good news vs what we might have expected previously - but I've no doubt the performance profile is going to be different than what was done in nVidia's solutions, for example. I don't think it needs to be at that same level to be meaningful, but we shouldn't take 'hardware RT' to be some singular, interchangeable concept.
This.