Hmmm, so they don't want to analyze Halo footage because it's compressed, but they will totally do look for VRS in ps5 game(compressed video) and say there isn't any(which goes perfectly with the narrative that ps5 doesn't have VRS)
I have a theory as to why PS5 doesn't have VRS, well, at least not as it's presented in a traditional RDNA 2 CU. If you read between the lines on Matt Hargett's tweet...
"Variable Rate Shading is nice for saving cycles, but VRS' optimization capability doesn't hold a handle to the Geometry Engine's capabilities. VRS without GE means you're still processing vertices you can/should eliminate in earlier stages to begin with. More free compute/memory."
My understanding of all these things is minimal at best, so take this with a grain of salt. Someone correct me if I'm wrong here, but from what I understand, VRS capabilities are within the compute unit itself?
Sony may have removed this feature because they're already doing this within the geometry engine, so
technically it has VRS, just not as it stands within the RDNA2 architecture. I know plenty of people poo-poo the idea of a part of the PS5 potentially being RDNA3, but it may not be that far fetched. Chances are PSVR2 will have gaze tracking, and this requires
foveated rendering (link is a interview with PlayStation's Dominic Mallison in 2019 about this tech). Again, my knowledge is limited, but if I were a betting man, I'd say PS5 is capable of this as it pertains to the necessity of PSVR2. VRS and mesh shading are more than likely built into the geometry engine and accelerated for the purpose of foveated rendering. It seems pretty logical to me, so
If true, maybe AMD will use Sony's implementation in future products. Could be complete bullshit, but I just keep making that connection. After all, Sony doesn't want to talk about all the PS5 capabilities since it will let the "cat out of the bag" on the existence of PSVR2.
Okay, I'm taking my tinfoil hat off now.