I’m trying to investigate the performance of Thrust reduce_by_key. Unfortunately, I’m away from my work machine for the next few days so I can’t write some simple code to test this, so I was wondering if someone could provide me with some insight into its performance. Basically, I’m trying to determine how the algorithm scales with the number of keys. The data I would be calling it over would have few repetitions of keys. Would this adversely affect performance?