This version partitions the input keys using a simple mod operator on the key. Note that this assumes that the keys are uniformly random. If they are not, the partitioning step should be changed to hash the inputs. We do not do this, since for our use case the keys are already random, and so the partitions are expected to be approximately equal-sized.
This version is significantly faster the previous attempts.
This version partitions the input keys using a simple mod operator on the key. Note that this assumes that the keys are uniformly random. If they are not, the partitioning step should be changed to hash the inputs. We do not do this, since for our use case the keys are already random, and so the partitions are expected to be approximately equal-sized.
This version is significantly faster the previous attempts.