You take 5700XT and compare it to 2070. The cards are essentially even in performance. 5700XT has a bit more actual, sustained tflops than 2070. Hence why RDNA1 is not as efficient in perf/flop as Turing but it's a lot more efficient than either GCN4 or GCN5 were against Pascal and Turing.
See this is exactly what I'm talking about, I believe that this is false. 5700XT Game Clock is 1755MHz while 2070S it's 1770MHz. Remember AMD reports 3 numbers Base/Game/Boost, Base is same for both Game Clock AMD=Boost Clock Nvidia, Boost Clock AMD (max theoretical burst) is unreported by Nvidia.
If you watch comparison videos you see 2070S consistently running 50MHz+ higher than RX5700. Example:
Buy Cheap PC Games: https://www.g2a.com/r/gr-5bec979e83485Sapphire Radeon RX 5700 XTZOTAC GAMING RTX 2070 Superintel core i7-8700K OCGames: 1440p/Ultra1-Resi...
youtu.be
Where do you get that RX 5700 Xt has higher sustained clock from? AMD boost clock does not mean same thing as Nvidia boost clock, for AMD it is theoretical maximum burst speed and a reason they list the card as
UP TO 9.75TF. If you calculate TF based on same method(using Nvidia method), using Boost/Game clock then RX 5700 Xt is 8.9856TF and 2070S is 9.0624TF. This however is not accurate representation either as Nvidia cards, due to better thermals, consistently run higher MHz than AMD in actual game performance comparisons. However it is a lot more accurate than calculating TF for one off of maximum theoretical clock and for the other off of manufacturer guaranteed minimum clock.
