1. Why this condor job run time is always about twice of the local machine run time?
No idea but if your local machine has enough RAM and your condor node does not, the condor job will swap. Swapping to hard drive will drastically increase the runtime.
2. How SIZE is calculated?
Note that in some cases linux kernel will lie about memory usage. We have where we explicitly request memory > 0 just to override condor's autoguesstimated requirement. Because we know condor in that instance is wrong and all nodes that will run those jobs have enough RAM to do so.
Dimitri