Changeset 2600 for palm/trunk/SOURCE/init_pegrid.f90
- Timestamp:
- Nov 1, 2017 2:11:20 PM (7 years ago)
- File:
-
- 1 edited
Legend:
- Unmodified
- Added
- Removed
-
palm/trunk/SOURCE/init_pegrid.f90
r2516 r2600 25 25 ! ----------------- 26 26 ! $Id$ 27 ! calculation of block-I/O quantitites removed (is now done in parin) 28 ! 29 ! 2516 2017-10-04 11:03:04Z suehring 27 30 ! Remove tabs 28 31 ! … … 209 212 coupling_topology, gathered_size, grid_level, & 210 213 grid_level_count, inflow_l, inflow_n, inflow_r, inflow_s, & 211 io_blocks, io_group, maximum_grid_level, & 212 maximum_parallel_io_streams, message_string, & 214 maximum_grid_level, message_string, & 213 215 mg_switch_to_pe0_level, momentum_advec, nest_bound_l, & 214 216 nest_bound_n, nest_bound_r, nest_bound_s, nest_domain, neutral, & … … 1372 1374 ENDIF 1373 1375 1374 !1375 !-- Calculate the number of groups into which parallel I/O is split.1376 !-- The default for files which are opened by all PEs (or where each1377 !-- PE opens his own independent file) is, that all PEs are doing input/output1378 !-- in parallel at the same time. This might cause performance or even more1379 !-- severe problems depending on the configuration of the underlying file1380 !-- system.1381 !-- First, set the default:1382 IF ( maximum_parallel_io_streams == -1 .OR. &1383 maximum_parallel_io_streams > numprocs ) THEN1384 maximum_parallel_io_streams = numprocs1385 ENDIF1386 1387 !1388 !-- Now calculate the number of io_blocks and the io_group to which the1389 !-- respective PE belongs. I/O of the groups is done in serial, but in parallel1390 !-- for all PEs belonging to the same group. A preliminary setting with myid1391 !-- based on MPI_COMM_WORLD has been done in parin.1392 io_blocks = numprocs / maximum_parallel_io_streams1393 io_group = MOD( myid+1, io_blocks )1394 1395 1396 1376 END SUBROUTINE init_pegrid
Note: See TracChangeset
for help on using the changeset viewer.