Titan RTX vs RTX 2080 Ti vs Ampere/RTX 3000 for DL Computer vision?

I’m working on semantic segmentation and GANs on images that range from 1000x1000 pixels to 4000x4000

So far, I’ve gotten DL models to work on our existing GeForce 1080 Ti, but I want to speedup the training

From what I’ve gathered, the Quadro RTX 8000/6000, Titan RTX, and RTX 2080 Ti are the best GPUs for computer vision-DL. Is this right? Are there other good GPUs for DL-Computer Vision?

I also read somewhere that unless you’re memory constrained, then it’s better to get 2x RTX 2080 Ti instead of the Titan or Quadro. Is this correct?

Price is not too much of an issue for me. But if 2x RTX 2080 Ti isn’t much worse than 1x Titan and has much better performance/price, then I would prefer the 2x RTX 2080 Ti

Or should I just wait for the Ampere and RTX GeForce 3000 series?