[petsc-users] petsc4py and MPI.COMM_SELF.Spawn
Barry Smith
bsmith at mcs.anl.gov
Wed Mar 1 13:18:21 CST 2017
Try putting the import petsc4py and petsc4py init AFTER the import mpi4py
> On Mar 1, 2017, at 11:40 AM, Rodrigo Felicio <Rodrigo.Felicio at iongeo.com> wrote:
>
> Thanks, Barry and Matt, for your prompt responses. I really appreciated them.
>
> Sorry I forgot to mention that I am using petsc 3.6.1, petsc4py 3.6.0 and mpi4py 2.0.0. (and also tested with mpi4py 1.2.2)
>
> from running env I get:
>
> PYTHONPATH=/home/XXXXX/Enthought/Canopy_64bit/User/lib/python2.7/site-packages:/home/XXXXX/dev/myPy:/home/XXXXX/ipnotebooks
>
> and also
>
> PETSC_DIR=/home/XXXXX/mylocal/petsc-3.6.1
> PETSC_ARCH=arch-linux2-c
> I_MPI_ROOT=/apps/tools/centos54-x86_64-intel14//impi/4.1.3.04
>
>
> Funny thing is that only inside a python session I see the directory where both petsc4py and mpi4py are installed in the python path, i.e.,
>
> only after
> import sys
> sys.path
>
> I see '/home/XXXX/.local/lib/python2.7/site-packages'
>
>
> Anyway, after running
> time mpirun -n 1 python ./dyn_mem_ex.py
>
> I get the error msg (once I killed the process running dyn_mem_ex.py. Mind you I edited out the the IP of the machine with XXX.XX.XX.XX)
> =====================================================================================
> = BAD TERMINATION OF ONE OF YOUR APPLICATION PROCESSES
> = EXIT CODE: 15
> = CLEANING UP REMAINING PROCESSES
> = YOU CAN IGNORE THE BELOW CLEANUP MESSAGES
> =====================================================================================
> Fatal error in PMPI_Init_thread: Invalid port, error stack:
> MPIR_Init_thread(674).....:
> MPID_Init(320)............: spawned process group was unable to connect back to the parent on port <tag#0$description#v1n29$port#51319$ifname#XXX.XX.XX.XX$>
> MPID_Comm_connect(206)....:
> MPIDI_Comm_connect(579)...: Named port tag#0$description#v1n29$port#51319$ifname#XXX.XX.XX.XX$ does not exist
> MPIDI_Comm_connect(415)...:
> dequeue_and_set_error(628): Communication error with rank 0
> Fatal error in PMPI_Init_thread: Invalid port, error stack:
> MPIR_Init_thread(674)..:
> MPID_Init(320).........: spawned process group was unable to connect back to the parent on port <tag#0$description#v1n29$port#51319$ifname#XXX.XX.XX.XX$>
> MPID_Comm_connect(206).:
> MPIDI_Comm_connect(432): Named port tag#0$description#v1n29$port#51319$ifname#XXX.XX.XX.XX$ does not exist
> Fatal error in PMPI_Init_thread: Invalid port, error stack:
> MPIR_Init_thread(674)..:
> MPID_Init(320).........: spawned process group was unable to connect back to the parent on port <tag#0$description#v1n29$port#51319$ifname#XXX.XX.XX.XX$>
> MPID_Comm_connect(206).:
> MPIDI_Comm_connect(432): Named port tag#0$description#v1n29$port#51319$ifname#XXX.XX.XX.XX$ does not exist
> Fatal error in PMPI_Init_thread: Invalid port, error stack:
> MPIR_Init_thread(674)..:
> MPID_Init(320).........: spawned process group was unable to connect back to the parent on port <tag#0$description#v1n29$port#51319$ifname#XXX.XX.XX.XX$>
> MPID_Comm_connect(206).:
> MPIDI_Comm_connect(432): Named port tag#0$description#v1n29$port#51319$ifname#XXX.XX.XX.XX$ does not exist
> APPLICATION TERMINATED WITH THE EXIT STRING: Terminated (signal 15)
>
> real 0m9.987s
> user 0m36.714s
> sys 0m11.541s
>
> if I comment out the lines that import PETSc and initialize petsc4py on the code for the cpi.py (codes attached below)
>
> import pestc4py
> #petsc4py.init(sys.argv)
> #from petsc4py import PETSc
> from mpi4py import MPI
>
> Then it runs without problems and the output is
>
> time mpirun -n 1 python ./dyn_mem_ex.py
> proc 2 of 4
> proc 3 of 4 proc 1 of 4
>
> proc 0 of 4
> proc 0 of 4, Adim=[10]
> proc 1 of 4, Adim=[10]
> proc 2 of 4, Adim=[10]
> proc 3 of 4, Adim=[10]
> Adata = [ 0. 1. 2. 3. 4. 5. 6. 7. 8. 9.]Adata = [ 0. 1. 2. 3. 4. 5. 6. 7. 8. 9.]
>
> Adata = [ 0. 1. 2. 3. 4. 5. 6. 7. 8. 9.]Adata = [ 0. 1. 2. 3. 4. 5. 6. 7. 8. 9.]
>
> 3.14160098692
> 2.65258238441e-06
>
> real 0m0.535s
> user 0m0.431s
> sys 0m0.633s
>
>
>
> the codes that I used in this example, have just minor modifications compared to old examples from petsc4py. For reference, I am also attaching them
>
> #---------------------------------
> # dyn_mem_ex.py
> # ------------------------------------
> import numpy
> import sys
> import petsc4py
> #petsc4py.init(sys.argv)
> #from petsc4py import PETSc
> from mpi4py import MPI
> mypath = '/home/XXX/study/mpi4py/'
> comm = MPI.COMM_SELF.Spawn(sys.executable,
> args=[mypath + 'cpi.py'],
> maxprocs=4)
>
> N = numpy.array(100, 'i')
> Adata = numpy.array(numpy.arange(10), dtype='f')
> Adim = numpy.array(Adata.shape[0], dtype='i')
>
>
> comm.Bcast([N, MPI.INT], root=MPI.ROOT)
> comm.Bcast([Adim, MPI.INT], root=MPI.ROOT)
> comm.Bcast([Adata, MPI.FLOAT], root=MPI.ROOT)
> PI = numpy.array(0.0, 'd')
> comm.Reduce(None, [PI, MPI.DOUBLE],
> op=MPI.SUM, root=MPI.ROOT)
> print(PI)
> print(PI/numpy.pi - 1.0)
>
> comm.Disconnect()
>
>
> #---------------------------------
> # cpi.py
> # ------------------------------------
>
> import numpy
> import sys, petsc4py
> #petsc4py.init(sys.argv)
> #from petsc4py import PETSc
> from mpi4py import MPI
>
> parent = MPI.Comm.Get_parent()
> size = parent.Get_size()
> rank = parent.Get_rank()
>
> print("proc {} of {} ".format(rank, size))
> N = numpy.array(0, dtype='i')
> Adim = numpy.zeros(1, dtype='i')
>
> parent.Bcast([N, MPI.INT], root=0)
> parent.Bcast([Adim, MPI.INT], root=0)
>
> Adata = numpy.zeros(Adim[0], dtype='f')
> parent.Bcast([Adata, MPI.FLOAT],root=0)
> print("proc {} of {}, Adim={}".format(rank, size, Adim))
> print("Adata = {}".format(Adata))
> h = 1.0 / N; s = 0.0
> for i in range(rank, N, size):
> x = h * (i + 0.5)
> #print(rank,size,Adim.shape)
> s += 4.0 / (1.0 + x**2)
> PI = numpy.array(s * h, dtype='d')
> parent.Reduce([PI, MPI.DOUBLE], None,
> op=MPI.SUM, root=0)
>
> parent.Disconnect()
>
>
> kind regards,
> Rodrigo
>
>
> ________________________________
>
>
> This email and any files transmitted with it are confidential and are intended solely for the use of the individual or entity to whom they are addressed. If you are not the original recipient or the person responsible for delivering the email to the intended recipient, be advised that you have received this email in error, and that any use, dissemination, forwarding, printing, or copying of this email is strictly prohibited. If you received this email in error, please immediately notify the sender and delete the original.
>
More information about the petsc-users
mailing list