Update Fortran MPI interface: "use mpi" instead of "include mpif.h"#46
Merged
danielhollas merged 2 commits intomasterfrom Nov 9, 2020
Merged
Update Fortran MPI interface: "use mpi" instead of "include mpif.h"#46danielhollas merged 2 commits intomasterfrom
danielhollas merged 2 commits intomasterfrom
Conversation
Codecov Report
@@ Coverage Diff @@
## master #46 +/- ##
=======================================
Coverage 54.67% 54.67%
=======================================
Files 36 36
Lines 5536 5536
=======================================
Hits 3027 3027
Misses 2509 2509
|
danielhollas
force-pushed
the
use-mpi
branch
12 times, most recently
from
033c723 to
404b2cd
Compare
danielhollas
commented
Nov 7, 2020
| @@ -1,5 +1,5 @@ | |||
| CXX = g++ | |||
| CXXFLAGS = -g -O2 -Wall | |||
| CXXFLAGS = -O2 -Wall -std=c++11 | |||
Contributor
Author
There was a problem hiding this comment.
This was overlooked in the last PR. :-(
| end if | ||
|
|
||
| timer = MPI_WTIME(ierr) | ||
| timer = MPI_WTIME() |
Contributor
Author
There was a problem hiding this comment.
MPI_WTIME() does not return any errors. This kind of error can be caught by using use mpi instead of include mpif.h.
| call MPI_recv(fyc_new ,size1*size2, MPI_DOUBLE_PRECISION, dest, tag_fy, MPI_COMM_WORLD, status, ierr ) | ||
| call MPI_recv(fzc_new, size1*size2, MPI_DOUBLE_PRECISION, dest, tag_fz, MPI_COMM_WORLD, status, ierr ) | ||
| call MPI_recv(irank, 1, MPI_INTEGER, dest, tag_fz, MPI_COMM_WORLD, status, ierr ) | ||
| call MPI_Send(x, size1*size2, MPI_DOUBLE_PRECISION, dest, tag_x, MPI_COMM_WORLD, ierr) |
Contributor
Author
There was a problem hiding this comment.
Here again, the status parameter should not be there in MPI_Send(). Hopefully, this did not introduce any practical problem in the REMD code, since we're not checking the ierr variable anyway (but we really should).
|
|
||
| # TODO: This is extremely brittle, we should at least | ||
| # verify that the number of parameters is correct! | ||
| if [[ $# -ne 7 ]]; then |
Contributor
Author
There was a problem hiding this comment.
Changes in test.sh and Makefile do not actually pertain to MPI, but they were a nice cleanup.
| .cpp.o: | ||
| $(CXX) $(CXXLAGS) $(DFLAGS) -c $< | ||
| %.o: %.cpp | ||
| $(CXX) $(CXXFLAGS) $(DFLAGS) -c $< |
Contributor
Author
There was a problem hiding this comment.
Whoops, previous PR had a typo, CXXLAGS -> CXXFLAGS
|
|
||
| .cpp.o: | ||
| $(CXX) $(CXXLAGS) $(DFLAGS) -c $< | ||
| %.o: %.cpp |
Contributor
Author
There was a problem hiding this comment.
This is a more modern approach, as per:
https://www.gnu.org/software/make/manual/html_node/Suffix-Rules.html
| testclean : | ||
| /bin/bash TESTS/test.sh ${BIN} clean ${MPI} ${FFTW} $(PLUMED) ${CP2K} | ||
| testclean: | ||
| /bin/bash TESTS/test.sh ${BIN} $(TEST) ${MPI} ${FFTW} $(PLUMED) ${CP2K} clean |
Contributor
Author
There was a problem hiding this comment.
This is now more consistent.
danielhollas
changed the title
Closes #33. We might be able to use 'use mpi_f08', which is recommended, but need to test whether it's supported by all compiler versions in our CI suite. It might also require some code changes. See https://www.mpi-forum.org/docs/mpi-3.1/mpi31-report/node408.htm As an unfortunate side-effect, we now need to build our own MPICH and OpenMPI in Github Actions, for each GFortran version, since use mpi does not allow using different GFortran versions for MPI compiler and ABIN. Probably a good thing to do anyway. The MPICH and OpenMPI builds take a long time, but we cache them so it's not so bad (GA cache should last at least a week).
suchanj
approved these changes
Nov 9, 2020
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Closes #33. This change uncovered a couple of bugs we had in various MPI calls, fortunately nothing critical.
As an unfortunate side-effect, we now need to build our own MPICH and OpenMPI in Github Actions, for each GFortran version,
since
use mpidoes not allow using different GFortran versions for MPI compiler and ABIN. That is unfortunate, but probably a good thing to do anyway. The MPICH and OpenMPI builds take a long time, but we cache them so it's not so bad (GA cache should last at least a week).As part of this, I also did a bit more cleanup of
MakefileandTEST/test.sh.