Open MPI logo

Open MPI User's Mailing List Archives

  |   Home   |   Support   |   FAQ   |   all Open MPI User's mailing list

Subject: Re: [OMPI users] SYSTEM CPU with OpenMPI 1.4.3
From: tmishima_at_[hidden]
Date: 2010-11-15 19:35:35


Hi,

 I did the test with the simple program as shown below.
  (I use mumps, which is a parallel linear solver.)

 This test program does nothing but just calls intialize &
 finalize routine of MUMPS & MPI.

      INCLUDE 'mpif.h'
      INCLUDE 'dmumps_struc.h'
      TYPE (DMUMPS_STRUC) MUMPS_PAR
c
      MUMPS_PAR%COMM = MPI_COMM_WORLD
      MUMPS_PAR%SYM = 1
      MUMPS_PAR%PAR = 1
      MUMPS_PAR%JOB = -1 ! INITIALIZE MUMPS
      CALL MPI_INIT(IERR)
      CALL DMUMPS(MUMPS_PAR)
c
      CALL MPI_COMM_RANK( MPI_COMM_WORLD, MYID, IERR )
      IF ( MYID .EQ. 0 ) CALL SLEEP(180) ! WAIT 180 SEC.
c
      MUMPS_PAR%JOB = -2 ! FINALIZE MUMPS
      CALL DMUMPS(MUMPS_PAR)
      CALL MPI_FINALIZE(IERR)
c
      END

command line : mpirun -host node06,node05 -np 16 ./testrun

Results:

 User & System CPUs are quite different between two versions.
 Waiting processes are counted as USER CPU in openmpi 1.4.2,
 while they are counted as USER & SYSTEM CPU in openmpi 1.4.3.
 Something in the MUMPS intialize routine might affect SYSTEM CPU.

          node06 CPU | node05 CPU
openmpi usr sys idl | usr sys idl
1.4.2 88 0 12 | 100 0 0
1.4.3 41 47 12 | 44 56 0

* process id of node06: 0-07
  process id of node05: 08-15
  ( total 16 process )

* see attacehd file which is a snap shot of GANGLIA.

(See attached file: openmpi143.JPG)


openmpi143.JPG