Fastest partition method. Why?

[ds_is_fun]

I couldn't get the answer to this topic. I think the round robin would execute at the fastest rate. Am i right? Can someone pl. explain.
Thanks

[vmcburney]

Round robin is the fastest partition method because there is no intelligence to the partitioning, it is merely passing it around. I think this is the default partitioning method. There are some stages that require a specific type of partitioning in order to work correctly. For example the aggregation stage needs the data hash partitioned on the aggregation key fields. There is a good writeup on the different partitioning methods in the Parallel Job Developers Guide.