Performance analysis of the runaway kernel test for the traditional scheme and the bucketing scheme. The evolution of the runaway kernel is simulated until t = 70. The dashed lines show fitted curves for simplified cost models linear and quadratic in n. The solid lines show the runtime for the event-averaged number of representative particles. For each tested configuration (solid horizontal lines), the initial number of representative particles (circle) and the maximum number of representative particles in the simulation (right arrow) are indicated. The vertical dotted lines indicate the number of representative particles n for which the simulation data of either scheme exceeds the sizes of the per-core L2 cache, the per-socket inclusive L3 cache, and the main memory of the benchmark machine.