Had a hunch this might’ve been a thread already. So the benchmark can be optimized with additional masking right off the bat?