Kernel execution configuration

What’s meaning about the following code
bitonicSort<<<1, NUM, sizeof(int) * NUM>>>(dvalues);
Shouldn’t the execution configuration be dim3 type? NUM is int type.

yes that’ll default .y=.z=1

The single int provided is placed into the x field of the dim3. I’m not sure if this is done as an automatic upcast of an int to a dim3 or that nvcc handles the case specially.

Which means gird.x=1, grid.y=1, threads.x=NUM, threads.y=NUM?