The total number of rows each library virtualizes. Higher values increase initialization and scrolling load.
The number of times the same initial-render condition is measured. More rounds stabilize results but take longer.
The number of round-trip scrolls before samples are recorded. Warm-up reduces the influence of initial caching and layout preparation.
The number of scroll samples recorded in each initial-render round. More samples improve confidence in the median and p95.
The number of batches used to measure insert, move, remove, and height-change conditions. More batches reveal variation between repetitions.
The number of operations performed in each mutation batch. More samples stabilize the median and p95.
Actual time varies with device performance and adaptive early stopping.