Why is cudaLaunchCooperativeKernel() returning not permitted?

Question

So I am using GTX 1050 with a compute capability of 6.1 with CUDA 11.0. I need to use grid synchronization in my program so cudaLaunchCooperativeKernel() is nee