Something I wanted to do for a while, since I have both cards, is test the performance gains from Pascal generation's biggest consumer graphics chip, GP102, to Ampere's, the similarly codenamed GA102. Of course, this would mean testing NVIDIA TITAN Xp (2017) against RTX 3090. Well, almost full performance on the latter; keep in mind that 3090 isn't the full GA102, it has 2 SM disabled (82/84), however it is so close (2.5%~) it might as well be the full chip.
Anyway, because I am a master procrastinator, and even with my new medication, I just can't seem to settle to do the testing fully... this might not get done. It might, though, so there's that!
Anyway, I did test two games I am playing right now, and surprisingly, no, it's not Warframe (I know, right?) So, here you go for two game tests. I'll give you a hint, GA102 nearly doubles GP102 performance, as I'd expect from the resources on tap... Also, I threw in a TU104 result from my friend Maik's RTX 2080 SUPER in Vermintide 2. (Thanks for testing, Maik!) Though I eyeballed the clock speeds from what most 2080S run at (1.8-1.9 GHz) and since Maik doesn't overclock, and it's a reasonable AIB model, it will be around that.
Anyway! The charts! Here!