I want to get a CUDA occupancy from the Quadro RTX 6000 GPU.
But ,the RTX 6000 has 476 tensor cores as well as 4608 streaming multiprocessors. How can I get the CUDA occupancy considering of the tensor core?
I want to get a CUDA occupancy from the Quadro RTX 6000 GPU.
But ,the RTX 6000 has 476 tensor cores as well as 4608 streaming multiprocessors. How can I get the CUDA occupancy considering of the tensor core?