Cuda没有将大型阵列复制到设备上

I am new to CUDA, so I am sorry if I made any silly mistake, but this seems puzzling to me. The following code works perfectly for arrays of size up to 620 elements. When we change the NV def (number of vortexes) from 621 on, all arrays in the kernel become of NAN. I hope someone can explain this.I am new to CUDA, so I am sorry if I made any s