The current test cases of scheduler_perf are basically simple; create Pods with a specific template (i.e., specific scheduling constraint etc) and measure the metrics.

But, the real cluster actually often has various unschedulable Pods, and each of them has different unschedulable plugins.
By adding such a scenario in scheduler-perf, we can observe QHint's impact on the scheduling throughput more, since QHints accurately selects Pods to move to activeQ/backoffQ based on each Pod's unschedulable reason.

/sig scheduling
/kind feature