Re: [AMBER-Developers] mpi4py issues in AMBER22

From: Hai Nguyen via AMBER-Developers <amber-developers.ambermd.org>
Date: Sun, 18 Sep 2022 19:46:56 -0400

On Wed, Aug 17, 2022 at 5:12 PM Ray Luo via AMBER-Developers <
amber-developers.ambermd.org> wrote:

> We also met the same problem ... mpi4py is not installed in Amber's
> miniconda distribution.


I guess the conda distribution is for serial code while mpi4py is for
parallel one.

.Dave: i think the ambertools conda maintainer can add mpi4py as dependency
for ambertools serial.

Hai

>
>
> Ray
> --
> Ray Luo, Ph.D.
> Professor of Structural Biology/Biochemistry/Biophysics,
> Chemical and Materials Physics, Chemical and Biomolecular Engineering,
> Biomedical Engineering, and Materials Science and Engineering
> Department of Molecular Biology and Biochemistry
> University of California, Irvine, CA 92697-3900
>
> On Wed, Aug 17, 2022 at 1:24 PM Ilyas Yildirim via AMBER-Developers
> <amber-developers.ambermd.org> wrote:
> >
> > There seems to be an issue again with mpi4py, which is required by
> mmpbsa calculations. I am not sure where to start but I do not think that
> cmake installs mpi4py at all (in my case). After serial installation, the
> parallel installation is done with the following options:
> >
> > # Assume this is Linux:
> >
> > cmake $AMBER_PREFIX/amber22_src \
> > -DCMAKE_INSTALL_PREFIX=$AMBER_PREFIX/amber22 \
> > -DCOMPILER=GNU \
> > -DMPI=YES -DCUDA=FALSE -DINSTALL_TESTS=TRUE \
> > -DDOWNLOAD_MINICONDA=TRUE \
> > 2>&1 | tee cmake.log
> >
> > When I check out the cmake.log file after running run_cmake, I am seeing
> in two lines the word "mpi4py":
> >
> > -- Checking for Python package mpi4py -- not found
> >
> > -- ---building bundled:
> -----------------------------------------------------.
> > .
> > .
> > .
> > -- [[32mmpi4py[[0m [[1;29m- MPI support library for MMPBSA.py[[0m
> > .
> > .
> >
> > After the compilation, when I perform mmpbsa calculations as I normally
> do, I see the following output:
> >
> > Traceback (most recent call last):
> > File "/home/iyildirim/programs/amber22/bin/MMPBSA.py.MPI", line 55, in
> <module>
> > raise MMPBSA_Error('Could not import mpi4py package! Use serial
> version '
> > MMPBSA_mods.exceptions.MMPBSA_Error: Could not import mpi4py package!
> Use serial version or install mpi4py.
> > Traceback (most recent call last):
> > File "/home/iyildirim/programs/amber22/bin/MMPBSA.py.MPI", line 53, in
> <module>
> > Traceback (most recent call last):
> > File "/home/iyildirim/programs/amber22/bin/MMPBSA.py.MPI", line 53, in
> <module>
> > from mpi4py import MPI
> > ModuleNotFoundError: No module named 'mpi4py'
> >
> > When I check out if mpi4py is already installed in amber.python I am
> seeing the following output:
> >
> > iyildirim.rnachem:~/programs/amber20_src/build$ amber.python
> > Python 3.9.12 (main, Apr 5 2022, 06:56:58)
> > [GCC 7.5.0] :: Anaconda, Inc. on linux
> > Type "help", "copyright", "credits" or "license" for more information.
> > >>> import mpi4py
> > Traceback (most recent call last):
> > File "<stdin>", line 1, in <module>
> > ModuleNotFoundError: No module named 'mpi4py'
> > >>>
> >
> > PYTHONPATH is directed to the following directory:
> >
> > iyildirim.rnachem:~/programs/amber20_src/build$ echo $PYTHONPATH
> > /home/iyildirim/programs/amber22/lib/python3.9/site-packages
> >
> > Also, I do not see any mpi4py file inside $AMBERHOME:
> >
> > iyildirim.rnachem:~/programs/amber22$ find ./ -name "*mpi4py*"
> > iyildirim.rnachem:~/programs/amber22$
> >
> > Weird thing is that the files are present in the XXX_src directory:
> >
> > iyildirim.rnachem:~/programs/amber22_src$ find ./ -name "*mpi4py*"
> > ./build/AmberTools/src/mpi4py-3.1.3
> > ./build/AmberTools/src/mpi4py-3.1.3/CMakeFiles/mpi4py-build
> >
> ./build/AmberTools/src/mpi4py-3.1.3/CMakeFiles/mpi4py-build/mpi4py-build.stamp
> >
> ./build/AmberTools/src/mpi4py-3.1.3/CMakeFiles/mpi4py-build/lib.linux-x86_64-3.9/mpi4py
> >
> ./build/AmberTools/src/mpi4py-3.1.3/CMakeFiles/mpi4py-build/lib.linux-x86_64-3.9/mpi4py/include/mpi4py
> >
> ./build/AmberTools/src/mpi4py-3.1.3/CMakeFiles/mpi4py-build/lib.linux-x86_64-3.9/mpi4py/include/mpi4py/mpi4py.MPI.h
> >
> ./build/AmberTools/src/mpi4py-3.1.3/CMakeFiles/mpi4py-build/lib.linux-x86_64-3.9/mpi4py/include/mpi4py/mpi4py.MPI_api.h
> >
> ./build/AmberTools/src/mpi4py-3.1.3/CMakeFiles/mpi4py-build/lib.linux-x86_64-3.9/mpi4py/include/mpi4py/mpi4py.i
> >
> ./build/AmberTools/src/mpi4py-3.1.3/CMakeFiles/mpi4py-build/lib.linux-x86_64-3.9/mpi4py/include/mpi4py/mpi4py.h
> > ./build/AmberTools/src/mpi4py-3.1.3/CMakeFiles/mpi4py.dir
> > ./AmberTools/src/mpi4py-3.1.3
> > ./AmberTools/src/mpi4py-3.1.3/src/mpi4py
> > ./AmberTools/src/mpi4py-3.1.3/src/mpi4py/include/mpi4py
> > ./AmberTools/src/mpi4py-3.1.3/src/mpi4py/include/mpi4py/mpi4py.MPI.h
> > ./AmberTools/src/mpi4py-3.1.3/src/mpi4py/include/mpi4py/mpi4py.MPI_api.h
> > ./AmberTools/src/mpi4py-3.1.3/src/mpi4py/include/mpi4py/mpi4py.i
> > ./AmberTools/src/mpi4py-3.1.3/src/mpi4py/include/mpi4py/mpi4py.h
> > ./AmberTools/src/mpi4py-3.1.3/src/mpi4py.egg-info
> > ./AmberTools/src/mpi4py-3.1.3/src/mpi4py.MPI.c
> >
> > It seems that mpi4py is compiled within _src but not copied to
> $AMBERHOME. I am seeing also the following output in CMakeCache.txt:
> >
> > iyildirim.rnachem:~/programs/amber22_src/build$ grep mpi4py *
> > grep: amber-modules: Is a directory
> > grep: AmberTools: Is a directory
> > grep: benchmarks: Is a directory
> > grep: cmake: Is a directory
> > CMakeCache.txt://Whether the Python package mpi4py was found
> >
> CMakeCache.txt:mpi4py_BINARY_DIR:STATIC=/home/iyildirim/programs/amber22_src/build/AmberTools/src/mpi4py-3.1.3
> > CMakeCache.txt:mpi4py_IS_TOP_LEVEL:STATIC=OFF
> >
> CMakeCache.txt:mpi4py_SOURCE_DIR:STATIC=/home/iyildirim/programs/amber22_src/AmberTools/src/mpi4py-3.1.3
> > grep: CMakeFiles: Is a directory
> >
> > Is the line "mpi4py_IS_TOP_LEVEL:STATIC=OFF" normal or should it be ON?
> Any ideas on what is going on? Thanks.
> >
> > Best,
> >
> > Ilyas Yildirim, Ph.D.
> > ---------------------------------------------------------
> > = Assistant Professor =
> > = Department of Chemistry & Biochemistry =
> > = Florida Atlantic University =
> > = 5353 Parkside Drive, Jupiter, FL 33458 =
> > = Building MC17, Room # 224 =
> > = Phone: +1(561)799-8325 | E-mail: iyildirim_at_fau.edu =
> > ---------------------------------------------------------
> > = Website:
> https://urldefense.com/v3/__https://cescos.fau.edu/*iyildirim/__;fg!!CzAuKJ42GuquVTTmVmPViYEvSg!Lk5Jx5_6o9jQNim_0OAYWmZlu9vmsdR8IvFN0_PagMwM-Zeh9QBeyByS7xEUNzZ87gDtwzijcaiVFVQJv1MzfcrXdCzV$
> =
> > =
> https://urldefense.com/v3/__http://scholar.google.com/citations?user=O6RQCcwAAAAJ__;!!CzAuKJ42GuquVTTmVmPViYEvSg!Lk5Jx5_6o9jQNim_0OAYWmZlu9vmsdR8IvFN0_PagMwM-Zeh9QBeyByS7xEUNzZ87gDtwzijcaiVFVQJv1MzfUvYaLgU$
> =
> > ---------------------------------------------------------
> > _______________________________________________
> > AMBER-Developers mailing list
> > AMBER-Developers.ambermd.org
> >
> https://urldefense.com/v3/__http://lists.ambermd.org/mailman/listinfo/amber-developers__;!!CzAuKJ42GuquVTTmVmPViYEvSg!Lk5Jx5_6o9jQNim_0OAYWmZlu9vmsdR8IvFN0_PagMwM-Zeh9QBeyByS7xEUNzZ87gDtwzijcaiVFVQJv1MzfSefA3t7$
>
> _______________________________________________
> AMBER-Developers mailing list
> AMBER-Developers.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber-developers
>
_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Sun Sep 18 2022 - 17:00:03 PDT
Custom Search