Changeset 622 for palm/trunk/SOURCE/transpose.f90

Timestamp:

Dec 10, 2010 8:08:13 AM (13 years ago)

Author:

raasch

Message:

New:
---

Optional barriers included in order to speed up collective operations
MPI_ALLTOALL and MPI_ALLREDUCE. This feature is controlled with new initial
parameter collective_wait. Default is .FALSE, but .TRUE. on SGI-type
systems. (advec_particles, advec_s_bc, buoyancy, check_for_restart,
cpu_statistics, data_output_2d, data_output_ptseries, flow_statistics,
global_min_max, inflow_turbulence, init_3d_model, init_particles, init_pegrid,
init_slope, parin, pres, poismg, set_particle_attributes, timestep,
read_var_list, user_statistics, write_compressed, write_var_list)

Adjustments for Kyushu Univ. (lcrte, ibmku). Concerning hybrid
(MPI/openMP) runs, the number of openMP threads per MPI tasks can now
be given as an argument to mrun-option -O. (mbuild, mrun, subjob)

Changed:

---

Initialization of the module command changed for SGI-ICE/lcsgi (mbuild, subjob)

Errors:

---

File:

• palm/trunk/SOURCE/transpose.f90 (modified) (9 diffs)

palm/trunk/SOURCE/transpose.f90

@@ -4 +4 @@
 ! Current revisions:
 ! -----------------
-! 
+! optional barriers included in order to speed up collective operations
 !
 ! Former revisions:
@@ -69 +69 @@
 !-- Transpose array
     CALL cpu_log( log_point_s(32), 'mpi_alltoall', 'start' )
+    IF ( collective_wait )  CALL MPI_BARRIER( comm2d, ierr )
     CALL MPI_ALLTOALL( f_inv(nys_x,nzb_x,0), sendrecvcount_xy, MPI_REAL, &
                        work(1),              sendrecvcount_xy, MPI_REAL, &
@@ -143 +144 @@
 !--    Transpose array
        CALL cpu_log( log_point_s(32), 'mpi_alltoall', 'start' )
+       IF ( collective_wait )  CALL MPI_BARRIER( comm2d, ierr )
        CALL MPI_ALLTOALL( work(1),          sendrecvcount_zx, MPI_REAL, &
                           f_inv(nys,nxl,1), sendrecvcount_zx, MPI_REAL, &
@@ -229 +231 @@
 !-- Transpose array
     CALL cpu_log( log_point_s(32), 'mpi_alltoall', 'start' )
+    IF ( collective_wait )  CALL MPI_BARRIER( comm2d, ierr )
     CALL MPI_ALLTOALL( work(1),              sendrecvcount_xy, MPI_REAL, &
                        f_inv(nys_x,nzb_x,0), sendrecvcount_xy, MPI_REAL, &
@@ -291 +294 @@
 !-- Transpose array
     CALL cpu_log( log_point_s(32), 'mpi_alltoall', 'start' )
+    IF ( collective_wait )  CALL MPI_BARRIER( comm2d, ierr )
     CALL MPI_ALLTOALL( f_inv(nxl,1,nys), sendrecvcount_xy, MPI_REAL, &
                        work(1),          sendrecvcount_xy, MPI_REAL, &
@@ -373 +377 @@
 !-- Transpose array
     CALL cpu_log( log_point_s(32), 'mpi_alltoall', 'start' )
+    IF ( collective_wait )  CALL MPI_BARRIER( comm2d, ierr )
     CALL MPI_ALLTOALL( f_inv(nxl_y,nzb_y,0), sendrecvcount_yz, MPI_REAL, &
                        work(1),              sendrecvcount_yz, MPI_REAL, &
@@ -454 +459 @@
 !-- Transpose array
     CALL cpu_log( log_point_s(32), 'mpi_alltoall', 'start' )
+    IF ( collective_wait )  CALL MPI_BARRIER( comm2d, ierr )
     CALL MPI_ALLTOALL( f_inv(nys,nxl,1), sendrecvcount_zx, MPI_REAL, &
                        work(1),          sendrecvcount_zx, MPI_REAL, &
@@ -528 +534 @@
 !--    Transpose array
        CALL cpu_log( log_point_s(32), 'mpi_alltoall', 'start' )
+       IF ( collective_wait )  CALL MPI_BARRIER( comm2d, ierr )
        CALL MPI_ALLTOALL( work(1),              sendrecvcount_yz, MPI_REAL, &
                           f_inv(nxl_y,nzb_y,0), sendrecvcount_yz, MPI_REAL, &
@@ -627 +634 @@
 !-- Transpose array
     CALL cpu_log( log_point_s(32), 'mpi_alltoall', 'start' )
+    IF ( collective_wait )  CALL MPI_BARRIER( comm2d, ierr )
     CALL MPI_ALLTOALL( f_inv(nys,nxl,1), sendrecvcount_zyd, MPI_REAL, &
                        work(1),          sendrecvcount_zyd, MPI_REAL, &