Is this grid spec OK?

for a problem with 1,056,768 particles with a blocksize of 64 I am specifiying a grid for each kernel<<<16512, 64>>>(…);

is this OK? The compiler is not complaining.

The compiler does not check the validity of the block and grid dimensions, but the dimensions you list should be fine. The limits are given in Appendix A.1.1 of the programming guide.