Drake before DLSS is 6x the performance of the X1
I think people forget this.
Drake isn't 6X Nintendo Switch when all is said and done.
It's 6X Nintendo Switch. Before DLSS.
According to Microsoft, Nintendo Switch is the equivalent of 0.75TF to Xbox One's 1.4TF.
6*0.75=4.5
4.5TF
Before DLSS.
That's assuming clock speeds don't increase.
I don't think there will be a single game for a very, very, VERY long time that can't survive, can't at least function and run on a 4.5TF machine. And I doubt DLSS would struggle to make them run pretty alright, even then.
Don't underestimate this thing.
(Some notes:
Nintendo Switch is not literally 0.75TF, it's 0.75TF as a proportion of the performance of Xbox One, where Xbox One's performance = 1.4TF and Xbox Series S performance = 4TF.
Why are these numbers weird and non-literal? Because you can't really compare desktop AMD to mobile Nvidia. However, Microsoft has done the conversion for us, one Nintendo Switch has the EQUIVALENT of about one half an Xbox One. A little over, even.
So, Drake is not LITERALLY 4.5TF, but Microsoft could compare it to Xbox One and say it is the EQUIVALENT PERFORMANCE of an Xbox One IF it had 4.5TF of performance.
As others have pointed out, likely the raw score is probably around the 3.45TF range, but "equivalent" to 4TF of AMD performance. Within spitting distance. Still, that's very, very commendable, and will likely be helped along by extras like its RT cores and Tensor cores.
This also isn't accounting for the CPU, which is bound to be weaker than Xbox Series S. Though it can be noted that it's possible that the system may have more RAM than Series S, and developers may delegate usually CPU tasks to tensor cores, so the gap might not be as gulchtious as it seems.)