Add max supported SM for nvrtc-11.0 (#48151)
Summary:
Should fix the regression when nvrtc from CUDA-11.0 is used on the system with RTX3080
Addresses issue described in https://github.com/pytorch/pytorch/issues/47669#issuecomment-725073808
Pull Request resolved: https://github.com/pytorch/pytorch/pull/48151
Reviewed By: ngimel
Differential Revision: D25043899
Pulled By: malfet
fbshipit-source-id: 998ded59387e3971c2c1a5df4af595630515a72e