GPU recommendation?

[c10ck3r]

Hey guys,
I'm looking at upgrading my home GPU with the primary use running Python code, but with idle time going to either mfaktc or CudaPm1. I've currently got a GTX 1050 with a Threadripper 1950X CPU and 32GB DDR4 RAM @ 2666, but I noticed that the GTX 1050 isn't increasing performance much over my old GTX 460. From what I've seen, the GTX 2070 Super is about the highest price I'd be willing to go, but is there a better option to get increased performance without spending as much money?
TIA

[hansl]

I recently picked up a couple GTX 1660 to upgrade the graphics in my older systems.
These seemed like a nice price/performance sweet spot to me, which gets about 1500 GHz-d/day on mfaktc under linux. I noticed the Windows performance isn't quite as good, being closer to 1300-1350 though. I haven't tried CudaPM1 on them.

They are going for about $220, so you could get 2 for less than the price of a 2070 super, and performance/watt is pretty much the same.

[M344587487]

An nvidia card for TF, a Radeon VII for P-1 or PRP.

[Mark Rose]

Take any RTX card over a GTX card for mfaktc.

[nomead]

Originally Posted by Mark Rose

Take any RTX card over a GTX card for mfaktc.

RTX20xx or GTX16xx over anything else for mfaktc, depending on the budget. GTX16 is Turing too, so it's quite fast for factoring (for the price). On the mersenne.ca performance chart, the 1660Ti is currently even a bit above the 1080Ti, and the 1660 (non-Ti) isn't far behind...