see 

https://en.wikipedia.org/wiki/CUDA#GPUs_supported

and 

https://lammps.sandia.gov/doc/Build_extras.html