Re: [Gretl-devel] gretl-MPI process IDs
Am 14.01.2017 um 20:11 schrieb Allin Cottrell:
values) I noticed something that's a little awkward. Namely, for
almost
everything that's indexable in hansl, the index starts at 1, but with
MPI we've carried over the underlying MPI specification in which the
indexation of processes starts at 0
Yes, that struck me, too, but I decided to keep quiet about it ;-)
What I thought is, the term "mpirank" is so unintuitive and un-hanslish
that it is natural the indexing would also be un-hanslish.
2) If backward-compatibility turns out not to be a serious issue, do
we
think it's worthwhile to switch to 1-based MPI process IDs?
I agree that adjusting it to the 1-based indexing of the gretl world
would be good. And while we're moving away from the underlying MPI
terms, couldn't we also replace the term "rank"? For example you write
about the MPI process IDs, so why not "$mpiprocid" instead of
"$mpirank"?
BTW, this would also resolve the backwards incompatibility (which isn't
important here anyway, I think): Just define $mpiprocid = $mpirank + 1,
and allow both in parallel (sic!), at least for a while.
thanks,
sven