/home/cholod/wd/peps/ROMS/romsagrif_r215/Roms_tools/Roms_Agrif/compute_starts_counts.h File Reference

This graph shows which files directly or indirectly include this file:

Go to the source code of this file.

Functions/Subroutines

!$Id type and which specify
grid!type of the field to be
read written from to netCDF
and record !number (if there is an unlimited dimension for that field in the!netCDF file)

in the case of MPI code the
data can still be!written into
a single or (3) multiple files containing data!individually for each MPI-node(this option is activated by!CPP-switch PARALLEL_FILES)

In the case of MPI single file
mode each!MPI node (one at-a-time) writes a rectangular block of array

In the case of MPI single file
mode each!MPI so that starts
depend on the position of node
jj on the processor!while
counts corresponds to the size
of MPI subdomains In the!case
of PARALLEL_FILES starts
corresponding to
horizontal!dimensions are both
equal counts correspond to
subdomains!with physical
boundaries but MPI halos
striped out this module
computes ranges jmax!which
define starting ending indices
of the portion of model
array!to be written into the
file[hence in all cases count(1)=imax-imin+1,!and
count(2)=jmax-jmin+1] This is
necessary because model
arrays!have slightly different
shapes than the corresponding
arrays in the!netCDF files The
differences are due to index shifting (netCDF!array index must always start from 1, while Fortran does not)

and stripping periodic
computational!margins integer start (4)

and stripping periodic
computational!margins integer count (4)

and stripping periodic
computational!margins integer
horiz_type count (1:2) correspond to XI-and start(i)

and stripping periodic
computational!margins integer
horiz_type while index is for
either count (i)

Variables

!$Id __pad9__

!$Id v

!$Id type and record

!$Id type and which specify
grid!type of the field to be
read written from to netCDF file

!$Id type and which specify
grid!type of the field to be
read written from to netCDF
and record compute start and
count arrays which
contain!starting indices and
proper counts corresponding to
the field data!stored in the
file!This module supports
three different policies of
reading writing!the data

In the case of MPI single file
mode each!MPI so that starts
depend on the position of node ii = '

In the case of MPI single file
mode each!MPI so that starts
depend on the position of node
jj on the processor grid

In the case of MPI single file
mode each!MPI so that starts
depend on the position of node
jj on the processor!while
counts corresponds to the size
of MPI subdomains In the!case
of PARALLEL_FILES starts
corresponding to
horizontal!dimensions are both
equal to

In the case of MPI single file
mode each!MPI so that starts
depend on the position of node
jj on the processor!while
counts corresponds to the size
of MPI subdomains In the!case
of PARALLEL_FILES starts
corresponding to
horizontal!dimensions are both
equal counts correspond to
subdomains!with physical
boundaries kept

In the case of MPI single file
mode each!MPI so that starts
depend on the position of node
jj on the processor!while
counts corresponds to the size
of MPI subdomains In the!case
of PARALLEL_FILES starts
corresponding to
horizontal!dimensions are both
equal counts correspond to
subdomains!with physical
boundaries but MPI halos
striped out !Additionally

In the case of MPI single file
mode each!MPI so that starts
depend on the position of node
jj on the processor!while
counts corresponds to the size
of MPI subdomains In the!case
of PARALLEL_FILES starts
corresponding to
horizontal!dimensions are both
equal counts correspond to
subdomains!with physical
boundaries but MPI halos
striped out this module
computes ranges imin = '

In the case of MPI single file
mode each!MPI so that starts
depend on the position of node
jj on the processor!while
counts corresponds to the size
of MPI subdomains In the!case
of PARALLEL_FILES starts
corresponding to
horizontal!dimensions are both
equal counts correspond to
subdomains!with physical
boundaries but MPI halos
striped out this module
computes ranges imax = '

In the case of MPI single file
mode each!MPI so that starts
depend on the position of node
jj on the processor!while
counts corresponds to the size
of MPI subdomains In the!case
of PARALLEL_FILES starts
corresponding to
horizontal!dimensions are both
equal counts correspond to
subdomains!with physical
boundaries but MPI halos
striped out this module
computes ranges jmin

!Fortran array dimension padding

and stripping periodic
computational!margins integer jmax = '

and stripping periodic
computational!margins integer & vert_type

and stripping periodic
computational!margins integer
horiz_type ierr

and stripping periodic
computational!margins integer
horiz_type !start

is for time record only then mynode

is for time record only then horiz_grid

is for time record only then horiz_type

is for time record only then c
C **& jj = '

Function Documentation

! $Id type and which specify grid !type of the field to be read written from to netCDF and record !number ( if there is an unlimited dimension for that field in the!netCDF file )

Type Constraints

and stripping periodic computational !margins integer horiz_type while index is for either count ( i )

and stripping periodic computational !margins integer horiz_type count ( 1:2 )

Type Constraints

and stripping periodic computational !margins integer count ( 4 )

Here is the caller graph for this function:

In the case of MPI single file mode each !MPI node ( one at-a- time )

Here is the caller graph for this function:

in the case of MPI code the data can still be !written into a single or ( 3 )

Type Constraints

Here is the caller graph for this function:

In the case of MPI single file mode each !MPI so that starts depend on the position of node jj on the processor !while counts corresponds to the size of MPI subdomains In the !case of PARALLEL_FILES starts corresponding to horizontal !dimensions are both equal counts correspond to subdomains !with physical boundaries but MPI halos striped out this module computes ranges jmax !which define starting ending indices of the portion of model array !to be written into the file [hence in all cases count(1)=imax-imin+1,!and count(2)=jmax-jmin+1] This is necessary because model arrays !have slightly different shapes than the corresponding arrays in the !netCDF files The differences are due to index shifting	(	netCDF!array index must always start from	1,
		while Fortran does	not
	)

Type Constraints

and stripping periodic computational !margins integer start ( 4 )

Type Constraints

Here is the caller graph for this function:

Variable Documentation

In the case of MPI single file mode each !MPI so that starts depend on the position of node jj on the processor !while counts corresponds to the size of MPI subdomains In the !case of PARALLEL_FILES starts corresponding to horizontal !dimensions are both equal counts correspond to subdomains !with physical boundaries but MPI halos striped out !Additionally

Definition at line 22 of file compute_starts_counts.h.

and stripping periodic computational !margins integer horiz_type !start

Definition at line 41 of file compute_starts_counts.h.

! $Id __pad9__

Definition at line 2 of file compute_starts_counts.h.

! $Id type and which specify grid !type of the field to be read written from to netCDF and record compute start and count arrays which contain !starting indices and proper counts corresponding to the field data !stored in the file ! This module supports three different policies of reading writing !the data

Definition at line 15 of file compute_starts_counts.h.

in the case of MPI code the data can still be written into a single file

Definition at line 2 of file compute_starts_counts.h.

In the case of MPI single file mode each !MPI so that starts depend on the position of node jj on the processor grid

Definition at line 22 of file compute_starts_counts.h.

is for time record only then horiz_grid

Definition at line 49 of file compute_starts_counts.h.

is for time record only then horiz_type

Definition at line 49 of file compute_starts_counts.h.

CC CC CC ROUTINE trcini pisces h CC ************************CC CC mo INTEGER jpan CHARACTER *clname INTEGER numdep REAL denitide REAL jpj REAL zdate0 REAL zmaskt c INCLUDE netcdf inc integer irec C C Compute record length for direct access FILE C rdt rfact2 C C INITIALISE DUST INPUT FROM ATMOSPHERE C C bioname endif nrec_dust irec endif enddo ierr

Initial value:

0            ! These are default settings. In all cases
      do i=1

Definition at line 41 of file compute_starts_counts.h.

is for time record only then c ii = '

Definition at line 22 of file compute_starts_counts.h.

is for time record only then c imax = '

Definition at line 22 of file compute_starts_counts.h.

is for time record only then c imin = '

Definition at line 22 of file compute_starts_counts.h.

is for time record only then c C jj = '

Definition at line 106 of file compute_starts_counts.h.

is for time record only then c C jmax = '

Definition at line 38 of file compute_starts_counts.h.

is for time record only then c C jmin

Definition at line 22 of file compute_starts_counts.h.

In the case of MPI single file mode each !MPI so that starts depend on the position of node jj on the processor !while counts corresponds to the size of MPI subdomains In the !case of PARALLEL_FILES starts corresponding to horizontal !dimensions are both equal counts correspond to subdomains !with physical boundaries kept

Definition at line 22 of file compute_starts_counts.h.

is for time record only then mynode

Definition at line 49 of file compute_starts_counts.h.

!Fortran array dimension padding

Definition at line 35 of file compute_starts_counts.h.

! $Id type and record

Definition at line 2 of file compute_starts_counts.h.

In the case of MPI single file mode each !MPI so that starts depend on the position of node jj on the processor !while counts corresponds to the size of MPI subdomains In the !case of PARALLEL_FILES starts corresponding to horizontal !dimensions are both equal to

Definition at line 22 of file compute_starts_counts.h.

! $Id v

Definition at line 2 of file compute_starts_counts.h.

is for time record only vert_type

Initial value:

type/4              ! Decode grid type into vertical
      horiz_type=type-4*vert_type   ! and horizontal grid types

Definition at line 39 of file compute_starts_counts.h.


Functions/Subroutines
!$Id type and which specify grid!type of the field to be read written from to netCDF and record	!number (if there is an unlimited dimension for that field in the!netCDF file)
in the case of MPI code the data can still be!written into a single	or (3) multiple files containing data!individually for each MPI-node(this option is activated by!CPP-switch PARALLEL_FILES)
In the case of MPI single file mode each!MPI	node (one at-a-time) writes a rectangular block of array
In the case of MPI single file mode each!MPI so that starts depend on the position of node jj on the processor!while counts corresponds to the size of MPI subdomains In the!case of PARALLEL_FILES starts corresponding to horizontal!dimensions are both equal counts correspond to subdomains!with physical boundaries but MPI halos striped out this module computes ranges jmax!which define starting ending indices of the portion of model array!to be written into the file[hence in all cases count(1)=imax-imin+1,!and count(2)=jmax-jmin+1] This is necessary because model arrays!have slightly different shapes than the corresponding arrays in the!netCDF files The differences are due to index	shifting (netCDF!array index must always start from 1, while Fortran does not)
and stripping periodic computational!margins integer	start (4)
and stripping periodic computational!margins integer	count (4)
and stripping periodic computational!margins integer horiz_type	count (1:2) correspond to XI-and start(i)
and stripping periodic computational!margins integer horiz_type while index is for either	count (i)
Variables
!$Id	__pad9__
!$Id	v
!$Id type and	record
!$Id type and which specify grid!type of the field to be read written from to netCDF	file
!$Id type and which specify grid!type of the field to be read written from to netCDF and record compute start and count arrays which contain!starting indices and proper counts corresponding to the field data!stored in the file!This module supports three different policies of reading writing!the	data
In the case of MPI single file mode each!MPI so that starts depend on the position of node	ii = '
In the case of MPI single file mode each!MPI so that starts depend on the position of node jj on the processor	grid
In the case of MPI single file mode each!MPI so that starts depend on the position of node jj on the processor!while counts corresponds to the size of MPI subdomains In the!case of PARALLEL_FILES starts corresponding to horizontal!dimensions are both equal	to
In the case of MPI single file mode each!MPI so that starts depend on the position of node jj on the processor!while counts corresponds to the size of MPI subdomains In the!case of PARALLEL_FILES starts corresponding to horizontal!dimensions are both equal counts correspond to subdomains!with physical boundaries	kept
In the case of MPI single file mode each!MPI so that starts depend on the position of node jj on the processor!while counts corresponds to the size of MPI subdomains In the!case of PARALLEL_FILES starts corresponding to horizontal!dimensions are both equal counts correspond to subdomains!with physical boundaries but MPI halos striped out	!Additionally
In the case of MPI single file mode each!MPI so that starts depend on the position of node jj on the processor!while counts corresponds to the size of MPI subdomains In the!case of PARALLEL_FILES starts corresponding to horizontal!dimensions are both equal counts correspond to subdomains!with physical boundaries but MPI halos striped out this module computes ranges	imin = '
In the case of MPI single file mode each!MPI so that starts depend on the position of node jj on the processor!while counts corresponds to the size of MPI subdomains In the!case of PARALLEL_FILES starts corresponding to horizontal!dimensions are both equal counts correspond to subdomains!with physical boundaries but MPI halos striped out this module computes ranges	imax = '
In the case of MPI single file mode each!MPI so that starts depend on the position of node jj on the processor!while counts corresponds to the size of MPI subdomains In the!case of PARALLEL_FILES starts corresponding to horizontal!dimensions are both equal counts correspond to subdomains!with physical boundaries but MPI halos striped out this module computes ranges	jmin
!Fortran array dimension	padding
and stripping periodic computational!margins integer	jmax = '
and stripping periodic computational!margins integer &	vert_type
and stripping periodic computational!margins integer horiz_type	ierr
and stripping periodic computational!margins integer horiz_type	!start
is for time record only then	mynode
is for time record only then	horiz_grid
is for time record only then	horiz_type
is for time record only then c C **&	jj = '