I don't know but he said the patent looks like something linked to VRS and from my point of view it looks like this too.
Current Generation Hardware Speculation with a Technical Spin [post GDC 2020] [XBSX, PS5]
- Thread starter Proelite
- Start date
- Status
I don't know but he said the patent looks like something linked to VRS and from my point of view it looks like this too.
I second your skepticism. His Twitter bio lists #GameDesigner #GameDev #IndieGameDev but has no link to companies or games that he's worked on. That and typically real game developers have a much larger following on Twitter. Many grains of salt here, but you never know.
I don't care of what he is saying but the patent itself looks like a form of VRS. This is the first time I see this patent.
I'm aware that this is what Anandtech has been told, but it's not what's been happening in practice. These are the roadmaps shown during 2019:
Both Zen 3 and RDNA2 are releasing their first models before 2021.
As such, it's reasonable to expect RDNA3 on 5nm to release before 2022, even if AMD don't want to compromise on that.
Wouldn't surprise me they said it's "inclusive" just in case there's delays for whatever reason. So far they've always delivered as if the latter year marks the start of said year, not end.
Last edited:
Don't care who this twitter user is if what they referenced is legit.
If PS5 had this, sounds cool although we would need to see the actual impact.
And how much above and beyond in terms of performance and/or IQ is it beyond doing it manually in two steps of Mesh Shader then VRS.
Well, if it's legit then great!
cheapchips
Veteran
RDNA has improved culling in the front-end. Per the whitepaper:
PS5 would have an advantage with the higher clocks, but I think it would mainly amount to just reducing the bottleneck in that stage of the rendering pipeline in multiplatform titles.
That said, there can certainly be compute culling methods implemented as well that maybe has better knowledge of what to cull in the game, such as with idtech7 or Frostbite.
What about Mesh Shaders + VRS? Which is is the better approach?
Also causes the boat to grow apparently....
I don't see how you could replicate VRS using the geometry engine, as VRS tier 2 requires shader time decision making on how you process coverage of geometry that has already gone past the geometry stage in the pipleline.
I also have no confidence whatsoever in RGT's claim that Sony's Geometry Engine is replacing the current line of technology that's resulted in RDNA2. That would make no sense, as RDNA3 won't be in consoles and needs to support DX12U and Mesh Shading, and be an evolution of what they're bring to the PC in RDNA2. Furthermore, for RGT to have had this info months ago would require them to have a source inside the most sensitive part of AMD chip development labs, at a time when they (RGT) are still fumbling around talking in vague terms about hardware that's been in the hands of actual developers for months.
And this is definitely not what Cerny was talking about in Road to PS5, as Cerny was crystal clear and very specific that we should look for a product that appears at the same time to bear the fruits of their collaboration. That would not be an entire GPU generation - and probably ~18 months - after the arrival of PS5.
What he's saying is an incoherent mess.
- VRS is not about altering effective resolution by screen location. You could force it to do just that, but it's fundamentally far more controllable than that.
- The Geometry Engine is almost certainly not there to "procedurally generate high quality textures" (wat??)
- VRS isn't supposed to be a substitute for removing unnecessary geometry, that's one of the things programmable "mesh shaders" are there for (which in DX12U supersede Vertex shaders and Geometry shaders).
The patents are legit. The interpretations of what they mean leave much to be desired.
To me it's probably the case that SFS isn't there because look at the raw specs of the PS5 storage system. Not sure if there is total overlap between the two in terms of what it's doing/performance impact.... but maybe Sony diverged there to get the chip quicker to develop the cooling system/get dev kits out quicker.
Not being a dick, but there's two different things.
SFS which is an xbox exclusive feature, as in its not in RDNA2.
SF is what you mean, which is in RDNA2.
Screen location can be anything 2x2 or other size of pixel and can probably be done by primitive too. Here it means shade pixel at different rate by altering rasterization parameter.
I don't care of his explanation but about the patent.
Seems ideal for front end changes, it's definitely possible any work Sony did with the front end could show up in RDNA 3.
But the impact is somewhat muted by the fact that we continue to move further away from the front end and into compute.
It's just very difficult to keep up with the number of concurrent threads that can be crunched by compute.
I don't think it is RDNA 3 maybe it is custom to Sony I just comment about the patent.
I don't care about all this story about RDNA 1, RDNA 2, RDNA 3 but about the feature set and the patent.
- Status
Similar threads
- Locked