scnc
/
quantum-espresso
mirror of https://gitlab.com/QEF/q-e.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity
quantum-espresso/Modules/check_stop.f90

168 lines
4.8 KiB
Fortran
Raw Normal View History

- Almost all neb routines moved to Modules - New module check_stop used by all codes to check for exit conditions ( maximum time or EXIT file ) git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@718 c92efa57-630b-4861-b058-cf58834340f0
2004-03-15 08:08:26 +08:00
!
Added signal processing by RdM. Disabled by default, enabled by option -D__TRAP_SIGUSR1 or using "./configure --enable-signals" git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@6370 c92efa57-630b-4861-b058-cf58834340f0
2010-02-09 03:25:34 +08:00
! Copyright (C) 2002-2010 Quantum ESPRESSO group
- Almost all neb routines moved to Modules - New module check_stop used by all codes to check for exit conditions ( maximum time or EXIT file ) git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@718 c92efa57-630b-4861-b058-cf58834340f0
2004-03-15 08:08:26 +08:00
! This file is distributed under the terms of the
! GNU General Public License. See the file `License'
! in the root directory of the present distribution,
! or http://www.gnu.org/copyleft/gpl.txt .
!
1) An experimental parallelization of NEB images has been implemented. This required a deep modification of the parallelism in PWscf: there are two new communicators (intra_image_comm and inter_image_comm) and the existing "pool" communicators (intra_pool_comm and inter_pool_comm) are now vectors of length given by the number of parallel images. #ifdef __PARA is no longer needed because all "parallel" variables are always initialized for a serial run and all parallel routines are, in the case of a serial run, dummy routines. The wrappers to MPI routines used only by PWscf are in the PW/para.f90 file. The others (mp_***) are in the Modules/mp.f90. All explicit referencies to mpif.h should be replaced by an "USE parallel_include" (in a serial run parallel_include is simply a dummy module). 2) The extrapolation of both potential and wavefunctions has been rewritten in order to be smarter than before: on the basis of the required extrapolation order, on the basis of the history and on the basis of which files are really present on the disk, the algorithm chooses the extrapolation order. All the algorithms in which ions are moved can use the extrapolation. These are both unstable features: I need the help of everybody to test them. C.S. git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@742 c92efa57-630b-4861-b058-cf58834340f0
2004-03-24 17:36:50 +08:00
!
! ... This module contains functions to check if the code should
When reading UPF potentials it is better to check for end-of-file rather than saving a useless value of 'iostat' that is never verified git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@2417 c92efa57-630b-4861-b058-cf58834340f0
2005-11-03 23:59:28 +08:00
! ... be smoothly stopped.
More installation cleanup (or maybe messup): "configure" should now recognize the correct length of fortran integer variables containing C pointers, writing it into file include/fft_defs.h . Side effects not unlikely, please check and correct if you can (and dare to). Version number moved to 3.2cvs git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@3407 c92efa57-630b-4861-b058-cf58834340f0
2006-09-21 23:23:47 +08:00
! ... In particular the function check_stop_now returns .TRUE. if
Fix for check_stop (N. Nemec) git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@6376 c92efa57-630b-4861-b058-cf58834340f0
2010-02-09 23:16:28 +08:00
! ... either the user has created a given file or if the
1) An experimental parallelization of NEB images has been implemented. This required a deep modification of the parallelism in PWscf: there are two new communicators (intra_image_comm and inter_image_comm) and the existing "pool" communicators (intra_pool_comm and inter_pool_comm) are now vectors of length given by the number of parallel images. #ifdef __PARA is no longer needed because all "parallel" variables are always initialized for a serial run and all parallel routines are, in the case of a serial run, dummy routines. The wrappers to MPI routines used only by PWscf are in the PW/para.f90 file. The others (mp_***) are in the Modules/mp.f90. All explicit referencies to mpif.h should be replaced by an "USE parallel_include" (in a serial run parallel_include is simply a dummy module). 2) The extrapolation of both potential and wavefunctions has been rewritten in order to be smarter than before: on the basis of the required extrapolation order, on the basis of the history and on the basis of which files are really present on the disk, the algorithm chooses the extrapolation order. All the algorithms in which ions are moved can use the extrapolation. These are both unstable features: I need the help of everybody to test them. C.S. git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@742 c92efa57-630b-4861-b058-cf58834340f0
2004-03-24 17:36:50 +08:00
! ... elapsed time is larger than max_seconds
!
- Almost all neb routines moved to Modules - New module check_stop used by all codes to check for exit conditions ( maximum time or EXIT file ) git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@718 c92efa57-630b-4861-b058-cf58834340f0
2004-03-15 08:08:26 +08:00
!------------------------------------------------------------------------------!
1) An experimental parallelization of NEB images has been implemented. This required a deep modification of the parallelism in PWscf: there are two new communicators (intra_image_comm and inter_image_comm) and the existing "pool" communicators (intra_pool_comm and inter_pool_comm) are now vectors of length given by the number of parallel images. #ifdef __PARA is no longer needed because all "parallel" variables are always initialized for a serial run and all parallel routines are, in the case of a serial run, dummy routines. The wrappers to MPI routines used only by PWscf are in the PW/para.f90 file. The others (mp_***) are in the Modules/mp.f90. All explicit referencies to mpif.h should be replaced by an "USE parallel_include" (in a serial run parallel_include is simply a dummy module). 2) The extrapolation of both potential and wavefunctions has been rewritten in order to be smarter than before: on the basis of the required extrapolation order, on the basis of the history and on the basis of which files are really present on the disk, the algorithm chooses the extrapolation order. All the algorithms in which ions are moved can use the extrapolation. These are both unstable features: I need the help of everybody to test them. C.S. git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@742 c92efa57-630b-4861-b058-cf58834340f0
2004-03-24 17:36:50 +08:00
MODULE check_stop
- Almost all neb routines moved to Modules - New module check_stop used by all codes to check for exit conditions ( maximum time or EXIT file ) git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@718 c92efa57-630b-4861-b058-cf58834340f0
2004-03-15 08:08:26 +08:00
!------------------------------------------------------------------------------!
Fix for check_stop (N. Nemec) git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@6376 c92efa57-630b-4861-b058-cf58834340f0
2010-02-09 23:16:28 +08:00
!
1) An experimental parallelization of NEB images has been implemented. This required a deep modification of the parallelism in PWscf: there are two new communicators (intra_image_comm and inter_image_comm) and the existing "pool" communicators (intra_pool_comm and inter_pool_comm) are now vectors of length given by the number of parallel images. #ifdef __PARA is no longer needed because all "parallel" variables are always initialized for a serial run and all parallel routines are, in the case of a serial run, dummy routines. The wrappers to MPI routines used only by PWscf are in the PW/para.f90 file. The others (mp_***) are in the Modules/mp.f90. All explicit referencies to mpif.h should be replaced by an "USE parallel_include" (in a serial run parallel_include is simply a dummy module). 2) The extrapolation of both potential and wavefunctions has been rewritten in order to be smarter than before: on the basis of the required extrapolation order, on the basis of the history and on the basis of which files are really present on the disk, the algorithm chooses the extrapolation order. All the algorithms in which ions are moved can use the extrapolation. These are both unstable features: I need the help of everybody to test them. C.S. git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@742 c92efa57-630b-4861-b058-cf58834340f0
2004-03-24 17:36:50 +08:00
USE kinds
!
IMPLICIT NONE
!
SAVE
!
All occurrences of d0, d+N, d-N, replaced with fortran-90 style _DP. Please do not re-introduce old-style syntax. A uniform syntax allows if the need arises to change the precision in less than no time. Courtesy of Pascal Thibeaudeau. Modules/ directory only. git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@3964 c92efa57-630b-4861-b058-cf58834340f0
2007-06-12 01:13:15 +08:00
REAL(DP) :: max_seconds = 1.E+7_DP
Some cleanup: the info message about reading/writing the restart file now indicates the restart directory (the only file users are aware of) and not the xml descriptor. C.S. git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@2432 c92efa57-630b-4861-b058-cf58834340f0
2005-11-08 02:56:23 +08:00
!
1) An experimental parallelization of NEB images has been implemented. This required a deep modification of the parallelism in PWscf: there are two new communicators (intra_image_comm and inter_image_comm) and the existing "pool" communicators (intra_pool_comm and inter_pool_comm) are now vectors of length given by the number of parallel images. #ifdef __PARA is no longer needed because all "parallel" variables are always initialized for a serial run and all parallel routines are, in the case of a serial run, dummy routines. The wrappers to MPI routines used only by PWscf are in the PW/para.f90 file. The others (mp_***) are in the Modules/mp.f90. All explicit referencies to mpif.h should be replaced by an "USE parallel_include" (in a serial run parallel_include is simply a dummy module). 2) The extrapolation of both potential and wavefunctions has been rewritten in order to be smarter than before: on the basis of the required extrapolation order, on the basis of the history and on the basis of which files are really present on the disk, the algorithm chooses the extrapolation order. All the algorithms in which ions are moved can use the extrapolation. These are both unstable features: I need the help of everybody to test them. C.S. git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@742 c92efa57-630b-4861-b058-cf58834340f0
2004-03-24 17:36:50 +08:00
LOGICAL, PRIVATE :: tinit = .FALSE.
!
- fixed a configuration problem with pgi compiler and preprocessor - elapsed_seconds substituted by cclock - fortran version of scnds substituted by c version. now time information is taken only through: cclock for the real wall clock time scnds for cpu time of the running process defined in file clib/cptimer.c git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@3400 c92efa57-630b-4861-b058-cf58834340f0
2006-09-20 01:00:18 +08:00
REAL(DP) :: init_second
1) An experimental parallelization of NEB images has been implemented. This required a deep modification of the parallelism in PWscf: there are two new communicators (intra_image_comm and inter_image_comm) and the existing "pool" communicators (intra_pool_comm and inter_pool_comm) are now vectors of length given by the number of parallel images. #ifdef __PARA is no longer needed because all "parallel" variables are always initialized for a serial run and all parallel routines are, in the case of a serial run, dummy routines. The wrappers to MPI routines used only by PWscf are in the PW/para.f90 file. The others (mp_***) are in the Modules/mp.f90. All explicit referencies to mpif.h should be replaced by an "USE parallel_include" (in a serial run parallel_include is simply a dummy module). 2) The extrapolation of both potential and wavefunctions has been rewritten in order to be smarter than before: on the basis of the required extrapolation order, on the basis of the history and on the basis of which files are really present on the disk, the algorithm chooses the extrapolation order. All the algorithms in which ions are moved can use the extrapolation. These are both unstable features: I need the help of everybody to test them. C.S. git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@742 c92efa57-630b-4861-b058-cf58834340f0
2004-03-24 17:36:50 +08:00
!
- Almost all neb routines moved to Modules - New module check_stop used by all codes to check for exit conditions ( maximum time or EXIT file ) git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@718 c92efa57-630b-4861-b058-cf58834340f0
2004-03-15 08:08:26 +08:00
CONTAINS
1) An experimental parallelization of NEB images has been implemented. This required a deep modification of the parallelism in PWscf: there are two new communicators (intra_image_comm and inter_image_comm) and the existing "pool" communicators (intra_pool_comm and inter_pool_comm) are now vectors of length given by the number of parallel images. #ifdef __PARA is no longer needed because all "parallel" variables are always initialized for a serial run and all parallel routines are, in the case of a serial run, dummy routines. The wrappers to MPI routines used only by PWscf are in the PW/para.f90 file. The others (mp_***) are in the Modules/mp.f90. All explicit referencies to mpif.h should be replaced by an "USE parallel_include" (in a serial run parallel_include is simply a dummy module). 2) The extrapolation of both potential and wavefunctions has been rewritten in order to be smarter than before: on the basis of the required extrapolation order, on the basis of the history and on the basis of which files are really present on the disk, the algorithm chooses the extrapolation order. All the algorithms in which ions are moved can use the extrapolation. These are both unstable features: I need the help of everybody to test them. C.S. git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@742 c92efa57-630b-4861-b058-cf58834340f0
2004-03-24 17:36:50 +08:00
!
! ... internal procedures
!
!-----------------------------------------------------------------------
Stop by user request (prefix.EXIT) was not working for CP. Fixed and reorganised. C.S. git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@2602 c92efa57-630b-4861-b058-cf58834340f0
2005-12-18 01:40:37 +08:00
SUBROUTINE check_stop_init()
1) An experimental parallelization of NEB images has been implemented. This required a deep modification of the parallelism in PWscf: there are two new communicators (intra_image_comm and inter_image_comm) and the existing "pool" communicators (intra_pool_comm and inter_pool_comm) are now vectors of length given by the number of parallel images. #ifdef __PARA is no longer needed because all "parallel" variables are always initialized for a serial run and all parallel routines are, in the case of a serial run, dummy routines. The wrappers to MPI routines used only by PWscf are in the PW/para.f90 file. The others (mp_***) are in the Modules/mp.f90. All explicit referencies to mpif.h should be replaced by an "USE parallel_include" (in a serial run parallel_include is simply a dummy module). 2) The extrapolation of both potential and wavefunctions has been rewritten in order to be smarter than before: on the basis of the required extrapolation order, on the basis of the history and on the basis of which files are really present on the disk, the algorithm chooses the extrapolation order. All the algorithms in which ions are moved can use the extrapolation. These are both unstable features: I need the help of everybody to test them. C.S. git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@742 c92efa57-630b-4861-b058-cf58834340f0
2004-03-24 17:36:50 +08:00
!-----------------------------------------------------------------------
!
Stop by user request (prefix.EXIT) was not working for CP. Fixed and reorganised. C.S. git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@2602 c92efa57-630b-4861-b058-cf58834340f0
2005-12-18 01:40:37 +08:00
USE input_parameters, ONLY : max_seconds_ => max_seconds
Added signal processing by RdM. Disabled by default, enabled by option -D__TRAP_SIGUSR1 or using "./configure --enable-signals" git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@6370 c92efa57-630b-4861-b058-cf58834340f0
2010-02-09 03:25:34 +08:00
USE io_global, ONLY : stdout
More minor I/O cleanup git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@6832 c92efa57-630b-4861-b058-cf58834340f0
2010-06-11 22:43:50 +08:00
USE io_files, ONLY : prefix, exit_file
Fix for check_stop (N. Nemec) git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@6376 c92efa57-630b-4861-b058-cf58834340f0
2010-02-09 23:16:28 +08:00
#if defined __TRAP_SIGUSR1
Added signal processing by RdM. Disabled by default, enabled by option -D__TRAP_SIGUSR1 or using "./configure --enable-signals" git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@6370 c92efa57-630b-4861-b058-cf58834340f0
2010-02-09 03:25:34 +08:00
USE set_signal, ONLY : signal_trap_init
Fix for check_stop (N. Nemec) git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@6376 c92efa57-630b-4861-b058-cf58834340f0
2010-02-09 23:16:28 +08:00
#endif
Added signal processing by RdM. Disabled by default, enabled by option -D__TRAP_SIGUSR1 or using "./configure --enable-signals" git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@6370 c92efa57-630b-4861-b058-cf58834340f0
2010-02-09 03:25:34 +08:00
1) An experimental parallelization of NEB images has been implemented. This required a deep modification of the parallelism in PWscf: there are two new communicators (intra_image_comm and inter_image_comm) and the existing "pool" communicators (intra_pool_comm and inter_pool_comm) are now vectors of length given by the number of parallel images. #ifdef __PARA is no longer needed because all "parallel" variables are always initialized for a serial run and all parallel routines are, in the case of a serial run, dummy routines. The wrappers to MPI routines used only by PWscf are in the PW/para.f90 file. The others (mp_***) are in the Modules/mp.f90. All explicit referencies to mpif.h should be replaced by an "USE parallel_include" (in a serial run parallel_include is simply a dummy module). 2) The extrapolation of both potential and wavefunctions has been rewritten in order to be smarter than before: on the basis of the required extrapolation order, on the basis of the history and on the basis of which files are really present on the disk, the algorithm chooses the extrapolation order. All the algorithms in which ions are moved can use the extrapolation. These are both unstable features: I need the help of everybody to test them. C.S. git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@742 c92efa57-630b-4861-b058-cf58834340f0
2004-03-24 17:36:50 +08:00
!
IMPLICIT NONE
!
- fixed a configuration problem with pgi compiler and preprocessor - elapsed_seconds substituted by cclock - fortran version of scnds substituted by c version. now time information is taken only through: cclock for the real wall clock time scnds for cpu time of the running process defined in file clib/cptimer.c git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@3400 c92efa57-630b-4861-b058-cf58834340f0
2006-09-20 01:00:18 +08:00
REAL(DP), EXTERNAL :: cclock
1) An experimental parallelization of NEB images has been implemented. This required a deep modification of the parallelism in PWscf: there are two new communicators (intra_image_comm and inter_image_comm) and the existing "pool" communicators (intra_pool_comm and inter_pool_comm) are now vectors of length given by the number of parallel images. #ifdef __PARA is no longer needed because all "parallel" variables are always initialized for a serial run and all parallel routines are, in the case of a serial run, dummy routines. The wrappers to MPI routines used only by PWscf are in the PW/para.f90 file. The others (mp_***) are in the Modules/mp.f90. All explicit referencies to mpif.h should be replaced by an "USE parallel_include" (in a serial run parallel_include is simply a dummy module). 2) The extrapolation of both potential and wavefunctions has been rewritten in order to be smarter than before: on the basis of the required extrapolation order, on the basis of the history and on the basis of which files are really present on the disk, the algorithm chooses the extrapolation order. All the algorithms in which ions are moved can use the extrapolation. These are both unstable features: I need the help of everybody to test them. C.S. git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@742 c92efa57-630b-4861-b058-cf58834340f0
2004-03-24 17:36:50 +08:00
!
Some cleanup: the info message about reading/writing the restart file now indicates the restart directory (the only file users are aware of) and not the xml descriptor. C.S. git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@2432 c92efa57-630b-4861-b058-cf58834340f0
2005-11-08 02:56:23 +08:00
IF ( tinit ) &
WRITE( UNIT = stdout, &
FMT = '(/,5X,"WARNING: check_stop already initialized")' )
1) An experimental parallelization of NEB images has been implemented. This required a deep modification of the parallelism in PWscf: there are two new communicators (intra_image_comm and inter_image_comm) and the existing "pool" communicators (intra_pool_comm and inter_pool_comm) are now vectors of length given by the number of parallel images. #ifdef __PARA is no longer needed because all "parallel" variables are always initialized for a serial run and all parallel routines are, in the case of a serial run, dummy routines. The wrappers to MPI routines used only by PWscf are in the PW/para.f90 file. The others (mp_***) are in the Modules/mp.f90. All explicit referencies to mpif.h should be replaced by an "USE parallel_include" (in a serial run parallel_include is simply a dummy module). 2) The extrapolation of both potential and wavefunctions has been rewritten in order to be smarter than before: on the basis of the required extrapolation order, on the basis of the history and on the basis of which files are really present on the disk, the algorithm chooses the extrapolation order. All the algorithms in which ions are moved can use the extrapolation. These are both unstable features: I need the help of everybody to test them. C.S. git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@742 c92efa57-630b-4861-b058-cf58834340f0
2004-03-24 17:36:50 +08:00
!
Stop by user request (prefix.EXIT) was not working for CP. Fixed and reorganised. C.S. git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@2602 c92efa57-630b-4861-b058-cf58834340f0
2005-12-18 01:40:37 +08:00
! ... the exit_file name is set here
!
exit_file = TRIM( prefix ) // '.EXIT'
!
All occurrences of d0, d+N, d-N, replaced with fortran-90 style _DP. Please do not re-introduce old-style syntax. A uniform syntax allows if the need arises to change the precision in less than no time. Courtesy of Pascal Thibeaudeau. Modules/ directory only. git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@3964 c92efa57-630b-4861-b058-cf58834340f0
2007-06-12 01:13:15 +08:00
IF ( max_seconds_ > 0.0_DP ) max_seconds = max_seconds_
1) An experimental parallelization of NEB images has been implemented. This required a deep modification of the parallelism in PWscf: there are two new communicators (intra_image_comm and inter_image_comm) and the existing "pool" communicators (intra_pool_comm and inter_pool_comm) are now vectors of length given by the number of parallel images. #ifdef __PARA is no longer needed because all "parallel" variables are always initialized for a serial run and all parallel routines are, in the case of a serial run, dummy routines. The wrappers to MPI routines used only by PWscf are in the PW/para.f90 file. The others (mp_***) are in the Modules/mp.f90. All explicit referencies to mpif.h should be replaced by an "USE parallel_include" (in a serial run parallel_include is simply a dummy module). 2) The extrapolation of both potential and wavefunctions has been rewritten in order to be smarter than before: on the basis of the required extrapolation order, on the basis of the history and on the basis of which files are really present on the disk, the algorithm chooses the extrapolation order. All the algorithms in which ions are moved can use the extrapolation. These are both unstable features: I need the help of everybody to test them. C.S. git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@742 c92efa57-630b-4861-b058-cf58834340f0
2004-03-24 17:36:50 +08:00
!
- fixed a configuration problem with pgi compiler and preprocessor - elapsed_seconds substituted by cclock - fortran version of scnds substituted by c version. now time information is taken only through: cclock for the real wall clock time scnds for cpu time of the running process defined in file clib/cptimer.c git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@3400 c92efa57-630b-4861-b058-cf58834340f0
2006-09-20 01:00:18 +08:00
init_second = cclock()
1) An experimental parallelization of NEB images has been implemented. This required a deep modification of the parallelism in PWscf: there are two new communicators (intra_image_comm and inter_image_comm) and the existing "pool" communicators (intra_pool_comm and inter_pool_comm) are now vectors of length given by the number of parallel images. #ifdef __PARA is no longer needed because all "parallel" variables are always initialized for a serial run and all parallel routines are, in the case of a serial run, dummy routines. The wrappers to MPI routines used only by PWscf are in the PW/para.f90 file. The others (mp_***) are in the Modules/mp.f90. All explicit referencies to mpif.h should be replaced by an "USE parallel_include" (in a serial run parallel_include is simply a dummy module). 2) The extrapolation of both potential and wavefunctions has been rewritten in order to be smarter than before: on the basis of the required extrapolation order, on the basis of the history and on the basis of which files are really present on the disk, the algorithm chooses the extrapolation order. All the algorithms in which ions are moved can use the extrapolation. These are both unstable features: I need the help of everybody to test them. C.S. git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@742 c92efa57-630b-4861-b058-cf58834340f0
2004-03-24 17:36:50 +08:00
tinit = .TRUE.
!
Fix for check_stop (N. Nemec) git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@6376 c92efa57-630b-4861-b058-cf58834340f0
2010-02-09 23:16:28 +08:00
#if defined __TRAP_SIGUSR1
Added signal processing by RdM. Disabled by default, enabled by option -D__TRAP_SIGUSR1 or using "./configure --enable-signals" git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@6370 c92efa57-630b-4861-b058-cf58834340f0
2010-02-09 03:25:34 +08:00
CALL signal_trap_init ( )
Fix for check_stop (N. Nemec) git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@6376 c92efa57-630b-4861-b058-cf58834340f0
2010-02-09 23:16:28 +08:00
#endif
Added signal processing by RdM. Disabled by default, enabled by option -D__TRAP_SIGUSR1 or using "./configure --enable-signals" git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@6370 c92efa57-630b-4861-b058-cf58834340f0
2010-02-09 03:25:34 +08:00
!
1) An experimental parallelization of NEB images has been implemented. This required a deep modification of the parallelism in PWscf: there are two new communicators (intra_image_comm and inter_image_comm) and the existing "pool" communicators (intra_pool_comm and inter_pool_comm) are now vectors of length given by the number of parallel images. #ifdef __PARA is no longer needed because all "parallel" variables are always initialized for a serial run and all parallel routines are, in the case of a serial run, dummy routines. The wrappers to MPI routines used only by PWscf are in the PW/para.f90 file. The others (mp_***) are in the Modules/mp.f90. All explicit referencies to mpif.h should be replaced by an "USE parallel_include" (in a serial run parallel_include is simply a dummy module). 2) The extrapolation of both potential and wavefunctions has been rewritten in order to be smarter than before: on the basis of the required extrapolation order, on the basis of the history and on the basis of which files are really present on the disk, the algorithm chooses the extrapolation order. All the algorithms in which ions are moved can use the extrapolation. These are both unstable features: I need the help of everybody to test them. C.S. git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@742 c92efa57-630b-4861-b058-cf58834340f0
2004-03-24 17:36:50 +08:00
RETURN
!
more end subroutine --> end subroutine name [Gerardo] git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@1873 c92efa57-630b-4861-b058-cf58834340f0
2005-05-18 17:38:45 +08:00
END SUBROUTINE check_stop_init
1) An experimental parallelization of NEB images has been implemented. This required a deep modification of the parallelism in PWscf: there are two new communicators (intra_image_comm and inter_image_comm) and the existing "pool" communicators (intra_pool_comm and inter_pool_comm) are now vectors of length given by the number of parallel images. #ifdef __PARA is no longer needed because all "parallel" variables are always initialized for a serial run and all parallel routines are, in the case of a serial run, dummy routines. The wrappers to MPI routines used only by PWscf are in the PW/para.f90 file. The others (mp_***) are in the Modules/mp.f90. All explicit referencies to mpif.h should be replaced by an "USE parallel_include" (in a serial run parallel_include is simply a dummy module). 2) The extrapolation of both potential and wavefunctions has been rewritten in order to be smarter than before: on the basis of the required extrapolation order, on the basis of the history and on the basis of which files are really present on the disk, the algorithm chooses the extrapolation order. All the algorithms in which ions are moved can use the extrapolation. These are both unstable features: I need the help of everybody to test them. C.S. git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@742 c92efa57-630b-4861-b058-cf58834340f0
2004-03-24 17:36:50 +08:00
!
!-----------------------------------------------------------------------
Tests on cpu time and soft exit adaptet to "parallel" NEB. C.S. git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@746 c92efa57-630b-4861-b058-cf58834340f0
2004-03-25 20:09:40 +08:00
FUNCTION check_stop_now( inunit )
1) An experimental parallelization of NEB images has been implemented. This required a deep modification of the parallelism in PWscf: there are two new communicators (intra_image_comm and inter_image_comm) and the existing "pool" communicators (intra_pool_comm and inter_pool_comm) are now vectors of length given by the number of parallel images. #ifdef __PARA is no longer needed because all "parallel" variables are always initialized for a serial run and all parallel routines are, in the case of a serial run, dummy routines. The wrappers to MPI routines used only by PWscf are in the PW/para.f90 file. The others (mp_***) are in the Modules/mp.f90. All explicit referencies to mpif.h should be replaced by an "USE parallel_include" (in a serial run parallel_include is simply a dummy module). 2) The extrapolation of both potential and wavefunctions has been rewritten in order to be smarter than before: on the basis of the required extrapolation order, on the basis of the history and on the basis of which files are really present on the disk, the algorithm chooses the extrapolation order. All the algorithms in which ions are moved can use the extrapolation. These are both unstable features: I need the help of everybody to test them. C.S. git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@742 c92efa57-630b-4861-b058-cf58834340f0
2004-03-24 17:36:50 +08:00
!-----------------------------------------------------------------------
!
Added signal processing by RdM. Disabled by default, enabled by option -D__TRAP_SIGUSR1 or using "./configure --enable-signals" git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@6370 c92efa57-630b-4861-b058-cf58834340f0
2010-02-09 03:25:34 +08:00
USE mp, ONLY : mp_bcast
Fix for check_stop (N. Nemec) git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@6376 c92efa57-630b-4861-b058-cf58834340f0
2010-02-09 23:16:28 +08:00
USE mp_global, ONLY : intra_image_comm
Added signal processing by RdM. Disabled by default, enabled by option -D__TRAP_SIGUSR1 or using "./configure --enable-signals" git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@6370 c92efa57-630b-4861-b058-cf58834340f0
2010-02-09 03:25:34 +08:00
USE io_global, ONLY : ionode, ionode_id, meta_ionode, stdout
More minor I/O cleanup git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@6832 c92efa57-630b-4861-b058-cf58834340f0
2010-06-11 22:43:50 +08:00
USE io_files, ONLY : exit_file, iunexit
Fix for check_stop (N. Nemec) git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@6376 c92efa57-630b-4861-b058-cf58834340f0
2010-02-09 23:16:28 +08:00
#if defined __TRAP_SIGUSR1
Added signal processing by RdM. Disabled by default, enabled by option -D__TRAP_SIGUSR1 or using "./configure --enable-signals" git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@6370 c92efa57-630b-4861-b058-cf58834340f0
2010-02-09 03:25:34 +08:00
USE set_signal, ONLY : signal_detected
Fix for check_stop (N. Nemec) git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@6376 c92efa57-630b-4861-b058-cf58834340f0
2010-02-09 23:16:28 +08:00
#endif
Small changes and comment lines to run CP with nimage>1. Grep for "KNK_nimage" and uncomment the lines underneath. The images will write trajectory files as in "prefix_N.*", the restarts will be "RESTART_N", and the outputs will be "out.N_0". For the first image things are as usual ("_N" is skipped). Now, this won't do anything useful, and will just run non-communicating copies. For useful stuff (path integral, string method, etc) setup some communication between images in cpr.f90 Kostya git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@5084 c92efa57-630b-4861-b058-cf58834340f0
2008-07-29 10:05:18 +08:00
! KNK_image
! USE mp_global, ONLY : mpime, root, world_comm
1) An experimental parallelization of NEB images has been implemented. This required a deep modification of the parallelism in PWscf: there are two new communicators (intra_image_comm and inter_image_comm) and the existing "pool" communicators (intra_pool_comm and inter_pool_comm) are now vectors of length given by the number of parallel images. #ifdef __PARA is no longer needed because all "parallel" variables are always initialized for a serial run and all parallel routines are, in the case of a serial run, dummy routines. The wrappers to MPI routines used only by PWscf are in the PW/para.f90 file. The others (mp_***) are in the Modules/mp.f90. All explicit referencies to mpif.h should be replaced by an "USE parallel_include" (in a serial run parallel_include is simply a dummy module). 2) The extrapolation of both potential and wavefunctions has been rewritten in order to be smarter than before: on the basis of the required extrapolation order, on the basis of the history and on the basis of which files are really present on the disk, the algorithm chooses the extrapolation order. All the algorithms in which ions are moved can use the extrapolation. These are both unstable features: I need the help of everybody to test them. C.S. git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@742 c92efa57-630b-4861-b058-cf58834340f0
2004-03-24 17:36:50 +08:00
!
IMPLICIT NONE
!
Tests on cpu time and soft exit adaptet to "parallel" NEB. C.S. git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@746 c92efa57-630b-4861-b058-cf58834340f0
2004-03-25 20:09:40 +08:00
INTEGER, OPTIONAL, INTENT(IN) :: inunit
The check on stop conditions must be done by ionode(s) and not by meta_ionode, otherwise parallelization on neb images doesn't work anymore. C.S. git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@3244 c92efa57-630b-4861-b058-cf58834340f0
2006-07-06 05:31:35 +08:00
!
INTEGER :: unit
LOGICAL :: check_stop_now, tex
Added signal processing by RdM. Disabled by default, enabled by option -D__TRAP_SIGUSR1 or using "./configure --enable-signals" git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@6370 c92efa57-630b-4861-b058-cf58834340f0
2010-02-09 03:25:34 +08:00
LOGICAL :: signaled
The check on stop conditions must be done by ionode(s) and not by meta_ionode, otherwise parallelization on neb images doesn't work anymore. C.S. git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@3244 c92efa57-630b-4861-b058-cf58834340f0
2006-07-06 05:31:35 +08:00
REAL(DP) :: seconds
- fixed a configuration problem with pgi compiler and preprocessor - elapsed_seconds substituted by cclock - fortran version of scnds substituted by c version. now time information is taken only through: cclock for the real wall clock time scnds for cpu time of the running process defined in file clib/cptimer.c git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@3400 c92efa57-630b-4861-b058-cf58834340f0
2006-09-20 01:00:18 +08:00
REAL(DP), EXTERNAL :: cclock
1) An experimental parallelization of NEB images has been implemented. This required a deep modification of the parallelism in PWscf: there are two new communicators (intra_image_comm and inter_image_comm) and the existing "pool" communicators (intra_pool_comm and inter_pool_comm) are now vectors of length given by the number of parallel images. #ifdef __PARA is no longer needed because all "parallel" variables are always initialized for a serial run and all parallel routines are, in the case of a serial run, dummy routines. The wrappers to MPI routines used only by PWscf are in the PW/para.f90 file. The others (mp_***) are in the Modules/mp.f90. All explicit referencies to mpif.h should be replaced by an "USE parallel_include" (in a serial run parallel_include is simply a dummy module). 2) The extrapolation of both potential and wavefunctions has been rewritten in order to be smarter than before: on the basis of the required extrapolation order, on the basis of the history and on the basis of which files are really present on the disk, the algorithm chooses the extrapolation order. All the algorithms in which ions are moved can use the extrapolation. These are both unstable features: I need the help of everybody to test them. C.S. git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@742 c92efa57-630b-4861-b058-cf58834340f0
2004-03-24 17:36:50 +08:00
!
!
Fix for check_stop (N. Nemec) git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@6376 c92efa57-630b-4861-b058-cf58834340f0
2010-02-09 23:16:28 +08:00
! ... cclock is a C function returning the elapsed solar
- fixed a configuration problem with pgi compiler and preprocessor - elapsed_seconds substituted by cclock - fortran version of scnds substituted by c version. now time information is taken only through: cclock for the real wall clock time scnds for cpu time of the running process defined in file clib/cptimer.c git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@3400 c92efa57-630b-4861-b058-cf58834340f0
2006-09-20 01:00:18 +08:00
! ... time in seconds since the Epoch ( 00:00:00 1/1/1970 )
1) An experimental parallelization of NEB images has been implemented. This required a deep modification of the parallelism in PWscf: there are two new communicators (intra_image_comm and inter_image_comm) and the existing "pool" communicators (intra_pool_comm and inter_pool_comm) are now vectors of length given by the number of parallel images. #ifdef __PARA is no longer needed because all "parallel" variables are always initialized for a serial run and all parallel routines are, in the case of a serial run, dummy routines. The wrappers to MPI routines used only by PWscf are in the PW/para.f90 file. The others (mp_***) are in the Modules/mp.f90. All explicit referencies to mpif.h should be replaced by an "USE parallel_include" (in a serial run parallel_include is simply a dummy module). 2) The extrapolation of both potential and wavefunctions has been rewritten in order to be smarter than before: on the basis of the required extrapolation order, on the basis of the history and on the basis of which files are really present on the disk, the algorithm chooses the extrapolation order. All the algorithms in which ions are moved can use the extrapolation. These are both unstable features: I need the help of everybody to test them. C.S. git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@742 c92efa57-630b-4861-b058-cf58834340f0
2004-03-24 17:36:50 +08:00
!
Some cleanup: the info message about reading/writing the restart file now indicates the restart directory (the only file users are aware of) and not the xml descriptor. C.S. git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@2432 c92efa57-630b-4861-b058-cf58834340f0
2005-11-08 02:56:23 +08:00
IF ( .NOT. tinit ) &
1) An experimental parallelization of NEB images has been implemented. This required a deep modification of the parallelism in PWscf: there are two new communicators (intra_image_comm and inter_image_comm) and the existing "pool" communicators (intra_pool_comm and inter_pool_comm) are now vectors of length given by the number of parallel images. #ifdef __PARA is no longer needed because all "parallel" variables are always initialized for a serial run and all parallel routines are, in the case of a serial run, dummy routines. The wrappers to MPI routines used only by PWscf are in the PW/para.f90 file. The others (mp_***) are in the Modules/mp.f90. All explicit referencies to mpif.h should be replaced by an "USE parallel_include" (in a serial run parallel_include is simply a dummy module). 2) The extrapolation of both potential and wavefunctions has been rewritten in order to be smarter than before: on the basis of the required extrapolation order, on the basis of the history and on the basis of which files are really present on the disk, the algorithm chooses the extrapolation order. All the algorithms in which ions are moved can use the extrapolation. These are both unstable features: I need the help of everybody to test them. C.S. git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@742 c92efa57-630b-4861-b058-cf58834340f0
2004-03-24 17:36:50 +08:00
CALL errore( 'check_stop_now', 'check_stop not initialized', 1 )
!
Tests on cpu time and soft exit adaptet to "parallel" NEB. C.S. git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@746 c92efa57-630b-4861-b058-cf58834340f0
2004-03-25 20:09:40 +08:00
unit = stdout
IF ( PRESENT( inunit ) ) unit = inunit
!
1) An experimental parallelization of NEB images has been implemented. This required a deep modification of the parallelism in PWscf: there are two new communicators (intra_image_comm and inter_image_comm) and the existing "pool" communicators (intra_pool_comm and inter_pool_comm) are now vectors of length given by the number of parallel images. #ifdef __PARA is no longer needed because all "parallel" variables are always initialized for a serial run and all parallel routines are, in the case of a serial run, dummy routines. The wrappers to MPI routines used only by PWscf are in the PW/para.f90 file. The others (mp_***) are in the Modules/mp.f90. All explicit referencies to mpif.h should be replaced by an "USE parallel_include" (in a serial run parallel_include is simply a dummy module). 2) The extrapolation of both potential and wavefunctions has been rewritten in order to be smarter than before: on the basis of the required extrapolation order, on the basis of the history and on the basis of which files are really present on the disk, the algorithm chooses the extrapolation order. All the algorithms in which ions are moved can use the extrapolation. These are both unstable features: I need the help of everybody to test them. C.S. git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@742 c92efa57-630b-4861-b058-cf58834340f0
2004-03-24 17:36:50 +08:00
check_stop_now = .FALSE.
Added signal processing by RdM. Disabled by default, enabled by option -D__TRAP_SIGUSR1 or using "./configure --enable-signals" git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@6370 c92efa57-630b-4861-b058-cf58834340f0
2010-02-09 03:25:34 +08:00
!
signaled = .FALSE.
Fix for check_stop (N. Nemec) git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@6376 c92efa57-630b-4861-b058-cf58834340f0
2010-02-09 23:16:28 +08:00
!
Small changes and comment lines to run CP with nimage>1. Grep for "KNK_nimage" and uncomment the lines underneath. The images will write trajectory files as in "prefix_N.*", the restarts will be "RESTART_N", and the outputs will be "out.N_0". For the first image things are as usual ("_N" is skipped). Now, this won't do anything useful, and will just run non-communicating copies. For useful stuff (path integral, string method, etc) setup some communication between images in cpr.f90 Kostya git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@5084 c92efa57-630b-4861-b058-cf58834340f0
2008-07-29 10:05:18 +08:00
! KNK_image
! IF ( mpime == root ) THEN
The check on stop conditions must be done by ionode(s) and not by meta_ionode, otherwise parallelization on neb images doesn't work anymore. C.S. git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@3244 c92efa57-630b-4861-b058-cf58834340f0
2006-07-06 05:31:35 +08:00
IF ( ionode ) THEN
1) An experimental parallelization of NEB images has been implemented. This required a deep modification of the parallelism in PWscf: there are two new communicators (intra_image_comm and inter_image_comm) and the existing "pool" communicators (intra_pool_comm and inter_pool_comm) are now vectors of length given by the number of parallel images. #ifdef __PARA is no longer needed because all "parallel" variables are always initialized for a serial run and all parallel routines are, in the case of a serial run, dummy routines. The wrappers to MPI routines used only by PWscf are in the PW/para.f90 file. The others (mp_***) are in the Modules/mp.f90. All explicit referencies to mpif.h should be replaced by an "USE parallel_include" (in a serial run parallel_include is simply a dummy module). 2) The extrapolation of both potential and wavefunctions has been rewritten in order to be smarter than before: on the basis of the required extrapolation order, on the basis of the history and on the basis of which files are really present on the disk, the algorithm chooses the extrapolation order. All the algorithms in which ions are moved can use the extrapolation. These are both unstable features: I need the help of everybody to test them. C.S. git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@742 c92efa57-630b-4861-b058-cf58834340f0
2004-03-24 17:36:50 +08:00
!
INQUIRE( FILE = TRIM( exit_file ), EXIST = tex )
!
IF ( tex ) THEN
Fix for check_stop (N. Nemec) git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@6376 c92efa57-630b-4861-b058-cf58834340f0
2010-02-09 23:16:28 +08:00
!
1) An experimental parallelization of NEB images has been implemented. This required a deep modification of the parallelism in PWscf: there are two new communicators (intra_image_comm and inter_image_comm) and the existing "pool" communicators (intra_pool_comm and inter_pool_comm) are now vectors of length given by the number of parallel images. #ifdef __PARA is no longer needed because all "parallel" variables are always initialized for a serial run and all parallel routines are, in the case of a serial run, dummy routines. The wrappers to MPI routines used only by PWscf are in the PW/para.f90 file. The others (mp_***) are in the Modules/mp.f90. All explicit referencies to mpif.h should be replaced by an "USE parallel_include" (in a serial run parallel_include is simply a dummy module). 2) The extrapolation of both potential and wavefunctions has been rewritten in order to be smarter than before: on the basis of the required extrapolation order, on the basis of the history and on the basis of which files are really present on the disk, the algorithm chooses the extrapolation order. All the algorithms in which ions are moved can use the extrapolation. These are both unstable features: I need the help of everybody to test them. C.S. git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@742 c92efa57-630b-4861-b058-cf58834340f0
2004-03-24 17:36:50 +08:00
check_stop_now = .TRUE.
!
OPEN( UNIT = iunexit, FILE = TRIM( exit_file ) )
CLOSE( UNIT = iunexit, STATUS = 'DELETE' )
!
prefix.EXIT was broken by the last update (QE-3.1.1 does not work). Kostya git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@3263 c92efa57-630b-4861-b058-cf58834340f0
2006-07-11 04:15:57 +08:00
ELSE
!
- fixed a configuration problem with pgi compiler and preprocessor - elapsed_seconds substituted by cclock - fortran version of scnds substituted by c version. now time information is taken only through: cclock for the real wall clock time scnds for cpu time of the running process defined in file clib/cptimer.c git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@3400 c92efa57-630b-4861-b058-cf58834340f0
2006-09-20 01:00:18 +08:00
seconds = cclock() - init_second
prefix.EXIT was broken by the last update (QE-3.1.1 does not work). Kostya git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@3263 c92efa57-630b-4861-b058-cf58834340f0
2006-07-11 04:15:57 +08:00
!
check_stop_now = ( seconds > max_seconds )
!
1) An experimental parallelization of NEB images has been implemented. This required a deep modification of the parallelism in PWscf: there are two new communicators (intra_image_comm and inter_image_comm) and the existing "pool" communicators (intra_pool_comm and inter_pool_comm) are now vectors of length given by the number of parallel images. #ifdef __PARA is no longer needed because all "parallel" variables are always initialized for a serial run and all parallel routines are, in the case of a serial run, dummy routines. The wrappers to MPI routines used only by PWscf are in the PW/para.f90 file. The others (mp_***) are in the Modules/mp.f90. All explicit referencies to mpif.h should be replaced by an "USE parallel_include" (in a serial run parallel_include is simply a dummy module). 2) The extrapolation of both potential and wavefunctions has been rewritten in order to be smarter than before: on the basis of the required extrapolation order, on the basis of the history and on the basis of which files are really present on the disk, the algorithm chooses the extrapolation order. All the algorithms in which ions are moved can use the extrapolation. These are both unstable features: I need the help of everybody to test them. C.S. git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@742 c92efa57-630b-4861-b058-cf58834340f0
2004-03-24 17:36:50 +08:00
END IF
!
The check on stop conditions must be done by ionode(s) and not by meta_ionode, otherwise parallelization on neb images doesn't work anymore. C.S. git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@3244 c92efa57-630b-4861-b058-cf58834340f0
2006-07-06 05:31:35 +08:00
END IF
!
Fix for check_stop (N. Nemec) git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@6376 c92efa57-630b-4861-b058-cf58834340f0
2010-02-09 23:16:28 +08:00
#if defined __TRAP_SIGUSR1
Added signal processing by RdM. Disabled by default, enabled by option -D__TRAP_SIGUSR1 or using "./configure --enable-signals" git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@6370 c92efa57-630b-4861-b058-cf58834340f0
2010-02-09 03:25:34 +08:00
signaled = signal_detected()
check_stop_now = check_stop_now .OR. signaled
tex = tex .OR. signaled
Fix for check_stop (N. Nemec) git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@6376 c92efa57-630b-4861-b058-cf58834340f0
2010-02-09 23:16:28 +08:00
#endif
Added signal processing by RdM. Disabled by default, enabled by option -D__TRAP_SIGUSR1 or using "./configure --enable-signals" git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@6370 c92efa57-630b-4861-b058-cf58834340f0
2010-02-09 03:25:34 +08:00
!
Small changes and comment lines to run CP with nimage>1. Grep for "KNK_nimage" and uncomment the lines underneath. The images will write trajectory files as in "prefix_N.*", the restarts will be "RESTART_N", and the outputs will be "out.N_0". For the first image things are as usual ("_N" is skipped). Now, this won't do anything useful, and will just run non-communicating copies. For useful stuff (path integral, string method, etc) setup some communication between images in cpr.f90 Kostya git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@5084 c92efa57-630b-4861-b058-cf58834340f0
2008-07-29 10:05:18 +08:00
! KNK_image
! CALL mp_bcast( check_stop_now, root, world_comm )
The check on stop conditions must be done by ionode(s) and not by meta_ionode, otherwise parallelization on neb images doesn't work anymore. C.S. git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@3244 c92efa57-630b-4861-b058-cf58834340f0
2006-07-06 05:31:35 +08:00
CALL mp_bcast( check_stop_now, ionode_id, intra_image_comm )
!
IF ( check_stop_now .AND. meta_ionode ) THEN
!
IF ( tex ) THEN
1) An experimental parallelization of NEB images has been implemented. This required a deep modification of the parallelism in PWscf: there are two new communicators (intra_image_comm and inter_image_comm) and the existing "pool" communicators (intra_pool_comm and inter_pool_comm) are now vectors of length given by the number of parallel images. #ifdef __PARA is no longer needed because all "parallel" variables are always initialized for a serial run and all parallel routines are, in the case of a serial run, dummy routines. The wrappers to MPI routines used only by PWscf are in the PW/para.f90 file. The others (mp_***) are in the Modules/mp.f90. All explicit referencies to mpif.h should be replaced by an "USE parallel_include" (in a serial run parallel_include is simply a dummy module). 2) The extrapolation of both potential and wavefunctions has been rewritten in order to be smarter than before: on the basis of the required extrapolation order, on the basis of the history and on the basis of which files are really present on the disk, the algorithm chooses the extrapolation order. All the algorithms in which ions are moved can use the extrapolation. These are both unstable features: I need the help of everybody to test them. C.S. git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@742 c92efa57-630b-4861-b058-cf58834340f0
2004-03-24 17:36:50 +08:00
!
The check on stop conditions must be done by ionode(s) and not by meta_ionode, otherwise parallelization on neb images doesn't work anymore. C.S. git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@3244 c92efa57-630b-4861-b058-cf58834340f0
2006-07-06 05:31:35 +08:00
WRITE( UNIT = unit, &
FMT = '(/,5X,"Program stopped by user request")' )
!
ELSE
1) An experimental parallelization of NEB images has been implemented. This required a deep modification of the parallelism in PWscf: there are two new communicators (intra_image_comm and inter_image_comm) and the existing "pool" communicators (intra_pool_comm and inter_pool_comm) are now vectors of length given by the number of parallel images. #ifdef __PARA is no longer needed because all "parallel" variables are always initialized for a serial run and all parallel routines are, in the case of a serial run, dummy routines. The wrappers to MPI routines used only by PWscf are in the PW/para.f90 file. The others (mp_***) are in the Modules/mp.f90. All explicit referencies to mpif.h should be replaced by an "USE parallel_include" (in a serial run parallel_include is simply a dummy module). 2) The extrapolation of both potential and wavefunctions has been rewritten in order to be smarter than before: on the basis of the required extrapolation order, on the basis of the history and on the basis of which files are really present on the disk, the algorithm chooses the extrapolation order. All the algorithms in which ions are moved can use the extrapolation. These are both unstable features: I need the help of everybody to test them. C.S. git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@742 c92efa57-630b-4861-b058-cf58834340f0
2004-03-24 17:36:50 +08:00
!
Tests on cpu time and soft exit adaptet to "parallel" NEB. C.S. git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@746 c92efa57-630b-4861-b058-cf58834340f0
2004-03-25 20:09:40 +08:00
WRITE( UNIT = unit, &
FMT = '(/,5X,"Maximum CPU time exceeded")' )
WRITE( UNIT = unit, &
FMT = '(/,5X,"max_seconds = ",F10.2)' ) max_seconds
WRITE( UNIT = unit, &
FMT = '(5X,"elapsed seconds = ",F10.2)' ) seconds
1) An experimental parallelization of NEB images has been implemented. This required a deep modification of the parallelism in PWscf: there are two new communicators (intra_image_comm and inter_image_comm) and the existing "pool" communicators (intra_pool_comm and inter_pool_comm) are now vectors of length given by the number of parallel images. #ifdef __PARA is no longer needed because all "parallel" variables are always initialized for a serial run and all parallel routines are, in the case of a serial run, dummy routines. The wrappers to MPI routines used only by PWscf are in the PW/para.f90 file. The others (mp_***) are in the Modules/mp.f90. All explicit referencies to mpif.h should be replaced by an "USE parallel_include" (in a serial run parallel_include is simply a dummy module). 2) The extrapolation of both potential and wavefunctions has been rewritten in order to be smarter than before: on the basis of the required extrapolation order, on the basis of the history and on the basis of which files are really present on the disk, the algorithm chooses the extrapolation order. All the algorithms in which ions are moved can use the extrapolation. These are both unstable features: I need the help of everybody to test them. C.S. git-svn-id: http://qeforge.qe-forge.org/svn/q-e/trunk/espresso@742 c92efa57-630b-4861-b058-cf58834340f0
2004-03-24 17:36:50 +08:00
!
END IF
!
END IF
!
RETURN
!
END FUNCTION check_stop_now
!
END MODULE check_stop