As a point of clarification, the behavior of MPI_Comm_spawn (and
MPI_Comm_spawn_multiple) is such that it does not grow the MPI Universe
size, which is set when you lamboot. If you request N more processes by
using MPI_Comm_spawn then those processes will be distributed over the
resources in the universe essentially oversubscribing them, but will
not try to expand the universe by requesting more resources from PBS.
On Jan 23, 2005, at 7:21 PM, Anthony J. Ciani wrote:
> Hello All,
>
> On Sat, 22 Jan 2005, Siva Bandhamravuri wrote:
>> Hi,
>> I was wondering if there are any issues in dynamic process creation
>> with
>> MPI_Comm_spawn under the openPBS scheduler.
>> Is the MPI_Comm_spawn supported under openPBS. If yes, where are the
>> newly
>> created processes allocated. Because under openPBS, only a specific
>> set of
>> nodes is given to a MPI job during the job start where as we require
>> the newly
>> spawned processes to be allocated to a new node.
>
> I should clarify my previous response. I was assuming that you did
> not know the number of daughter processes to start a priori. If you
> DO KNOW the number of daughters you will need before hand (say you'll
> need 16 total, but 8 to start), then you could submit a 16 processor
> job and use "mpirun -np 8 n8-15 your_program" to start. This will
> create the first 8 instances on nodes 8 through 15. After this, you
> could use;
>
> MPI_Comm_spawn("your_child_program", argv, 8, info, 0, MPI_COMM_WORLD,
> child_comm, ierr);
>
> to have rank 0 in MPI_COMM_WORLD start 8 instances of
> "your_child_program", which will execute on nodes 0 through 7. It
> would then be up to you to figure out how to split the work up amongst
> your child programs, as each call to MPI_Comm_spawn starts new
> programs on n0 up.
>
>
> BTW, if your parent program will be idle while the children run, then
> you should just reuse the same nodes.
>
>
> ------------------------------------------------------------
> Anthony Ciani (aciani1_at_[hidden])
> Computational Condensed Matter Physics
> Department of Physics, University of Illinois, Chicago
> http://ciani.phy.uic.edu/~tony
> ------------------------------------------------------------
> _______________________________________________
> This list is archived at http://www.lam-mpi.org/MailArchives/lam/
>
----
Josh Hursey
jjhursey_at_[hidden]
http://www.lam-mpi.org/
|
