Work-preserving scaling results for the uniform-emission benchmark on a log-log scale. Wall-clock time (average of the last three cycle step times) is shown as a function of the number of MPI ranks for the OFI (RDMA) and point-to-point (P2P) backends. Dashed lines show the ideal A/P scaling anchored at 2240 ranks. The total number of cell crossings is held fixed by scaling the time step as Δt ∝ P^−4/3 while keeping 5000 cells per rank.