Open marcovirgolin opened 1 year ago
It is furthermore important to check that there will be no problem with RNG: it's OK if we have to sacrifice reproducibility (when > 1 thread is used), but NOT OK if different threads generate the same random numbers.
Would be great to have the possibility to run
gpg
in parallel.Probably the best way is to use
omp
and its pragma's.Things that can be parallelized:
Note that the computational bottlenecks are in reversed order w.r.t. the list just given (generating offspring, which includes fitness evaluation, is the most expensive).