Some comparisons on higher end hardware!
For folding and crunching with an NVidia GPU, the RTX 2080 Ti is probably the best out there, with occasional projects benefiting the RTX 2080/2080 Super, thanks to it's higher core boost clock.
Those projects also won't have enough parallel threads, to make fully use of all the 2080 Ti's cores.
But if you want to go a step beyond the 2080 Ti, these GPUs are phenomenal (and costly).
Enjoy reading up on some of the technical data (and don't buy their servers, as more than likely, the CPU is too slow for these cards and crunching/folding; for reference: a 2080 Super needs 3,1Ghz CPU to keep it fed; these servers only provide 2.2-2.3Ghz sustained):
https://lambdalabs.com/blog/best-gpu-tensorflow-2080-ti-vs-v100-vs-titan-v-vs-1080-ti-benchmark/

I see nothing about BOINC or Folding in this article and none of us are running any type of a Datacenter Environments or Deep Learning Training Centers.
This is about "Deep Learning GPU Benchmarks" for "Deep Learning Training" and would be best placed under Overclocking, Cooling and Benchmarking or even Computer Industry News