False.
That is a mobile GPU. Chassis cooling performance makes a large difference in terms of actual performance. Better to compare the 5500XT desktop part to the RX590 desktop part.
Summary 5.2Tflops RDNA ~= 7.1 Tflops Polaris.
This shows that a 4Tflop RDNA card would be around the same as a 5.4Tflop Polaris card. RDNA2 is likely to have an FPS/Flop uplift over RDNA otherwise there is no way AMD is hitting their +50% perf/watt metric.
Indeed they have. It will be due to the Ram config rather than the GPU horsepower because Series S has roughly equal GPU horsepower vs the One X.
Tflops measure one thing and if you know the rough FPS/Flop figure can be used to ballpark the performance of announced but unreleased GPUs. RDNA has far more FPS/Flop than GCN. RDNA 2 will be a further step.
This test shows that RDNA has a 39% FPS/Flop uplift over Polaris. The 5500XT vs RX590 comparison above also shows a similar FPS/Flop uplift. 4 Tflop RDNA would exceed a 5Tflop Polaris card, we already have empirical evidence of this yet you seem to think that a 4Tflop RDNA2 card will only match a 5Tflop Polaris one. Not sure what more I can say.
I don't conflate anything. I show that it is two routes to the same location. perf/watt increases show that X1X performance will be available in an ~ 80W SoC. An 80W SoC would be easy to cool inside the Series S chassis so that adds up. Next we have IPC and clockspeed. That also adds up to 4Tflop RDNA2 ~= to 6Tflop Polaris. Two ways of getting there and they both indicate the same thing. The Series S GPU is comparable to the X1X GPU.