ports/openmpi
ports/openmpi
1
1
Форкнуть 0
Код Релизы Активность
d70e2e8c2b
openmpi/ompi/include/mpif-common.h

468 строки
19 KiB
C
Исходник Обычный вид История

Bring over changes from the /tmp/fortran-stuff series: - Make the F90 bindings compile and link properly with gfortran 4.0, 4.1, Intel 9.0, PGI 6.1, Sun (don't know version offhand -- the most current as of this writing, I think), and NAG 5.2, although some have limitations (e.g., NAG can't seem to handle the medium and large sizes) - Building the F90 "small" module size is now the default, even for developers - Split up mpif.h into multiple files because parts of it were toxic to the F90 bindings - Properly specify unsized/unshaped arrays to make the bindings work on all known compilers - Make ompi_info show Fortran 90 bindings size - XML somewhat lags the generated scripts as of this commit, but functionality was my main goal -- the XML can be updated later (if at all). This commit was SVN r10118.
2006-05-30 18:37:41 +04:00
! -*- fortran -*-
!
! Copyright (c) 2004-2006 The Trustees of Indiana University and Indiana
! University Research and Technology
! Corporation. All rights reserved.
! Copyright (c) 2004-2005 The University of Tennessee and The University
! of Tennessee Research Foundation. All rights
! reserved.
- Add the missing parts: add MPI_REAL2 to the end of the list of Fortran datatypes (mpif-common.h) and the list of registered datatypes: MOOG(REAL2). Configure and Compilation with ia32/gcc just finished, naturally without real2. This commit was SVN r15137.
2007-06-20 00:41:28 +04:00
! Copyright (c) 2004-2007 High Performance Computing Center Stuttgart,
Bring over changes from the /tmp/fortran-stuff series: - Make the F90 bindings compile and link properly with gfortran 4.0, 4.1, Intel 9.0, PGI 6.1, Sun (don't know version offhand -- the most current as of this writing, I think), and NAG 5.2, although some have limitations (e.g., NAG can't seem to handle the medium and large sizes) - Building the F90 "small" module size is now the default, even for developers - Split up mpif.h into multiple files because parts of it were toxic to the F90 bindings - Properly specify unsized/unshaped arrays to make the bindings work on all known compilers - Make ompi_info show Fortran 90 bindings size - XML somewhat lags the generated scripts as of this commit, but functionality was my main goal -- the XML can be updated later (if at all). This commit was SVN r10118.
2006-05-30 18:37:41 +04:00
! University of Stuttgart. All rights reserved.
! Copyright (c) 2004-2005 The Regents of the University of California.
! All rights reserved.
! Copyright (c) 2006 Cisco Systems, Inc. All rights reserved.
! $COPYRIGHT$
!
! Additional copyrights may follow
!
! $HEADER$
!
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
! WARNING WARNING WARNING WARNING WARNING WARNING WARNING WARNING WARNING
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
!
! Do ***not*** copy this file to the directory where your Fortran
! fortran application is compiled unless it is absolutely necessary! Most
! modern Fortran compilers now support the -I command line flag, which
! tells the compiler where to find .h files (specifically, this one). For
! example:
!
! shell$ mpif77 foo.f -o foo -I$OMPI_HOME/include
!
! will probably do the trick (assuming that you have set OMPI_HOME
! properly).
!
! That being said, OMPI's "mpif77" wrapper compiler should
! automatically include the -I option for you. The following command
! should be equivalent to the command listed above:
!
! shell$ mpif77 foo.f -o foo
!
! You should not copy this file to your local directory because it is
! possible that this file will be changed between versions of Open MPI.
! Indeed, this mpif.h is incompatible with the mpif.f of other
! implementations of MPI. Using this mpif.h with other implementations
! of MPI, or with other versions of Open MPI will result in undefined
! behavior (to include incorrect results, segmentation faults,
! unexplainable "hanging" in your application, etc.). Always use the
! -I command line option instead (or let mpif77 do it for you).
!
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
! WARNING WARNING WARNING WARNING WARNING WARNING WARNING WARNING WARNING
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
!
! This file contains the bulk of the Open MPI Fortran interface. It
! is included as a back-end file to both mpif.h (i.e., the
! standardized MPI Fortran header file) and mpi.f90 (the MPI-2
! Fortran module source file, found in ompi/mpi/f90).
!
! This file is marginally different than mpif.h. mpif.h includes
! some "external" statements that are not suitable for use with the
! MPI-2 F90 module, and therefore cannot be included in the mpi.f90
! source file. Hence, this file is essentially everything that
! needs to be in the standardized mpif.h *except* the "external"
! statements, and is therefore suitable to be included in mpi.f90.
!
! First, however, include some output from configure.
!
include 'mpif-config.h'
!
! MPI version
!
integer MPI_VERSION, MPI_SUBVERSION
parameter (MPI_VERSION=2)
parameter (MPI_SUBVERSION=0)
!
! Miscellaneous constants
!
integer MPI_ANY_SOURCE, MPI_ANY_TAG
integer MPI_PROC_NULL, MPI_MAX_PROCESSOR_NAME
Add ability to completely turn off MPI one-sided support, so that users can experiment with using ROMIO directly. This commit was SVN r15922.
2007-08-19 01:35:51 +04:00
integer MPI_ROOT
Bring over changes from the /tmp/fortran-stuff series: - Make the F90 bindings compile and link properly with gfortran 4.0, 4.1, Intel 9.0, PGI 6.1, Sun (don't know version offhand -- the most current as of this writing, I think), and NAG 5.2, although some have limitations (e.g., NAG can't seem to handle the medium and large sizes) - Building the F90 "small" module size is now the default, even for developers - Split up mpif.h into multiple files because parts of it were toxic to the F90 bindings - Properly specify unsized/unshaped arrays to make the bindings work on all known compilers - Make ompi_info show Fortran 90 bindings size - XML somewhat lags the generated scripts as of this commit, but functionality was my main goal -- the XML can be updated later (if at all). This commit was SVN r10118.
2006-05-30 18:37:41 +04:00
integer MPI_MAX_ERROR_STRING, MPI_UNDEFINED
integer MPI_CART, MPI_GRAPH, MPI_KEYVAL_INVALID
integer MPI_SOURCE, MPI_TAG, MPI_ERROR
integer MPI_TAG_UB, MPI_HOST, MPI_IO, MPI_WTIME_IS_GLOBAL
integer MPI_APPNUM, MPI_LASTUSEDCODE, MPI_UNIVERSE_SIZE
integer IMPI_CLIENT_SIZE, IMPI_CLIENT_COLOR
integer IMPI_HOST_SIZE, IMPI_HOST_COLOR
integer MPI_BSEND_OVERHEAD
integer MPI_MAX_INFO_KEY, MPI_MAX_INFO_VAL
integer MPI_MAX_PORT_NAME, MPI_MAX_OBJECT_NAME
integer MPI_ORDER_C, MPI_ORDER_FORTRAN
integer MPI_DISTRIBUTE_BLOCK, MPI_DISTRIBUTE_CYCLIC
integer MPI_DISTRIBUTE_NONE, MPI_DISTRIBUTE_DFLT_DARG
integer MPI_TYPECLASS_INTEGER, MPI_TYPECLASS_REAL
integer MPI_TYPECLASS_COMPLEX
* Add constants related to one-sided support that were missing from the fortran include files This commit was SVN r11416.
2006-08-25 01:14:05 +04:00
integer MPI_MODE_NOCHECK, MPI_MODE_NOPRECEDE, MPI_MODE_NOPUT
Fix a typo caught by MTT. This commit was SVN r11450.
2006-08-27 19:43:27 +04:00
integer MPI_MODE_NOSTORE, MPI_MODE_NOSUCCEED
* Add constants related to one-sided support that were missing from the fortran include files This commit was SVN r11416.
2006-08-25 01:14:05 +04:00
integer MPI_LOCK_EXCLUSIVE, MPI_LOCK_SHARED
Bring over changes from the /tmp/fortran-stuff series: - Make the F90 bindings compile and link properly with gfortran 4.0, 4.1, Intel 9.0, PGI 6.1, Sun (don't know version offhand -- the most current as of this writing, I think), and NAG 5.2, although some have limitations (e.g., NAG can't seem to handle the medium and large sizes) - Building the F90 "small" module size is now the default, even for developers - Split up mpif.h into multiple files because parts of it were toxic to the F90 bindings - Properly specify unsized/unshaped arrays to make the bindings work on all known compilers - Make ompi_info show Fortran 90 bindings size - XML somewhat lags the generated scripts as of this commit, but functionality was my main goal -- the XML can be updated later (if at all). This commit was SVN r10118.
2006-05-30 18:37:41 +04:00
integer MPI_WIN_BASE, MPI_WIN_SIZE, MPI_WIN_DISP_UNIT
parameter (MPI_ANY_SOURCE=-1)
parameter (MPI_ANY_TAG=-1)
parameter (MPI_PROC_NULL=-2)
parameter (MPI_ROOT=-4)
parameter (MPI_MAX_PROCESSOR_NAME=255)
parameter (MPI_MAX_ERROR_STRING=255)
parameter (MPI_UNDEFINED=-32766)
parameter (MPI_CART=1)
parameter (MPI_GRAPH=2)
parameter (MPI_KEYVAL_INVALID=-1)
parameter (MPI_SOURCE=1)
parameter (MPI_TAG=2)
parameter (MPI_ERROR=3)
parameter (MPI_TAG_UB=0)
parameter (MPI_HOST=1)
parameter (MPI_IO=2)
parameter (MPI_WTIME_IS_GLOBAL=3)
parameter (MPI_APPNUM=4)
parameter (MPI_LASTUSEDCODE=5)
parameter (MPI_UNIVERSE_SIZE=6)
Un-friggen' believable. The values for MPI_WIN_BASE, MPI_WIN_SIZE, and MPI_WIN_DISP_UNIT were off by one from their C counterparts. This fixes trac:304. This commit was SVN r11385. The following Trac tickets were found above: Ticket 304 --> https://svn.open-mpi.org/trac/ompi/ticket/304
2006-08-24 07:33:35 +04:00
parameter (MPI_WIN_BASE=7)
parameter (MPI_WIN_SIZE=8)
parameter (MPI_WIN_DISP_UNIT=9)
Bring over changes from the /tmp/fortran-stuff series: - Make the F90 bindings compile and link properly with gfortran 4.0, 4.1, Intel 9.0, PGI 6.1, Sun (don't know version offhand -- the most current as of this writing, I think), and NAG 5.2, although some have limitations (e.g., NAG can't seem to handle the medium and large sizes) - Building the F90 "small" module size is now the default, even for developers - Split up mpif.h into multiple files because parts of it were toxic to the F90 bindings - Properly specify unsized/unshaped arrays to make the bindings work on all known compilers - Make ompi_info show Fortran 90 bindings size - XML somewhat lags the generated scripts as of this commit, but functionality was my main goal -- the XML can be updated later (if at all). This commit was SVN r10118.
2006-05-30 18:37:41 +04:00
parameter (IMPI_CLIENT_SIZE=10)
parameter (IMPI_CLIENT_COLOR=11)
parameter (IMPI_HOST_SIZE=12)
parameter (IMPI_HOST_COLOR=13)
parameter (MPI_BSEND_OVERHEAD=128)
parameter (MPI_MAX_INFO_KEY=35)
parameter (MPI_MAX_INFO_VAL=255)
Merge the ORTE devel branch into the main trunk. Details of what this means will be circulated separately. Remains to be tested to ensure everything came over cleanly, so please continue to withhold commits a little longer This commit was SVN r17632.
2008-02-28 04:57:57 +03:00
parameter (MPI_MAX_PORT_NAME=255)
Bring over changes from the /tmp/fortran-stuff series: - Make the F90 bindings compile and link properly with gfortran 4.0, 4.1, Intel 9.0, PGI 6.1, Sun (don't know version offhand -- the most current as of this writing, I think), and NAG 5.2, although some have limitations (e.g., NAG can't seem to handle the medium and large sizes) - Building the F90 "small" module size is now the default, even for developers - Split up mpif.h into multiple files because parts of it were toxic to the F90 bindings - Properly specify unsized/unshaped arrays to make the bindings work on all known compilers - Make ompi_info show Fortran 90 bindings size - XML somewhat lags the generated scripts as of this commit, but functionality was my main goal -- the XML can be updated later (if at all). This commit was SVN r10118.
2006-05-30 18:37:41 +04:00
parameter (MPI_MAX_OBJECT_NAME=63)
parameter (MPI_ORDER_C=0)
parameter (MPI_ORDER_FORTRAN=1)
parameter (MPI_DISTRIBUTE_BLOCK=0)
parameter (MPI_DISTRIBUTE_CYCLIC=1)
parameter (MPI_DISTRIBUTE_NONE=2)
parameter (MPI_DISTRIBUTE_DFLT_DARG=-1)
parameter (MPI_TYPECLASS_INTEGER=1)
parameter (MPI_TYPECLASS_REAL=2)
parameter (MPI_TYPECLASS_COMPLEX=3)
* Add constants related to one-sided support that were missing from the fortran include files This commit was SVN r11416.
2006-08-25 01:14:05 +04:00
parameter (MPI_MODE_NOCHECK=1)
parameter (MPI_MODE_NOPRECEDE=2)
parameter (MPI_MODE_NOPUT=4)
parameter (MPI_MODE_NOSTORE=8)
parameter (MPI_MODE_NOSUCCEED=16)
parameter (MPI_LOCK_EXCLUSIVE=1)
parameter (MPI_LOCK_SHARED=2)
Bring over changes from the /tmp/fortran-stuff series: - Make the F90 bindings compile and link properly with gfortran 4.0, 4.1, Intel 9.0, PGI 6.1, Sun (don't know version offhand -- the most current as of this writing, I think), and NAG 5.2, although some have limitations (e.g., NAG can't seem to handle the medium and large sizes) - Building the F90 "small" module size is now the default, even for developers - Split up mpif.h into multiple files because parts of it were toxic to the F90 bindings - Properly specify unsized/unshaped arrays to make the bindings work on all known compilers - Make ompi_info show Fortran 90 bindings size - XML somewhat lags the generated scripts as of this commit, but functionality was my main goal -- the XML can be updated later (if at all). This commit was SVN r10118.
2006-05-30 18:37:41 +04:00
!
Clean up the Fortran MPI sentinel values per problem reported on the users mailing list: http://www.open-mpi.org/community/lists/users/2006/07/1680.php Warning: this log message is not for the weak. Read at your own risk. The problem was that we had several variables in Fortran common blocks of various types, but their C counterparts were all of a type equivalent to a fortran double complex. This didn't seem to matter for the compilers that we tested, but we never tested static builds (which is where this problem seems to occur, at least with the Intel compiler: the linker compilains that the variable in the common block in the user's .o file was of one size/alignment but the one in the C library was a different size/alignment). So this patch fixes the sizes/types of the Fortran common block variables and their corresponding C instantiations to be of the same sizes/types. But wait, there's more. We recently introduced a fix for the OSX linker where some C versions of the fortran common block variables (e.g., _ompi_fortran_status_ignore) were not being found when linking ompi_info (!). Further research shows that the code path for ompi_info to require ompi_fortran_status_ignore is, unfortunately, necessary (a quirk of how various components pull in different portions of the code base -- nothing in ompi_info itself requires fortran or MPI knowledge, of course). Hence, the real problem was that there was no code path from ompi_info to the portion of the code base where the C globals corresponding to the Fortran common block variables were instantiated. This is because the OSX linker does not automatically pull in .o files that only contain unintialized global variables; the OSX linker typically only pulls in a .o file from a library if it either has a function that is used or have a global variable that is initialized (that's the short version; lots of details and corner cases omitted). Hence, we changed the global C variables corresponding to the fortran common blocks to be initialized, thereby causing the OSX linker to pull them in automatically -- problem solved. At the same time, we moved the constants to another .c file with a function, just for good measure. However, this didn't really solve the problem: 1. The function in the file with the C versions of the fortran common block variables (ompi/mpi/f77/test_constants_f.c) did not have a code path that was reachable from ompi_info, so the only reason that the constants were found (on OSX) was because they were initialized in the global scope (i.e., causing the OSX compiler to pull in that .o file). 2. Initializing these variable in the global scope causes problems for some linkers where -- once all the size/type problems mentioned above were fixed -- the alignments of fortran common blocks and C global variables do not match (even though the types of the Fortran and C variables match -- wow!). Hence, initializing the C variables would not necessarily match the alignment of what Fortran expected, and the linker would issue a warning (i.e., the alignment warnings referenced in the original post). The solution is two-fold: 1. Move the Fortran variables from test_constants_f.c to ompi/mpi/runtime/ompi_mpi_init.c where there are other global constants that *are* initialized (that had nothing to do with fortran, so the alignment issues described above are not a factor), and therefore all linkers (including the OSX linker) will pull in this .o file and find all the symbols that it needs. 2. Do not initialize the C variables corresponding to the Fortran common blocks in the global scope. Indeed, never initialize them at all (because we never need their *values* - we only check for their *locations*). Since nothing is ever written to these variables (particularly in the global scope), the linker does not see any alignment differences during initialization, but does make both the C and Fortran variables have the same addresses (this method has been working in LAM/MPI for over a decade). There were some comments here in the OMPI code base and in the LAM code base that stated/implied that C variables corresponding to Fortran common blocks had to have the same alignment as the Fortran common blocks (i.e., 16). There were attempts in both code bases to ensure that this was true. However, the attempts were wrong (in both code bases), and I have now read enough Fortran compiler documentation to convince myself that matching alignments is not required (indeed, it's beyond our control). As long as C variables corresponding to Fortran common blocks are not initialized in the global scope, the linker will "figure it out" and adjust the alignment to whatever is required (i.e., the greater of the alignments). Specifically (to counter comments that no longer exist in the OMPI code base but still exist in the LAM code base): - there is no need to make attempts to specially align C variables corresponding to Fortran common blocks - the types and sizes of C variables corresponding to Fortran common blocks should match, but do not need to be on any particular alignment Finally, as a side effect of this effort, I found a bunch of inconsistencies with the intent of status/array_of_statuses parameters. For all the functions that I modified they should be "out" (not inout). This commit was SVN r11057.
2006-07-31 19:07:09 +04:00
! MPI sentinel values
!
! Several of these types were chosen with care to match specific
! overloaded functions in the F90 bindings. They should also match
! the types of their corresponding C variables. Do not arbitrarily
! change their types without also updating the F90 bindings and
! their corresponding types in ompi/mpi/f77/constants.h and
! ompi/mpi/runtime/ompi_init.c!
!
! MPI_BOTTOM is only used where choice buffers can be used (meaning
! that we already have overloaded F90 bindings for all available
! types), so any type is fine.
integer MPI_BOTTOM
! MPI_IN_PLACE has the same rationale as MPI_BOTTOM.
integer MPI_IN_PLACE
! Making MPI_ARGV_NULL be the same type as the parameter that is
! exepected in the F90 binding for MPI_COMM_SPAWN means that we
! don't need another binding for MPI_COMM_SPAWN.
Bring over changes from the /tmp/fortran-stuff series: - Make the F90 bindings compile and link properly with gfortran 4.0, 4.1, Intel 9.0, PGI 6.1, Sun (don't know version offhand -- the most current as of this writing, I think), and NAG 5.2, although some have limitations (e.g., NAG can't seem to handle the medium and large sizes) - Building the F90 "small" module size is now the default, even for developers - Split up mpif.h into multiple files because parts of it were toxic to the F90 bindings - Properly specify unsized/unshaped arrays to make the bindings work on all known compilers - Make ompi_info show Fortran 90 bindings size - XML somewhat lags the generated scripts as of this commit, but functionality was my main goal -- the XML can be updated later (if at all). This commit was SVN r10118.
2006-05-30 18:37:41 +04:00
character MPI_ARGV_NULL(1)
Clean up the Fortran MPI sentinel values per problem reported on the users mailing list: http://www.open-mpi.org/community/lists/users/2006/07/1680.php Warning: this log message is not for the weak. Read at your own risk. The problem was that we had several variables in Fortran common blocks of various types, but their C counterparts were all of a type equivalent to a fortran double complex. This didn't seem to matter for the compilers that we tested, but we never tested static builds (which is where this problem seems to occur, at least with the Intel compiler: the linker compilains that the variable in the common block in the user's .o file was of one size/alignment but the one in the C library was a different size/alignment). So this patch fixes the sizes/types of the Fortran common block variables and their corresponding C instantiations to be of the same sizes/types. But wait, there's more. We recently introduced a fix for the OSX linker where some C versions of the fortran common block variables (e.g., _ompi_fortran_status_ignore) were not being found when linking ompi_info (!). Further research shows that the code path for ompi_info to require ompi_fortran_status_ignore is, unfortunately, necessary (a quirk of how various components pull in different portions of the code base -- nothing in ompi_info itself requires fortran or MPI knowledge, of course). Hence, the real problem was that there was no code path from ompi_info to the portion of the code base where the C globals corresponding to the Fortran common block variables were instantiated. This is because the OSX linker does not automatically pull in .o files that only contain unintialized global variables; the OSX linker typically only pulls in a .o file from a library if it either has a function that is used or have a global variable that is initialized (that's the short version; lots of details and corner cases omitted). Hence, we changed the global C variables corresponding to the fortran common blocks to be initialized, thereby causing the OSX linker to pull them in automatically -- problem solved. At the same time, we moved the constants to another .c file with a function, just for good measure. However, this didn't really solve the problem: 1. The function in the file with the C versions of the fortran common block variables (ompi/mpi/f77/test_constants_f.c) did not have a code path that was reachable from ompi_info, so the only reason that the constants were found (on OSX) was because they were initialized in the global scope (i.e., causing the OSX compiler to pull in that .o file). 2. Initializing these variable in the global scope causes problems for some linkers where -- once all the size/type problems mentioned above were fixed -- the alignments of fortran common blocks and C global variables do not match (even though the types of the Fortran and C variables match -- wow!). Hence, initializing the C variables would not necessarily match the alignment of what Fortran expected, and the linker would issue a warning (i.e., the alignment warnings referenced in the original post). The solution is two-fold: 1. Move the Fortran variables from test_constants_f.c to ompi/mpi/runtime/ompi_mpi_init.c where there are other global constants that *are* initialized (that had nothing to do with fortran, so the alignment issues described above are not a factor), and therefore all linkers (including the OSX linker) will pull in this .o file and find all the symbols that it needs. 2. Do not initialize the C variables corresponding to the Fortran common blocks in the global scope. Indeed, never initialize them at all (because we never need their *values* - we only check for their *locations*). Since nothing is ever written to these variables (particularly in the global scope), the linker does not see any alignment differences during initialization, but does make both the C and Fortran variables have the same addresses (this method has been working in LAM/MPI for over a decade). There were some comments here in the OMPI code base and in the LAM code base that stated/implied that C variables corresponding to Fortran common blocks had to have the same alignment as the Fortran common blocks (i.e., 16). There were attempts in both code bases to ensure that this was true. However, the attempts were wrong (in both code bases), and I have now read enough Fortran compiler documentation to convince myself that matching alignments is not required (indeed, it's beyond our control). As long as C variables corresponding to Fortran common blocks are not initialized in the global scope, the linker will "figure it out" and adjust the alignment to whatever is required (i.e., the greater of the alignments). Specifically (to counter comments that no longer exist in the OMPI code base but still exist in the LAM code base): - there is no need to make attempts to specially align C variables corresponding to Fortran common blocks - the types and sizes of C variables corresponding to Fortran common blocks should match, but do not need to be on any particular alignment Finally, as a side effect of this effort, I found a bunch of inconsistencies with the intent of status/array_of_statuses parameters. For all the functions that I modified they should be "out" (not inout). This commit was SVN r11057.
2006-07-31 19:07:09 +04:00
! The array_of_argv parameter in the F90 bindings for
! MPI_COMM_SPAWN_MULTIPLE takes a variable number of dimensions
! (specified by the "count" parameter), so it's not possible to have
! a single variable match all possible values. Hence, make it an
! entirely different type (one that would never likely be used by a
! correct program, e.g., double) and have a separate F90 binding for
! matching just this type.
double precision MPI_ARGVS_NULL
! MPI_ERRCODES_IGNORE has similar rationale to MPI_ARGV_NULL. The
! F77 functions are all smart enough to check that the errcodes
! parameter is not ERRCODES_IGNORE before assigning values into it
! (hence, the fact that this is an array of only 1 element does not
! matter -- we'll never overrun it because we never assign values
! into it).
Bring over changes from the /tmp/fortran-stuff series: - Make the F90 bindings compile and link properly with gfortran 4.0, 4.1, Intel 9.0, PGI 6.1, Sun (don't know version offhand -- the most current as of this writing, I think), and NAG 5.2, although some have limitations (e.g., NAG can't seem to handle the medium and large sizes) - Building the F90 "small" module size is now the default, even for developers - Split up mpif.h into multiple files because parts of it were toxic to the F90 bindings - Properly specify unsized/unshaped arrays to make the bindings work on all known compilers - Make ompi_info show Fortran 90 bindings size - XML somewhat lags the generated scripts as of this commit, but functionality was my main goal -- the XML can be updated later (if at all). This commit was SVN r10118.
2006-05-30 18:37:41 +04:00
integer MPI_ERRCODES_IGNORE(1)
Clean up the Fortran MPI sentinel values per problem reported on the users mailing list: http://www.open-mpi.org/community/lists/users/2006/07/1680.php Warning: this log message is not for the weak. Read at your own risk. The problem was that we had several variables in Fortran common blocks of various types, but their C counterparts were all of a type equivalent to a fortran double complex. This didn't seem to matter for the compilers that we tested, but we never tested static builds (which is where this problem seems to occur, at least with the Intel compiler: the linker compilains that the variable in the common block in the user's .o file was of one size/alignment but the one in the C library was a different size/alignment). So this patch fixes the sizes/types of the Fortran common block variables and their corresponding C instantiations to be of the same sizes/types. But wait, there's more. We recently introduced a fix for the OSX linker where some C versions of the fortran common block variables (e.g., _ompi_fortran_status_ignore) were not being found when linking ompi_info (!). Further research shows that the code path for ompi_info to require ompi_fortran_status_ignore is, unfortunately, necessary (a quirk of how various components pull in different portions of the code base -- nothing in ompi_info itself requires fortran or MPI knowledge, of course). Hence, the real problem was that there was no code path from ompi_info to the portion of the code base where the C globals corresponding to the Fortran common block variables were instantiated. This is because the OSX linker does not automatically pull in .o files that only contain unintialized global variables; the OSX linker typically only pulls in a .o file from a library if it either has a function that is used or have a global variable that is initialized (that's the short version; lots of details and corner cases omitted). Hence, we changed the global C variables corresponding to the fortran common blocks to be initialized, thereby causing the OSX linker to pull them in automatically -- problem solved. At the same time, we moved the constants to another .c file with a function, just for good measure. However, this didn't really solve the problem: 1. The function in the file with the C versions of the fortran common block variables (ompi/mpi/f77/test_constants_f.c) did not have a code path that was reachable from ompi_info, so the only reason that the constants were found (on OSX) was because they were initialized in the global scope (i.e., causing the OSX compiler to pull in that .o file). 2. Initializing these variable in the global scope causes problems for some linkers where -- once all the size/type problems mentioned above were fixed -- the alignments of fortran common blocks and C global variables do not match (even though the types of the Fortran and C variables match -- wow!). Hence, initializing the C variables would not necessarily match the alignment of what Fortran expected, and the linker would issue a warning (i.e., the alignment warnings referenced in the original post). The solution is two-fold: 1. Move the Fortran variables from test_constants_f.c to ompi/mpi/runtime/ompi_mpi_init.c where there are other global constants that *are* initialized (that had nothing to do with fortran, so the alignment issues described above are not a factor), and therefore all linkers (including the OSX linker) will pull in this .o file and find all the symbols that it needs. 2. Do not initialize the C variables corresponding to the Fortran common blocks in the global scope. Indeed, never initialize them at all (because we never need their *values* - we only check for their *locations*). Since nothing is ever written to these variables (particularly in the global scope), the linker does not see any alignment differences during initialization, but does make both the C and Fortran variables have the same addresses (this method has been working in LAM/MPI for over a decade). There were some comments here in the OMPI code base and in the LAM code base that stated/implied that C variables corresponding to Fortran common blocks had to have the same alignment as the Fortran common blocks (i.e., 16). There were attempts in both code bases to ensure that this was true. However, the attempts were wrong (in both code bases), and I have now read enough Fortran compiler documentation to convince myself that matching alignments is not required (indeed, it's beyond our control). As long as C variables corresponding to Fortran common blocks are not initialized in the global scope, the linker will "figure it out" and adjust the alignment to whatever is required (i.e., the greater of the alignments). Specifically (to counter comments that no longer exist in the OMPI code base but still exist in the LAM code base): - there is no need to make attempts to specially align C variables corresponding to Fortran common blocks - the types and sizes of C variables corresponding to Fortran common blocks should match, but do not need to be on any particular alignment Finally, as a side effect of this effort, I found a bunch of inconsistencies with the intent of status/array_of_statuses parameters. For all the functions that I modified they should be "out" (not inout). This commit was SVN r11057.
2006-07-31 19:07:09 +04:00
! MPI_STATUS_IGNORE has similar rationale to MPI_ERRCODES_IGNORE.
Bring over changes from the /tmp/fortran-stuff series: - Make the F90 bindings compile and link properly with gfortran 4.0, 4.1, Intel 9.0, PGI 6.1, Sun (don't know version offhand -- the most current as of this writing, I think), and NAG 5.2, although some have limitations (e.g., NAG can't seem to handle the medium and large sizes) - Building the F90 "small" module size is now the default, even for developers - Split up mpif.h into multiple files because parts of it were toxic to the F90 bindings - Properly specify unsized/unshaped arrays to make the bindings work on all known compilers - Make ompi_info show Fortran 90 bindings size - XML somewhat lags the generated scripts as of this commit, but functionality was my main goal -- the XML can be updated later (if at all). This commit was SVN r10118.
2006-05-30 18:37:41 +04:00
integer MPI_STATUS_IGNORE(MPI_STATUS_SIZE)
Clean up the Fortran MPI sentinel values per problem reported on the users mailing list: http://www.open-mpi.org/community/lists/users/2006/07/1680.php Warning: this log message is not for the weak. Read at your own risk. The problem was that we had several variables in Fortran common blocks of various types, but their C counterparts were all of a type equivalent to a fortran double complex. This didn't seem to matter for the compilers that we tested, but we never tested static builds (which is where this problem seems to occur, at least with the Intel compiler: the linker compilains that the variable in the common block in the user's .o file was of one size/alignment but the one in the C library was a different size/alignment). So this patch fixes the sizes/types of the Fortran common block variables and their corresponding C instantiations to be of the same sizes/types. But wait, there's more. We recently introduced a fix for the OSX linker where some C versions of the fortran common block variables (e.g., _ompi_fortran_status_ignore) were not being found when linking ompi_info (!). Further research shows that the code path for ompi_info to require ompi_fortran_status_ignore is, unfortunately, necessary (a quirk of how various components pull in different portions of the code base -- nothing in ompi_info itself requires fortran or MPI knowledge, of course). Hence, the real problem was that there was no code path from ompi_info to the portion of the code base where the C globals corresponding to the Fortran common block variables were instantiated. This is because the OSX linker does not automatically pull in .o files that only contain unintialized global variables; the OSX linker typically only pulls in a .o file from a library if it either has a function that is used or have a global variable that is initialized (that's the short version; lots of details and corner cases omitted). Hence, we changed the global C variables corresponding to the fortran common blocks to be initialized, thereby causing the OSX linker to pull them in automatically -- problem solved. At the same time, we moved the constants to another .c file with a function, just for good measure. However, this didn't really solve the problem: 1. The function in the file with the C versions of the fortran common block variables (ompi/mpi/f77/test_constants_f.c) did not have a code path that was reachable from ompi_info, so the only reason that the constants were found (on OSX) was because they were initialized in the global scope (i.e., causing the OSX compiler to pull in that .o file). 2. Initializing these variable in the global scope causes problems for some linkers where -- once all the size/type problems mentioned above were fixed -- the alignments of fortran common blocks and C global variables do not match (even though the types of the Fortran and C variables match -- wow!). Hence, initializing the C variables would not necessarily match the alignment of what Fortran expected, and the linker would issue a warning (i.e., the alignment warnings referenced in the original post). The solution is two-fold: 1. Move the Fortran variables from test_constants_f.c to ompi/mpi/runtime/ompi_mpi_init.c where there are other global constants that *are* initialized (that had nothing to do with fortran, so the alignment issues described above are not a factor), and therefore all linkers (including the OSX linker) will pull in this .o file and find all the symbols that it needs. 2. Do not initialize the C variables corresponding to the Fortran common blocks in the global scope. Indeed, never initialize them at all (because we never need their *values* - we only check for their *locations*). Since nothing is ever written to these variables (particularly in the global scope), the linker does not see any alignment differences during initialization, but does make both the C and Fortran variables have the same addresses (this method has been working in LAM/MPI for over a decade). There were some comments here in the OMPI code base and in the LAM code base that stated/implied that C variables corresponding to Fortran common blocks had to have the same alignment as the Fortran common blocks (i.e., 16). There were attempts in both code bases to ensure that this was true. However, the attempts were wrong (in both code bases), and I have now read enough Fortran compiler documentation to convince myself that matching alignments is not required (indeed, it's beyond our control). As long as C variables corresponding to Fortran common blocks are not initialized in the global scope, the linker will "figure it out" and adjust the alignment to whatever is required (i.e., the greater of the alignments). Specifically (to counter comments that no longer exist in the OMPI code base but still exist in the LAM code base): - there is no need to make attempts to specially align C variables corresponding to Fortran common blocks - the types and sizes of C variables corresponding to Fortran common blocks should match, but do not need to be on any particular alignment Finally, as a side effect of this effort, I found a bunch of inconsistencies with the intent of status/array_of_statuses parameters. For all the functions that I modified they should be "out" (not inout). This commit was SVN r11057.
2006-07-31 19:07:09 +04:00
! MPI_STATUSES_IGNORE has similar rationale to MPI_ARGVS_NULL.
double precision MPI_STATUSES_IGNORE
Bring over changes from the /tmp/fortran-stuff series: - Make the F90 bindings compile and link properly with gfortran 4.0, 4.1, Intel 9.0, PGI 6.1, Sun (don't know version offhand -- the most current as of this writing, I think), and NAG 5.2, although some have limitations (e.g., NAG can't seem to handle the medium and large sizes) - Building the F90 "small" module size is now the default, even for developers - Split up mpif.h into multiple files because parts of it were toxic to the F90 bindings - Properly specify unsized/unshaped arrays to make the bindings work on all known compilers - Make ompi_info show Fortran 90 bindings size - XML somewhat lags the generated scripts as of this commit, but functionality was my main goal -- the XML can be updated later (if at all). This commit was SVN r10118.
2006-05-30 18:37:41 +04:00
common/mpi_fortran_bottom/MPI_BOTTOM
common/mpi_fortran_in_place/MPI_IN_PLACE
common/mpi_fortran_argv_null/MPI_ARGV_NULL
common/mpi_fortran_argvs_null/MPI_ARGVS_NULL
common/mpi_fortran_errcodes_ignore/MPI_ERRCODES_IGNORE
common/mpi_fortran_status_ignore/MPI_STATUS_IGNORE
common/mpi_fortran_statuses_ignore/MPI_STATUSES_IGNORE
!
! NULL "handles" (indices)
!
integer MPI_GROUP_NULL, MPI_COMM_NULL, MPI_DATATYPE_NULL
integer MPI_REQUEST_NULL, MPI_OP_NULL, MPI_ERRHANDLER_NULL
integer MPI_INFO_NULL, MPI_WIN_NULL
parameter (MPI_GROUP_NULL=0)
parameter (MPI_COMM_NULL=2)
parameter (MPI_DATATYPE_NULL=0)
parameter (MPI_REQUEST_NULL=0)
parameter (MPI_OP_NULL=0)
parameter (MPI_ERRHANDLER_NULL=0)
parameter (MPI_INFO_NULL=0)
parameter (MPI_WIN_NULL=0)
!
! MPI_Init_thread constants
!
integer MPI_THREAD_SINGLE, MPI_THREAD_FUNNELED
integer MPI_THREAD_SERIALIZED, MPI_THREAD_MULTIPLE
parameter (MPI_THREAD_SINGLE=0)
parameter (MPI_THREAD_FUNNELED=1)
parameter (MPI_THREAD_SERIALIZED=2)
parameter (MPI_THREAD_MULTIPLE=3)
!
! error classes
!
integer MPI_SUCCESS
integer MPI_ERR_BUFFER
integer MPI_ERR_COUNT
integer MPI_ERR_TYPE
integer MPI_ERR_TAG
integer MPI_ERR_COMM
integer MPI_ERR_RANK
integer MPI_ERR_REQUEST
integer MPI_ERR_ROOT
integer MPI_ERR_GROUP
integer MPI_ERR_OP
integer MPI_ERR_TOPOLOGY
integer MPI_ERR_DIMS
integer MPI_ERR_ARG
integer MPI_ERR_UNKNOWN
integer MPI_ERR_TRUNCATE
integer MPI_ERR_OTHER
integer MPI_ERR_INTERN
integer MPI_ERR_IN_STATUS
integer MPI_ERR_PENDING
integer MPI_ERR_ACCESS
integer MPI_ERR_AMODE
integer MPI_ERR_ASSERT
integer MPI_ERR_BAD_FILE
integer MPI_ERR_BASE
integer MPI_ERR_CONVERSION
integer MPI_ERR_DISP
integer MPI_ERR_DUP_DATAREP
integer MPI_ERR_FILE_EXISTS
integer MPI_ERR_FILE_IN_USE
integer MPI_ERR_FILE
integer MPI_ERR_INFO_KEY
integer MPI_ERR_INFO_NOKEY
integer MPI_ERR_INFO_VALUE
integer MPI_ERR_INFO
integer MPI_ERR_IO
integer MPI_ERR_KEYVAL
integer MPI_ERR_LOCKTYPE
integer MPI_ERR_NAME
integer MPI_ERR_NO_MEM
integer MPI_ERR_NOT_SAME
integer MPI_ERR_NO_SPACE
integer MPI_ERR_NO_SUCH_FILE
integer MPI_ERR_PORT
integer MPI_ERR_QUOTA
integer MPI_ERR_READ_ONLY
integer MPI_ERR_RMA_CONFLICT
integer MPI_ERR_RMA_SYNC
integer MPI_ERR_SERVICE
integer MPI_ERR_SIZE
integer MPI_ERR_SPAWN
integer MPI_ERR_UNSUPPORTED_DATAREP
integer MPI_ERR_UNSUPPORTED_OPERATION
integer MPI_ERR_WIN
integer MPI_ERR_SYSRESOURCE
integer MPI_ERR_LASTCODE
parameter( MPI_SUCCESS = 0)
parameter( MPI_ERR_BUFFER = 1)
parameter( MPI_ERR_COUNT = 2)
parameter( MPI_ERR_TYPE = 3)
parameter( MPI_ERR_TAG = 4)
parameter( MPI_ERR_COMM = 5)
parameter( MPI_ERR_RANK = 6)
parameter( MPI_ERR_REQUEST = 7)
parameter( MPI_ERR_ROOT = 8)
parameter( MPI_ERR_GROUP = 9)
parameter( MPI_ERR_OP = 10)
parameter( MPI_ERR_TOPOLOGY = 11)
parameter( MPI_ERR_DIMS = 12)
parameter( MPI_ERR_ARG = 13)
parameter( MPI_ERR_UNKNOWN = 14)
parameter( MPI_ERR_TRUNCATE = 15)
parameter( MPI_ERR_OTHER = 16)
parameter( MPI_ERR_INTERN = 17)
parameter( MPI_ERR_IN_STATUS = 18)
parameter( MPI_ERR_PENDING = 19)
parameter( MPI_ERR_ACCESS = 20)
parameter( MPI_ERR_AMODE = 21)
parameter( MPI_ERR_ASSERT = 22)
parameter( MPI_ERR_BAD_FILE = 23)
parameter( MPI_ERR_BASE = 24)
parameter( MPI_ERR_CONVERSION = 25)
parameter( MPI_ERR_DISP = 26)
parameter( MPI_ERR_DUP_DATAREP = 27)
parameter( MPI_ERR_FILE_EXISTS = 28)
parameter( MPI_ERR_FILE_IN_USE = 29)
parameter( MPI_ERR_FILE = 30)
parameter( MPI_ERR_INFO_KEY = 31)
parameter( MPI_ERR_INFO_NOKEY = 32)
parameter( MPI_ERR_INFO_VALUE = 33)
parameter( MPI_ERR_INFO = 34)
parameter( MPI_ERR_IO = 35)
parameter( MPI_ERR_KEYVAL = 36)
parameter( MPI_ERR_LOCKTYPE = 37)
parameter( MPI_ERR_NAME = 38)
parameter( MPI_ERR_NO_MEM = 39)
parameter( MPI_ERR_NOT_SAME = 40)
parameter( MPI_ERR_NO_SPACE = 41)
parameter( MPI_ERR_NO_SUCH_FILE = 42)
parameter( MPI_ERR_PORT = 43)
parameter( MPI_ERR_QUOTA = 44)
parameter( MPI_ERR_READ_ONLY = 45)
parameter( MPI_ERR_RMA_CONFLICT = 46)
parameter( MPI_ERR_RMA_SYNC = 47)
parameter( MPI_ERR_SERVICE = 48)
parameter( MPI_ERR_SIZE = 49)
parameter( MPI_ERR_SPAWN = 50)
parameter( MPI_ERR_UNSUPPORTED_DATAREP = 51)
parameter( MPI_ERR_UNSUPPORTED_OPERATION= 52)
parameter( MPI_ERR_WIN = 53)
parameter( MPI_ERR_SYSRESOURCE = -2)
parameter( MPI_ERR_LASTCODE = 54)
!
! comparison results
!
integer MPI_IDENT, MPI_CONGRUENT, MPI_SIMILAR, MPI_UNEQUAL
parameter (MPI_IDENT=0)
parameter (MPI_CONGRUENT=1)
parameter (MPI_SIMILAR=2)
parameter (MPI_UNEQUAL=3)
!
! datatype combiners
!
integer MPI_COMBINER_NAMED
integer MPI_COMBINER_DUP
integer MPI_COMBINER_CONTIGUOUS
integer MPI_COMBINER_VECTOR
integer MPI_COMBINER_HVECTOR_INTEGER
integer MPI_COMBINER_HVECTOR
integer MPI_COMBINER_INDEXED
integer MPI_COMBINER_HINDEXED_INTEGER
integer MPI_COMBINER_HINDEXED
integer MPI_COMBINER_INDEXED_BLOCK
integer MPI_COMBINER_STRUCT_INTEGER
integer MPI_COMBINER_STRUCT
integer MPI_COMBINER_SUBARRAY
integer MPI_COMBINER_DARRAY
integer MPI_COMBINER_F90_REAL
integer MPI_COMBINER_F90_COMPLEX
integer MPI_COMBINER_F90_INTEGER
integer MPI_COMBINER_RESIZED
parameter (MPI_COMBINER_NAMED=0)
parameter (MPI_COMBINER_DUP=1)
parameter (MPI_COMBINER_CONTIGUOUS=2)
parameter (MPI_COMBINER_VECTOR=3)
parameter (MPI_COMBINER_HVECTOR_INTEGER=4)
parameter (MPI_COMBINER_HVECTOR=5)
parameter (MPI_COMBINER_INDEXED=6)
parameter (MPI_COMBINER_HINDEXED_INTEGER=7)
parameter (MPI_COMBINER_HINDEXED=8)
parameter (MPI_COMBINER_INDEXED_BLOCK=9)
parameter (MPI_COMBINER_STRUCT_INTEGER=10)
parameter (MPI_COMBINER_STRUCT=11)
parameter (MPI_COMBINER_SUBARRAY=12)
parameter (MPI_COMBINER_DARRAY=13)
parameter (MPI_COMBINER_F90_REAL=14)
parameter (MPI_COMBINER_F90_COMPLEX=15)
parameter (MPI_COMBINER_F90_INTEGER=16)
parameter (MPI_COMBINER_RESIZED=17)
!
! lookup table indices
!
integer MPI_COMM_WORLD, MPI_COMM_SELF
integer MPI_GROUP_EMPTY
integer MPI_ERRORS_ARE_FATAL, MPI_ERRORS_RETURN
parameter (MPI_COMM_WORLD=0)
parameter (MPI_COMM_SELF=1)
parameter (MPI_GROUP_EMPTY=1)
parameter (MPI_ERRORS_ARE_FATAL=1)
parameter (MPI_ERRORS_RETURN=2)
integer MPI_BYTE, MPI_PACKED, MPI_UB, MPI_LB
integer MPI_CHARACTER, MPI_LOGICAL
integer MPI_INTEGER, MPI_INTEGER1, MPI_INTEGER2, MPI_INTEGER4
integer MPI_INTEGER8, MPI_INTEGER16
- Add the missing parts: add MPI_REAL2 to the end of the list of Fortran datatypes (mpif-common.h) and the list of registered datatypes: MOOG(REAL2). Configure and Compilation with ia32/gcc just finished, naturally without real2. This commit was SVN r15137.
2007-06-20 00:41:28 +04:00
integer MPI_REAL, MPI_REAL2, MPI_REAL4, MPI_REAL8, MPI_REAL16
Bring over changes from the /tmp/fortran-stuff series: - Make the F90 bindings compile and link properly with gfortran 4.0, 4.1, Intel 9.0, PGI 6.1, Sun (don't know version offhand -- the most current as of this writing, I think), and NAG 5.2, although some have limitations (e.g., NAG can't seem to handle the medium and large sizes) - Building the F90 "small" module size is now the default, even for developers - Split up mpif.h into multiple files because parts of it were toxic to the F90 bindings - Properly specify unsized/unshaped arrays to make the bindings work on all known compilers - Make ompi_info show Fortran 90 bindings size - XML somewhat lags the generated scripts as of this commit, but functionality was my main goal -- the XML can be updated later (if at all). This commit was SVN r10118.
2006-05-30 18:37:41 +04:00
integer MPI_DOUBLE_PRECISION
integer MPI_COMPLEX, MPI_COMPLEX8, MPI_COMPLEX16, MPI_COMPLEX32
integer MPI_DOUBLE_COMPLEX
integer MPI_2REAL, MPI_2DOUBLE_PRECISION, MPI_2INTEGER
Add external support for MPI_2COMPLEX and MPI_2DOUBLE_COMPLEX. The machinery was already there; we were just missing the INTEGER's for them in mpi.f and the #define's for them in mpi.h. This commit was SVN r13432.
2007-02-01 22:37:38 +03:00
integer MPI_2COMPLEX, MPI_2DOUBLE_COMPLEX
- Add the missing parts: add MPI_REAL2 to the end of the list of Fortran datatypes (mpif-common.h) and the list of registered datatypes: MOOG(REAL2). Configure and Compilation with ia32/gcc just finished, naturally without real2. This commit was SVN r15137.
2007-06-20 00:41:28 +04:00
!
! Do NOT change the order of these parameters
!
Bring over changes from the /tmp/fortran-stuff series: - Make the F90 bindings compile and link properly with gfortran 4.0, 4.1, Intel 9.0, PGI 6.1, Sun (don't know version offhand -- the most current as of this writing, I think), and NAG 5.2, although some have limitations (e.g., NAG can't seem to handle the medium and large sizes) - Building the F90 "small" module size is now the default, even for developers - Split up mpif.h into multiple files because parts of it were toxic to the F90 bindings - Properly specify unsized/unshaped arrays to make the bindings work on all known compilers - Make ompi_info show Fortran 90 bindings size - XML somewhat lags the generated scripts as of this commit, but functionality was my main goal -- the XML can be updated later (if at all). This commit was SVN r10118.
2006-05-30 18:37:41 +04:00
parameter (MPI_BYTE=1)
parameter (MPI_PACKED=2)
parameter (MPI_UB=3)
parameter (MPI_LB=4)
parameter (MPI_CHARACTER=5)
parameter (MPI_LOGICAL=6)
parameter (MPI_INTEGER=7)
parameter (MPI_INTEGER1=8)
parameter (MPI_INTEGER2=9)
parameter (MPI_INTEGER4=10)
parameter (MPI_INTEGER8=11)
parameter (MPI_INTEGER16=12)
parameter (MPI_REAL=13)
parameter (MPI_REAL4=14)
parameter (MPI_REAL8=15)
parameter (MPI_REAL16=16)
parameter (MPI_DOUBLE_PRECISION=17)
parameter (MPI_COMPLEX=18)
parameter (MPI_COMPLEX8=19)
parameter (MPI_COMPLEX16=20)
parameter (MPI_COMPLEX32=21)
parameter (MPI_DOUBLE_COMPLEX=22)
parameter (MPI_2REAL=23)
parameter (MPI_2DOUBLE_PRECISION=24)
parameter (MPI_2INTEGER=25)
Add external support for MPI_2COMPLEX and MPI_2DOUBLE_COMPLEX. The machinery was already there; we were just missing the INTEGER's for them in mpi.f and the #define's for them in mpi.h. This commit was SVN r13432.
2007-02-01 22:37:38 +03:00
parameter (MPI_2COMPLEX=26)
parameter (MPI_2DOUBLE_COMPLEX=27)
- Add the missing parts: add MPI_REAL2 to the end of the list of Fortran datatypes (mpif-common.h) and the list of registered datatypes: MOOG(REAL2). Configure and Compilation with ia32/gcc just finished, naturally without real2. This commit was SVN r15137.
2007-06-20 00:41:28 +04:00
parameter (MPI_REAL2=28)
Bring over changes from the /tmp/fortran-stuff series: - Make the F90 bindings compile and link properly with gfortran 4.0, 4.1, Intel 9.0, PGI 6.1, Sun (don't know version offhand -- the most current as of this writing, I think), and NAG 5.2, although some have limitations (e.g., NAG can't seem to handle the medium and large sizes) - Building the F90 "small" module size is now the default, even for developers - Split up mpif.h into multiple files because parts of it were toxic to the F90 bindings - Properly specify unsized/unshaped arrays to make the bindings work on all known compilers - Make ompi_info show Fortran 90 bindings size - XML somewhat lags the generated scripts as of this commit, but functionality was my main goal -- the XML can be updated later (if at all). This commit was SVN r10118.
2006-05-30 18:37:41 +04:00
integer MPI_MAX, MPI_MIN, MPI_SUM, MPI_PROD, MPI_LAND
integer MPI_BAND, MPI_LOR, MPI_BOR, MPI_LXOR, MPI_BXOR
integer MPI_MAXLOC, MPI_MINLOC, MPI_REPLACE
parameter (MPI_MAX=1)
parameter (MPI_MIN=2)
parameter (MPI_SUM=3)
parameter (MPI_PROD=4)
parameter (MPI_LAND=5)
parameter (MPI_BAND=6)
parameter (MPI_LOR=7)
parameter (MPI_BOR=8)
parameter (MPI_LXOR=9)
parameter (MPI_BXOR=10)
parameter (MPI_MAXLOC=11)
parameter (MPI_MINLOC=12)
parameter (MPI_REPLACE=13)
Копировать постоянную ссылку