Thanks, I guess the only way to know is to run my own benchmarks. Perhaps they have their own partitioning code which allows for faster performance.