So, nothing at all. Have said it countless times that Gears 5 isn't a best benchmark to judge it and that too under controlled settings.
It was a nearly a straight up copy and paste job from x1x and the xsx gpu was able to perform as well as a 2080 did on same settings . Its not hard to believe in the fact that we know a 40CU RX 5700xt is able to compete with a RTX 2070 and why is so hard to believe that a 52CU RDNA 2.0 gpu would not match or beat a 2080/2080S? We are talking like 25% more CU's without even taking into account architecture improvements. a 52 CU RX 5700 would be in 2080 range as is.
Because we haven't seen anything about the RT/ML capabilities. How much die is dedicated to that? Also, if Epic's now deleted video about MaxQ performing better than PS5 for the UE5 demo is anything to go by, it doesn't paint a very good picture of XSX either.
Gears 5 wasn't even the most demanding game on PC. RDR 2 would have been a far better candidate to judge it.
Not saying it's impossible just there's not enough data on either side.
In pure rasterization performance we can do an educated guess in what to expect.And that the gears 5 is a good peak into what to expect.
The only RT performance info we have is the minecraft RT demo. But at the same time AMD in one of their RDNA 2.0 slides said " select lighting effects for real time gaming" Which tells me that AMD's RT will be singular effect just like current RTX cards do. Not being able to do multiple RT effects at the same time. But I suspect that AMD's solution wont stall their shader processors like RTX does not being utilized when RT is on, which Ampere is suppose to fix.
Log in to comment