CCL: Clever parallel Fast Multipole Algorithms?



 Sent to CCL by: "Don  Steiger" [sd00_2002(_)yahoo.com]
 I am looking for a parallel implementation of the FMM that has some clever
 (whatever that means) parallelization.  Does anybody know of any implementation
 that uses some interesting, novel, or clever approach, idea, or logic to do the
 parallelization?