|Original Message (ID '134536') By yehuda:|
|In Response To 'Re: Re: ParallelDo'
Sorry for the late response
If you are just reassigning it means memory copying of some sort
Now, for the model that Mathematica uses for parallel computation, using several kernels, means to copy the data to those kernels and this is additional overhead (thus poorer performance of the parallel implementation).
If on the other hand, calling QN1[j], QN2[j]
etc., performs a LONG computation, than parallelization would lead to better results.
So, without getting into too much details, you need to know how much computation (CPU time) vs memory copying is involved, and then decide according to the specific case|