Basically we can run the gradient worklet and compute the Divergence, Vorticity, and QCriterion without ever having to store the gradient tensor in global memory