Why do I calculate that the output is all zero? Is my call command wrong? (PGI, FORTRAN,CUDA)Please help me

Graphics card: NVIDIA Quadro gv100 32g
Environment: VS2017+SDK10.0+CUDA10.0+PGI 18.0
Programming language:Fortran
Command:pgfortran -Mcuda=cuda10.0,cc70,ptxinfo,maxregcount:32 1.cuf

result: the output of serial computing part is correct, but the output of parallel computing part is 0.

Sorry but there’s not enough information here to help determine why you’re getting incorrect results. Are you able to provide a small reproducing example?