1:2 for the titan V ~ 7.5 TeraFLOP/s
Radeon VII is 3.5 TeraFlops FP64, but it is on par with the 1080 TI for single precision, and besides, lacks the tensor cores if and when any projects start to use them. Also, I'm pretty sure the Radeon isn't going to run any projects that are CUDA based...
2080 Ti is at 420 GFLOP/s DP
1080 TI is at 354 GFLOP/s DP
2080 is at 315 GFLOP/s DP
(those are nVidia's "official" numbers - your mileage may vary. For example, the figure they quote for the 1080 Ti is based on a calculation of 2 IPC * 3584 cores * 1.582 boost clock. My FE card boosts to 1,911 GHz, so I would imagine that all (at least all the nVidia) figures are on the low side, but good luck actually realizing the theoretical peaks.)