I was thinking about this and here is a possible reason it is slower. The data most likely still resides within memory, so you take a hit for the overhead of having multiple servers trying to come together with a final results. Now, if you had a much larger set of information and it was primarily sitting on disk, it would be much faster parallel than serial. So, having multiple processes would speed up the build process by overcoming disk I/O.
______________________ Applications come and go,
but the data remains!
however, it didn't help me out.
1. i don't think 1000000 is small .
2. what do i need to do in order to use parallel
anything need to setupt first?
3. since i only have 8 cpu, don't you think setup degree to 20 is too big( i tried 20 degree, it didn't speed things up).