ParentFull threadbwat48·I think the limitation here is more with the process nodes then with the GPU architectureView on HN