[gmx-users] Tipps for compiling gromacs 5 on BlueGene/Q
Hi,
I try to compile gromacs 5 on a BlueGene/Q.
The non-mpi version works fine with letting gromacs build its own fftw libaries.
With mpi the generated file mdrun_mpi seems to be corrupt.
Running it from both the login nodes and the cluster nodes gives:
>mdrun_mpi
/bgsys/source/srcV1R2M2.15270/comm/sys/buildtools/pami/common/bgq/BgqPersonality.h:102:
/bgsys/source/srcV1R2M2.15270/comm/sys/buildtools/pami/common/bgq/BgqPersonality.h<102>
Aborted (core dumped)
I would appreciate some help.
In the following, I show my build script.
Is there anything wrong?
module purge
module load cmake/2.8.12.1
module load mpich2
export FFLAGS='-O3 -qarch=qp -qtune=qp -qmaxmem=-1 ' #it was recommended to
set this flags from the BGQ user guide
export CFLAGS='-O3 -qarch=qp -qtune=qp -qmaxmem=-1 '
cd ${BUILD_DIR}
cmake ${SOURCE_DIR} \
-DCMAKE_C_COMPILER=mpicc \
-DCMAKE_CXX_COMPILER=mpicxx \
-DCMAKE_INSTALL_PREFIX=${INSTALL_DIR} \
-DCMAKE_TOOLCHAIN_FILE=PATH/TO/BlueGeneQ-static-XL-CXX.cmake \
-DGMX_BUILD_OWN_FFTW=ON \
-DGMX_DEFAULT_SUFFIX=ON \
-DGMX_MPI=ON \
-DGMX_BUILD_MDRUN_ONLY=ON \
-DGMX_RELAXED_DOUBLE_PRECISION=ON
make -j `nproc`
make install
As I said, this produces a corrupt mdrun_mpi file.
Thanks
Soren
--
Gromacs Users mailing list
* Please search the archive at
http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before posting!
* Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
* For (un)subscribe requests visit
https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or send a
mail to gmx-users-requ...@gromacs.org.
Re: [gmx-users] tunepme does not start
Then I misunderstood the function of -tunepme. I thought it replaces g_tune_pme, but that is apparently not the case. Thanks Soren -- Gromacs Users mailing list * Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before posting! * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists * For (un)subscribe requests visit https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or send a mail to gmx-users-requ...@gromacs.org.
[gmx-users] genbox solvate insert_molecules
Hi, when I start genbox with GMX5.x, I get the message: This tool has been removed from Gromacs 5.0. Please see http://www.gromacs.org/Documentation/How-tos/Tool_Changes_for_5.0 at this website it is stated: "This tool has been split to gmx solvate and gmx insert-molecules." Apparently, these tools do not exist on my system. Should they be compiled by default? How can I generate these files if not? best Soren -- Gromacs Users mailing list * Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before posting! * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists * For (un)subscribe requests visit https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or send a mail to gmx-users-requ...@gromacs.org.
[gmx-users] maximum number of specbonds?
Hi, I am parameterizing an artifical ligand. I use small residue-like fragments and connect them with specbonds in a grid, since I need arbitrary shapes and sizes. Now, I ran into segmentation faults, when the grid becomes to large. Is there an upper limit for the specbonds? How can I fix that or do you see a better way to solve that issue? Thanks Soren -- Gromacs Users mailing list * Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before posting! * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists * For (un)subscribe requests visit https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or send a mail to gmx-users-requ...@gromacs.org.
Re: [gmx-users] maximum number of specbonds?
Something is really weird about that system. The error message changes with the number of residues I use. Everything works fine for 2 and 3 residues. With 4 residues I get : ___ Warning: Starting residue DUL1 in chain not identified as Protein/RNA/DNA. Warning: Starting residue DUL2 in chain not identified as Protein/RNA/DNA. Warning: Starting residue DUL3 in chain not identified as Protein/RNA/DNA. Warning: Starting residue DUL4 in chain not identified as Protein/RNA/DNA. Problem with chain definition, or missing terminal residues. This chain does not appear to contain a recognized chain molecule. If this is incorrect, you can edit residuetypes.dat to modify the behavior. 11 out of 11 lines of specbond.dat converted successfully Special Atom Distance matrix: DUL1DUL1DUL1DUL2DUL2DUL2DUL3 DU01DU12DU23DU04DU15DU26DU07 DUL1DU12 0.200 DUL1DU23 0.346 0.200 DUL2DU04 0.400 0.346 0.529 DUL2DU15 0.529 0.400 0.529 0.200 DUL2DU26 0.529 0.346 0.400 0.346 0.200 DUL3DU07 0.400 0.600 0.721 0.693 0.872 0.916 DUL3DU18 0.200 0.400 0.529 0.529 0.693 0.721 0.200 DUL3DU29 0.200 0.346 0.400 0.600 0.721 0.693 0.346 DUL4 DU010 0.400 0.529 0.721 0.400 0.600 0.721 0.400 DUL4 DU111 0.346 0.400 0.600 0.200 0.400 0.529 0.529 DUL4 DU212 0.200 0.200 0.400 0.200 0.346 0.400 0.529 DUL3DUL3DUL4DUL4 DU18DU29 DU010 DU111 DUL3DU29 0.200 DUL4 DU010 0.346 0.529 DUL4 DU111 0.400 0.529 0.200 DUL4 DU212 0.346 0.400 0.346 0.200 Linking DUL-1 DU0-1 and DUL-1 DU1-2... Linking DUL-1 DU0-1 and DUL-3 DU1-8... Linking DUL-1 DU0-1 and DUL-3 DU2-9... Linking DUL-1 DU0-1 and DUL-4 DU2-12... Linking DUL-1 DU1-2 and DUL-1 DU2-3... Linking DUL-1 DU1-2 and DUL-4 DU2-12... Linking DUL-2 DU0-4 and DUL-2 DU1-5... Linking DUL-2 DU0-4 and DUL-4 DU1-11... Linking DUL-2 DU0-4 and DUL-4 DU2-12... Linking DUL-2 DU1-5 and DUL-2 DU2-6... Linking DUL-3 DU0-7 and DUL-3 DU1-8... Linking DUL-3 DU1-8 and DUL-3 DU2-9... Linking DUL-4 DU0-10 and DUL-4 DU1-11... Linking DUL-4 DU1-11 and DUL-4 DU2-12... Segmentation fault (core dumped) ___ and with 5 residues: ___ inking DUL-1 DU0-1 and DUL-1 DU1-2... Linking DUL-1 DU0-1 and DUL-3 DU1-8... Linking DUL-1 DU0-1 and DUL-3 DU2-9... Linking DUL-1 DU0-1 and DUL-4 DU2-12... Linking DUL-1 DU1-2 and DUL-1 DU2-3... Linking DUL-1 DU1-2 and DUL-4 DU2-12... Linking DUL-2 DU0-4 and DUL-2 DU1-5... Linking DUL-2 DU0-4 and DUL-4 DU1-11... Linking DUL-2 DU0-4 and DUL-4 DU2-12... Linking DUL-2 DU0-4 and DUL-5 DU2-15... Linking DUL-2 DU1-5 and DUL-2 DU2-6... Linking DUL-2 DU1-5 and DUL-5 DU2-15... Linking DUL-3 DU0-7 and DUL-3 DU1-8... Linking DUL-3 DU1-8 and DUL-3 DU2-9... Linking DUL-4 DU0-10 and DUL-4 DU1-11... Linking DUL-4 DU1-11 and DUL-4 DU2-12... Linking DUL-5 DU0-13 and DUL-5 DU1-14... Linking DUL-5 DU1-14 and DUL-5 DU2-15... Opening force field file /home/swacker/System/gromacs/GMX46/amber99sb-ildn.ff/aminoacids.arn Opening force field file /home/swacker/System/gromacs/GMX46/amber99sb-ildn.ff/dna.arn Opening force field file /home/swacker/System/gromacs/GMX46/amber99sb-ildn.ff/rna.arn Checking for duplicate atoms Generating any missing hydrogen atoms and/or adding termini. Now there are 5 residues with 15 atoms Making bonds... --- Program gmx pdb2gmx, VERSION 5.1-dev-20140607-3db1d85 Source code file: /home/swacker/Install_Software/gromacs/src/gromacs/gmxpreprocess/pgutil.c, line: 125 Fatal error: Residue 1 named DUL of a molecule in the input file was mapped to an entry in the topology database, but the atom used in an interaction of type special bond in that entry is not found in the input file. Perhaps your atom and/or residue naming needs to be fixed. ___ my specbonds.dat: ___ 11 CYS SG 1 CYS SG 1 0.2 CYS2CYS2 CYS SG 1 HEM FE 2 0.25CYS2HEME CYS SG 1 HEM CAB 1 0.18CYS2HEME CYS SG 1 HEM CAC 1 0.18CYS2HEME HIS NE2 1 HEM FE 1 0.2 HIS1HEME MET SD 1 HEM FE 1 0.24MET HEME CO C 1 HEMEFE 1 0.19CO HEME CYM SG 1 CYM SG 1 0.2 CYS2CYS2 DUL DU0 4 DUL DU1 4 0.2 DUL DUL DUL DU0 4 DUL DU2 4 0.2 DUL DUL DUL DU1 4 DUL DU2 4 0.2 DUL DUL ___ example.pdb _
[gmx-users] Installing gromacs 5.0.2 double precision with shared fftw libraries
Hi,
I have trouble to install gromacs 5.0.2 in double precision with shared fftw
libraries. During the cmake process I get:
Could not find fftw3 library named libfftw3, please specify its location in
CMAKE_PREFIX_PATH or FFTW_LIBRARY by hand (e.g.
-DFFTW_LIBRARY='/path/to/libfftw3.so')
CMake Error at cmake/gmxManageFFTLibraries.cmake:76 (MESSAGE):
Cannot find FFTW 3 (with correct precision - libfftw3f for mixed-precision
GROMACS or libfftw3 for double-precision GROMACS). Either choose the right
precision, choose another FFT(W) library (-DGMX_FFT_LIBRARY), enable the
advanced option to let GROMACS build FFTW 3 for you
(-GMX_BUILD_OWN_FFTW=ON), or use the really slow GROMACS built-in fftpack
library (-DGMX_FFT_LIBRARY=fftpack).
However, I provided the path to libfftwr3.so:
cmake ${SOURCE_DIR} \
-DCMAKE_INSTALL_PREFIX=${INSTALL_DIR} \
-DFFTWF_INCLUDE_DIR=~/Programme/fftw3/double/include \
-DFFTWF_LIBRARY=~/Programme/fftw3/double/lib/libfftw3.so \
-DCMAKE_PREFIX_PATH="/global/software/boost/gcc" \
-DGMX_DOUBLE=on \
-DGMX_CPU_ACCELERATION="SSE4.1"
I also tried "-DFFTWF_LIBRARY=~/Programme/fftw3/double/lib/"
The path is correct:
>ls ~/Programme/fftw3/double/lib/libfftw3.so
~/Programme/fftw3/double/lib/libfftw3.so
What is going wrong here?
kind regards
Soren
--
Dr. Soeren Wacker
Centre For Molecular Simulation, BI # 547
The University of Calgary
2500 University Drive (NW)
Calgary, Alberta
Canada, T2N 1N4
Tel.: ++1 (403) 210 7860
Email: swac...@ucalgary.ca
--
Gromacs Users mailing list
* Please search the archive at
http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before posting!
* Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
* For (un)subscribe requests visit
https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or send a
mail to gmx-users-requ...@gromacs.org.
Re: [gmx-users] Installing gromacs 5.0.2 double precision with shared fftw libraries
Hi Johnny,
I am using intel 12.1.4 compiler:
gcc version 4.4.7 20120313 (Red Hat 4.4.7-3) (GCC)
icc version 12.1.4 (gcc version 4.4.7 compatibility)
icpc version 12.1.4 (gcc version 4.4.7 compatibility)
and
Intel(R) Xeon(R) CPU X5550 @ 2.67GHz
regards
Soren
From: gromacs.org_gmx-users-boun...@maillist.sys.kth.se
[gromacs.org_gmx-users-boun...@maillist.sys.kth.se] on behalf of Johnny Lu
[johnny.lu...@gmail.com]
Sent: Wednesday, November 19, 2014 11:55 AM
To: gmx-us...@gromacs.org
Subject: Re: [gmx-users] Installing gromacs 5.0.2 double precision with shared
fftw libraries
Or, just let it build its own fftw3?
cmake ${SOURCE_DIR} \
-DCMAKE_INSTALL_PREFIX=${INSTALL_DIR} \
-GMX_BUILD_OWN_FFTW=ON \
-DCMAKE_PREFIX_PATH="/global/software/boost/gcc" \
-DGMX_DOUBLE=on \
-DGMX_CPU_ACCELERATION="SSE4.1"
What compilers and cpu are you using?
On Wed, Nov 19, 2014 at 1:43 PM, Carsten Kutzner wrote:
> Hi Sören,
>
> instead of providing FFTWF_INCLUDE_DIR and FFTWF_LIBRARY you
> could also put ~/Programme/fftw3/double/ in CMAKE_PREFIX_PATH as well.
>
> Viele Grüße!
> Carsten
>
>
> On 19 Nov 2014, at 19:25, Soren Wacker wrote:
>
> > Hi,
> >
> > I have trouble to install gromacs 5.0.2 in double precision with shared
> fftw libraries. During the cmake process I get:
> >
> > Could not find fftw3 library named libfftw3, please specify its location
> in CMAKE_PREFIX_PATH or FFTW_LIBRARY by hand (e.g.
> -DFFTW_LIBRARY='/path/to/libfftw3.so')
> > CMake Error at cmake/gmxManageFFTLibraries.cmake:76 (MESSAGE):
> > Cannot find FFTW 3 (with correct precision - libfftw3f for
> mixed-precision
> > GROMACS or libfftw3 for double-precision GROMACS). Either choose the
> right
> > precision, choose another FFT(W) library (-DGMX_FFT_LIBRARY), enable the
> > advanced option to let GROMACS build FFTW 3 for you
> > (-GMX_BUILD_OWN_FFTW=ON), or use the really slow GROMACS built-in
> fftpack
> > library (-DGMX_FFT_LIBRARY=fftpack).
> >
> > However, I provided the path to libfftwr3.so:
> >
> > cmake ${SOURCE_DIR} \
> > -DCMAKE_INSTALL_PREFIX=${INSTALL_DIR} \
> > -DFFTWF_INCLUDE_DIR=~/Programme/fftw3/double/include \
> > -DFFTWF_LIBRARY=~/Programme/fftw3/double/lib/libfftw3.so \
> > -DCMAKE_PREFIX_PATH="/global/software/boost/gcc" \
> > -DGMX_DOUBLE=on \
> > -DGMX_CPU_ACCELERATION="SSE4.1"
> >
> > I also tried "-DFFTWF_LIBRARY=~/Programme/fftw3/double/lib/"
> >
> > The path is correct:
> >> ls ~/Programme/fftw3/double/lib/libfftw3.so
> > ~/Programme/fftw3/double/lib/libfftw3.so
> >
> > What is going wrong here?
> >
> > kind regards
> > Soren
> >
> > --
> > Dr. Soeren Wacker
> > Centre For Molecular Simulation, BI # 547
> > The University of Calgary
> > 2500 University Drive (NW)
> > Calgary, Alberta
> > Canada, T2N 1N4
> >
> > Tel.: ++1 (403) 210 7860
> > Email: swac...@ucalgary.ca
> >
> > --
> > Gromacs Users mailing list
> >
> > * Please search the archive at
> http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before
> posting!
> >
> > * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
> >
> > * For (un)subscribe requests visit
> > https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
> send a mail to gmx-users-requ...@gromacs.org.
>
>
> --
> Dr. Carsten Kutzner
> Max Planck Institute for Biophysical Chemistry
> Theoretical and Computational Biophysics
> Am Fassberg 11, 37077 Goettingen, Germany
> Tel. +49-551-2012313, Fax: +49-551-2012302
> http://www.mpibpc.mpg.de/grubmueller/kutzner
> http://www.mpibpc.mpg.de/grubmueller/sppexa
>
> --
> Gromacs Users mailing list
>
> * Please search the archive at
> http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before
> posting!
>
> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>
> * For (un)subscribe requests visit
> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
> send a mail to gmx-users-requ...@gromacs.org.
>
--
Gromacs Users mailing list
* Please search the archive at
http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before posting!
* Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
* For (un)subscribe requests visit
https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or send a
mail to gmx-users-requ...@gromacs.org.
--
Gromacs Users mailing list
* Please search the archive at
http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before posting!
* Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
* For (un)subscribe requests visit
https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or send a
mail to gmx-users-requ...@gromacs.org.
Re: [gmx-users] Installing gromacs 5.0.2 double precision with shared fftw libraries
So, when I compile in double precision it must be
> -DFFTW_INCLUDE_DIR=~/Programme/fftw3/double/include \
when it is single precision it should be
> -DFFTWF_INCLUDE_DIR=~/Programme/fftw3/single/include \
Always, these little details that make computation so much fun ;-)
--
Dr. Soeren Wacker
Centre For Molecular Simulation, BI # 547
The University of Calgary
2500 University Drive (NW)
Calgary, Alberta
Canada, T2N 1N4
Tel.: ++1 (403) 210 7860
Email: swac...@ucalgary.ca
From: gromacs.org_gmx-users-boun...@maillist.sys.kth.se
[gromacs.org_gmx-users-boun...@maillist.sys.kth.se] on behalf of Mark Abraham
[mark.j.abra...@gmail.com]
Sent: Wednesday, November 19, 2014 3:01 PM
To: Discussion list for GROMACS users
Subject: Re: [gmx-users] Installing gromacs 5.0.2 double precision with shared
fftw libraries
On Wed, Nov 19, 2014 at 7:25 PM, Soren Wacker wrote:
> Hi,
>
> I have trouble to install gromacs 5.0.2 in double precision with shared
> fftw libraries. During the cmake process I get:
>
> Could not find fftw3 library named libfftw3, please specify its location
> in CMAKE_PREFIX_PATH or FFTW_LIBRARY by hand (e.g.
> -DFFTW_LIBRARY='/path/to/libfftw3.so')
> CMake Error at cmake/gmxManageFFTLibraries.cmake:76 (MESSAGE):
> Cannot find FFTW 3 (with correct precision - libfftw3f for
> mixed-precision
> GROMACS or libfftw3 for double-precision GROMACS). Either choose the
> right
> precision, choose another FFT(W) library (-DGMX_FFT_LIBRARY), enable the
> advanced option to let GROMACS build FFTW 3 for you
> (-GMX_BUILD_OWN_FFTW=ON), or use the really slow GROMACS built-in fftpack
> library (-DGMX_FFT_LIBRARY=fftpack).
>
> However, I provided the path to libfftwr3.so:
>
> cmake ${SOURCE_DIR} \
> -DCMAKE_INSTALL_PREFIX=${INSTALL_DIR} \
> -DFFTWF_INCLUDE_DIR=~/Programme/fftw3/double/include \
> -DFFTWF_LIBRARY=~/Programme/fftw3/double/lib/libfftw3.so \
>
If you want to specify these paths, don't give the double-precision
libraries as values to the single-precision (*F*loating point) FFTW
variables. Use -DFFTW_INCLUDE_DIR= etc.
> -DCMAKE_PREFIX_PATH="/global/software/boost/gcc" \
>
Or avoid handling those naming problems yourself and use
-DCMAKE_PREFIX_PATH="/global/software/boost/gcc;$HOME/Programme/fftw3/double"
Mark
-DGMX_DOUBLE=on \
> -DGMX_CPU_ACCELERATION="SSE4.1"
>
> I also tried "-DFFTWF_LIBRARY=~/Programme/fftw3/double/lib/"
>
> The path is correct:
> >ls ~/Programme/fftw3/double/lib/libfftw3.so
> ~/Programme/fftw3/double/lib/libfftw3.so
>
> What is going wrong here?
>
> kind regards
> Soren
>
> --
> Dr. Soeren Wacker
> Centre For Molecular Simulation, BI # 547
> The University of Calgary
> 2500 University Drive (NW)
> Calgary, Alberta
> Canada, T2N 1N4
>
> Tel.: ++1 (403) 210 7860
> Email: swac...@ucalgary.ca
>
> --
> Gromacs Users mailing list
>
> * Please search the archive at
> http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before
> posting!
>
> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>
> * For (un)subscribe requests visit
> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
> send a mail to gmx-users-requ...@gromacs.org.
>
--
Gromacs Users mailing list
* Please search the archive at
http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before posting!
* Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
* For (un)subscribe requests visit
https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or send a
mail to gmx-users-requ...@gromacs.org.
--
Gromacs Users mailing list
* Please search the archive at
http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before posting!
* Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
* For (un)subscribe requests visit
https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or send a
mail to gmx-users-requ...@gromacs.org.
Re: [gmx-users] Installing gromacs 5.0.2 double precision with shared fftw libraries
Hi Carsten!
Like this??
-DCMAKE_PREFIX_PATH="/global/software/boost/gcc:~/Programme/fftw3/double/" \
or like this?
-DCMAKE_PREFIX_PATH="/global/software/boost/gcc" \
-DCMAKE_PREFIX_PATH="~/Programme/fftw3/double/" \
regards
Soren
--
Dr. Soeren Wacker
Centre For Molecular Simulation, BI # 547
The University of Calgary
2500 University Drive (NW)
Calgary, Alberta
Canada, T2N 1N4
Tel.: ++1 (403) 210 7860
Email: swac...@ucalgary.ca
From: gromacs.org_gmx-users-boun...@maillist.sys.kth.se
[gromacs.org_gmx-users-boun...@maillist.sys.kth.se] on behalf of Carsten
Kutzner [ckut...@gwdg.de]
Sent: Wednesday, November 19, 2014 11:43 AM
To: gmx-us...@gromacs.org
Subject: Re: [gmx-users] Installing gromacs 5.0.2 double precision with shared
fftw libraries
Hi Sören,
instead of providing FFTWF_INCLUDE_DIR and FFTWF_LIBRARY you
could also put ~/Programme/fftw3/double/ in CMAKE_PREFIX_PATH as well.
Viele Grüße!
Carsten
On 19 Nov 2014, at 19:25, Soren Wacker wrote:
> Hi,
>
> I have trouble to install gromacs 5.0.2 in double precision with shared fftw
> libraries. During the cmake process I get:
>
> Could not find fftw3 library named libfftw3, please specify its location in
> CMAKE_PREFIX_PATH or FFTW_LIBRARY by hand (e.g.
> -DFFTW_LIBRARY='/path/to/libfftw3.so')
> CMake Error at cmake/gmxManageFFTLibraries.cmake:76 (MESSAGE):
> Cannot find FFTW 3 (with correct precision - libfftw3f for mixed-precision
> GROMACS or libfftw3 for double-precision GROMACS). Either choose the right
> precision, choose another FFT(W) library (-DGMX_FFT_LIBRARY), enable the
> advanced option to let GROMACS build FFTW 3 for you
> (-GMX_BUILD_OWN_FFTW=ON), or use the really slow GROMACS built-in fftpack
> library (-DGMX_FFT_LIBRARY=fftpack).
>
> However, I provided the path to libfftwr3.so:
>
> cmake ${SOURCE_DIR} \
> -DCMAKE_INSTALL_PREFIX=${INSTALL_DIR} \
> -DFFTWF_INCLUDE_DIR=~/Programme/fftw3/double/include \
> -DFFTWF_LIBRARY=~/Programme/fftw3/double/lib/libfftw3.so \
> -DCMAKE_PREFIX_PATH="/global/software/boost/gcc" \
> -DGMX_DOUBLE=on \
> -DGMX_CPU_ACCELERATION="SSE4.1"
>
> I also tried "-DFFTWF_LIBRARY=~/Programme/fftw3/double/lib/"
>
> The path is correct:
>> ls ~/Programme/fftw3/double/lib/libfftw3.so
> ~/Programme/fftw3/double/lib/libfftw3.so
>
> What is going wrong here?
>
> kind regards
> Soren
>
> --
> Dr. Soeren Wacker
> Centre For Molecular Simulation, BI # 547
> The University of Calgary
> 2500 University Drive (NW)
> Calgary, Alberta
> Canada, T2N 1N4
>
> Tel.: ++1 (403) 210 7860
> Email: swac...@ucalgary.ca
>
> --
> Gromacs Users mailing list
>
> * Please search the archive at
> http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before posting!
>
> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>
> * For (un)subscribe requests visit
> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or send a
> mail to gmx-users-requ...@gromacs.org.
--
Dr. Carsten Kutzner
Max Planck Institute for Biophysical Chemistry
Theoretical and Computational Biophysics
Am Fassberg 11, 37077 Goettingen, Germany
Tel. +49-551-2012313, Fax: +49-551-2012302
http://www.mpibpc.mpg.de/grubmueller/kutzner
http://www.mpibpc.mpg.de/grubmueller/sppexa
--
Gromacs Users mailing list
* Please search the archive at
http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before posting!
* Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
* For (un)subscribe requests visit
https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or send a
mail to gmx-users-requ...@gromacs.org.
--
Gromacs Users mailing list
* Please search the archive at
http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before posting!
* Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
* For (un)subscribe requests visit
https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or send a
mail to gmx-users-requ...@gromacs.org.
Re: [gmx-users] Installing gromacs 5.0.2 single precision with CUDA and MPI ??
Hi,
another question:
Is it possible to install gromacs 5.0.2 with CUDA and MPI?
Since, when I try to generate mdrun_mpi everything seems to work fine until
>make install
then I get some cryptic error messages.
Some details first:
--
Intel(R) Xeon(R) CPU X5550 @ 2.67GHz
# Set up environment to work with the Intel 12.1.4 compilers
openmpi/1.6.5
intel/2011-sp1.10.319 (intel 12.1.4 or 12.1.0 compiler)
cuda/6.0(also tried cuda 5.5)
>mpicc -v
icc version 12.1.4 (gcc version 4.4.7 compatibility)
input commands:
CC=mpicc \
CXX=mpiCC \ (also tried mpicxx)
cmake ${SOURCE_DIR} \
-DCMAKE_INSTALL_PREFIX=${INSTALL_DIR} \
-DFFTWF_INCLUDE_DIR=$FFTWSINGLE/include \
-DFFTWF_LIBRARY=$FFTWSINGLE/lib/libfftw3f.a \
-DCMAKE_PREFIX_PATH="/global/software/boost/gcc" \
-DCMAKE_PREFIX_PATH="$FFTWSINGLE" \
-DGMX_BUILD_MDRUN_ONLY=ON \
-DGMX_MPI=on \
-DGMX_SIMD="SSE4.1"
Here the output:
-- Configuring done
-- Generating done
-- Build files have been written to:
/home/@USER/Install_Software/gromacs/builds/gromacs_50rc1_intel1214_sharedfftw3_openmp_sse41_mpi_single
Scanning dependencies of target git-version-info
Scanning dependencies of target mdrun_objlib
[ 0%] Generating git version information
[ 1%] [ 1%] [ 1%] Building CXX object
src/programs/CMakeFiles/mdrun_objlib.dir/mdrun/md.cpp.o
Building C object src/programs/CMakeFiles/mdrun_objlib.dir/mdrun/membed.c.o
Building CXX object src/programs/CMakeFiles/mdrun_objlib.dir/mdrun/repl_ex.cpp.o
warning #13000: could not open message catalog file: diagsc.cat
warning #13000: could not open message catalog file: diagsc.cat
warning #13000: could not open message catalog file: diagsc.cat
[ 1%] Built target git-version-info
[ 1%] Building CXX object
src/programs/CMakeFiles/mdrun_objlib.dir/mdrun/runner.cpp.o
warning #13000: could not open message catalog file: diagsc.cat
[ 1%] Building CXX object
src/programs/CMakeFiles/mdrun_objlib.dir/mdrun/mdrun.cpp.o
[ 1%] [ 1%] Building NVCC (Device) object
src/gromacs/CMakeFiles/libgromacs.dir/gmxlib/cuda_tools/./libgromacs_generated_cudautils.cu.o
Generating baseversion-gen.c
[ 1%] warning #13000: could not open message catalog file: diagsc.cat
Building NVCC (Device) object
src/gromacs/CMakeFiles/libgromacs.dir/gmxlib/cuda_tools/./libgromacs_generated_copyrite_gpu.cu.o
icc: command line warning #10006: ignoring unknown option '-dumpspecs'
icc: command line error: no files specified; for help type "icc -help"
CMake Error at libgromacs_generated_cudautils.cu.o.cmake:206 (message):
Error generating
/home/@USER/Install_Software/gromacs/builds/gromacs_50rc1_intel1214_sharedfftw3_openmp_sse41_mpi_single/src/gromacs/CMakeFiles/libgromacs.dir/gmxlib/cuda_tools/./libgromacs_generated_cudautils.cu.o
make[2]: ***
[src/gromacs/CMakeFiles/libgromacs.dir/gmxlib/cuda_tools/./libgromacs_generated_cudautils.cu.o]
Error 1
make[2]: *** Waiting for unfinished jobs
icc: command line warning #10006: ignoring unknown option '-dumpspecs'
icc: command line error: no files specified; for help type "icc -help"
CMake Error at libgromacs_generated_copyrite_gpu.cu.o.cmake:206 (message):
Error generating
/home/@USER/Install_Software/gromacs/builds/gromacs_50rc1_intel1214_sharedfftw3_openmp_sse41_mpi_single/src/gromacs/CMakeFiles/libgromacs.dir/gmxlib/cuda_tools/./libgromacs_generated_copyrite_gpu.cu.o
make[2]: ***
[src/gromacs/CMakeFiles/libgromacs.dir/gmxlib/cuda_tools/./libgromacs_generated_copyrite_gpu.cu.o]
Error 1
make[1]: *** [src/gromacs/CMakeFiles/libgromacs.dir/all] Error 2
make[1]: *** Waiting for unfinished jobs
[ 1%] Built target mdrun_objlib
make: *** [all] Error 2
[ 0%] Generating git version information
[ 0%] Built target git-version-info
[ 0%] Building NVCC (Device) object
src/gromacs/CMakeFiles/libgromacs.dir/gmxlib/cuda_tools/./libgromacs_generated_cudautils.cu.o
icc: command line warning #10006: ignoring unknown option '-dumpspecs'
icc: command line error: no files specified; for help type "icc -help"
CMake Error at libgromacs_generated_cudautils.cu.o.cmake:206 (message):
Error generating
/home/@USER/Install_Software/gromacs/builds/gromacs_50rc1_intel1214_sharedfftw3_openmp_sse41_mpi_single/src/gromacs/CMakeFiles/libgromacs.dir/gmxlib/cuda_tools/./libgromacs_generated_cudautils.cu.o
make[2]: ***
[src/gromacs/CMakeFiles/libgromacs.dir/gmxlib/cuda_tools/./libgromacs_generated_cudautils.cu.o]
Error 1
make[1]: *** [src/gromacs/CMakeFiles/libgromacs.dir/all] Error 2
make: *** [all] Error 2
Finished installing MPI single-precision version of mdrun at Wed Nov 19
15:34:18 MST 2014
--
Gromacs Users mailing list
* Please search the archive at
http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before posting!
* Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
* For (un)subscribe requests visit
https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or send a
mail to gmx-users-requ...@gromacs.org.
Re: [gmx-users] Installing gromacs 5.0.2 single precision with CUDA and MPI ??
Which gromacs version would you recommend then??
Soren
From: gromacs.org_gmx-users-boun...@maillist.sys.kth.se
[gromacs.org_gmx-users-boun...@maillist.sys.kth.se] on behalf of Mark Abraham
[mark.j.abra...@gmail.com]
Sent: Wednesday, November 19, 2014 4:43 PM
To: Discussion list for GROMACS users
Subject: Re: [gmx-users] Installing gromacs 5.0.2 single precision with CUDA
and MPI ??
On Thu, Nov 20, 2014 at 12:08 AM, Soren Wacker wrote:
> Hi,
>
> another question:
> Is it possible to install gromacs 5.0.2 with CUDA and MPI?
>
Yes, but you should use a compiler version that is supported by your CUDA
version, and not try to install release candidate versions Gromacs. ;-)
Mark
> Since, when I try to generate mdrun_mpi everything seems to work fine until
> >make install
>
> then I get some cryptic error messages.
>
> Some details first:
> --
> Intel(R) Xeon(R) CPU X5550 @ 2.67GHz
> # Set up environment to work with the Intel 12.1.4 compilers
>
> openmpi/1.6.5
> intel/2011-sp1.10.319 (intel 12.1.4 or 12.1.0 compiler)
> cuda/6.0(also tried cuda 5.5)
>
> >mpicc -v
> icc version 12.1.4 (gcc version 4.4.7 compatibility)
>
> input commands:
> CC=mpicc \
> CXX=mpiCC \ (also tried mpicxx)
> cmake ${SOURCE_DIR} \
> -DCMAKE_INSTALL_PREFIX=${INSTALL_DIR} \
> -DFFTWF_INCLUDE_DIR=$FFTWSINGLE/include \
> -DFFTWF_LIBRARY=$FFTWSINGLE/lib/libfftw3f.a \
> -DCMAKE_PREFIX_PATH="/global/software/boost/gcc" \
> -DCMAKE_PREFIX_PATH="$FFTWSINGLE" \
> -DGMX_BUILD_MDRUN_ONLY=ON \
> -DGMX_MPI=on \
> -DGMX_SIMD="SSE4.1"
>
>
>
>
>
> Here the output:
>
> -- Configuring done
> -- Generating done
> -- Build files have been written to:
> /home/@USER/Install_Software/gromacs/builds/gromacs_50rc1_intel1214_sharedfftw3_openmp_sse41_mpi_single
> Scanning dependencies of target git-version-info
> Scanning dependencies of target mdrun_objlib
> [ 0%] Generating git version information
> [ 1%] [ 1%] [ 1%] Building CXX object
> src/programs/CMakeFiles/mdrun_objlib.dir/mdrun/md.cpp.o
> Building C object src/programs/CMakeFiles/mdrun_objlib.dir/mdrun/membed.c.o
> Building CXX object
> src/programs/CMakeFiles/mdrun_objlib.dir/mdrun/repl_ex.cpp.o
> warning #13000: could not open message catalog file: diagsc.cat
> warning #13000: could not open message catalog file: diagsc.cat
> warning #13000: could not open message catalog file: diagsc.cat
> [ 1%] Built target git-version-info
> [ 1%] Building CXX object
> src/programs/CMakeFiles/mdrun_objlib.dir/mdrun/runner.cpp.o
> warning #13000: could not open message catalog file: diagsc.cat
> [ 1%] Building CXX object
> src/programs/CMakeFiles/mdrun_objlib.dir/mdrun/mdrun.cpp.o
> [ 1%] [ 1%] Building NVCC (Device) object
> src/gromacs/CMakeFiles/libgromacs.dir/gmxlib/cuda_tools/./libgromacs_generated_cudautils.cu.o
> Generating baseversion-gen.c
> [ 1%] warning #13000: could not open message catalog file: diagsc.cat
> Building NVCC (Device) object
> src/gromacs/CMakeFiles/libgromacs.dir/gmxlib/cuda_tools/./libgromacs_generated_copyrite_gpu.cu.o
> icc: command line warning #10006: ignoring unknown option '-dumpspecs'
> icc: command line error: no files specified; for help type "icc -help"
> CMake Error at libgromacs_generated_cudautils.cu.o.cmake:206 (message):
> Error generating
>
> /home/@USER/Install_Software/gromacs/builds/gromacs_50rc1_intel1214_sharedfftw3_openmp_sse41_mpi_single/src/gromacs/CMakeFiles/libgromacs.dir/gmxlib/cuda_tools/./libgromacs_generated_cudautils.cu.o
>
>
> make[2]: ***
> [src/gromacs/CMakeFiles/libgromacs.dir/gmxlib/cuda_tools/./libgromacs_generated_cudautils.cu.o]
> Error 1
> make[2]: *** Waiting for unfinished jobs
> icc: command line warning #10006: ignoring unknown option '-dumpspecs'
> icc: command line error: no files specified; for help type "icc -help"
> CMake Error at libgromacs_generated_copyrite_gpu.cu.o.cmake:206 (message):
> Error generating
>
> /home/@USER/Install_Software/gromacs/builds/gromacs_50rc1_intel1214_sharedfftw3_openmp_sse41_mpi_single/src/gromacs/CMakeFiles/libgromacs.dir/gmxlib/cuda_tools/./libgromacs_generated_copyrite_gpu.cu.o
>
>
> make[2]: ***
> [src/gromacs/CMakeFiles/libgromacs.dir/gmxlib/cuda_tools/./libgromacs_generated_copyrite_gpu.cu.o]
> Error 1
> make[1]: *** [src/gromacs/CMakeFiles/libgromacs.dir/all] Error 2
> make[1]: *** Waiting for unfinished jobs
> [ 1%] Built target mdrun_objlib
> make: *** [all] Error 2
> [ 0%] Generating git version information
> [ 0%] Built target git-version-info
> [ 0%] Building NVCC (Device) object
> src/groma
Re: [gmx-users] Installing gromacs 5.0.2 single precision with CUDA and MPI ??
I tried both, with the same result. I can compile gromacs 5.0.2 as non-MPI
version using cuda 5.5 and intel 12.1.4.
But, it is not working together with MPI 1.6.5 nor 1.6.2
Is there anything else to consider when compiling with cuda and mpi?
The error messages do not tell me much:
icc: command line warning #10006: ignoring unknown option '-dumpspecs'
icc: command line warning #10006: ignoring unknown option '-dumpspecs'
icc: command line error: no files specified; for help type "icc -help"
icc: command line error: no files specified; for help type "icc -help"
CMake Error at cuda_tools_generated_cudautils.cu.o.cmake:206 (message):
Error generating
$HOME/Install_Software/gromacs-5.0.2/builds/gromacs_5p0p2_intel1214_sharedfftw3_openmp_sse41_mpi_single/src/gromacs/gmxlib/cuda_tools/CMakeFiles/cuda_tools.dir//./cuda_tools_generated_cudautils.cu.o
Thanks
Soren
From: gromacs.org_gmx-users-boun...@maillist.sys.kth.se
[gromacs.org_gmx-users-boun...@maillist.sys.kth.se] on behalf of Justin Lemkul
[jalem...@vt.edu]
Sent: Wednesday, November 19, 2014 6:08 PM
To: gmx-us...@gromacs.org
Subject: Re: [gmx-users] Installing gromacs 5.0.2 single precision with CUDA
and MPI ??
On 11/19/14 8:01 PM, Soren Wacker wrote:
> Which gromacs version would you recommend then??
Are you installing 5.0.2, as the subject line states, or 5.0-rc1, as your
install directory name implies? If you're not actually installing 5.0.2, that's
the one you should be installing.
-Justin
> Soren
>
>
>
> From: gromacs.org_gmx-users-boun...@maillist.sys.kth.se
> [gromacs.org_gmx-users-boun...@maillist.sys.kth.se] on behalf of Mark Abraham
> [mark.j.abra...@gmail.com]
> Sent: Wednesday, November 19, 2014 4:43 PM
> To: Discussion list for GROMACS users
> Subject: Re: [gmx-users] Installing gromacs 5.0.2 single precision with CUDA
> and MPI ??
>
> On Thu, Nov 20, 2014 at 12:08 AM, Soren Wacker wrote:
>
>> Hi,
>>
>> another question:
>> Is it possible to install gromacs 5.0.2 with CUDA and MPI?
>>
>
> Yes, but you should use a compiler version that is supported by your CUDA
> version, and not try to install release candidate versions Gromacs. ;-)
>
> Mark
>
>
>> Since, when I try to generate mdrun_mpi everything seems to work fine until
>>> make install
>>
>> then I get some cryptic error messages.
>>
>> Some details first:
>> --
>> Intel(R) Xeon(R) CPU X5550 @ 2.67GHz
>> # Set up environment to work with the Intel 12.1.4 compilers
>>
>> openmpi/1.6.5
>> intel/2011-sp1.10.319 (intel 12.1.4 or 12.1.0 compiler)
>> cuda/6.0(also tried cuda 5.5)
>>
>>> mpicc -v
>> icc version 12.1.4 (gcc version 4.4.7 compatibility)
>>
>> input commands:
>> CC=mpicc \
>> CXX=mpiCC \ (also tried mpicxx)
>> cmake ${SOURCE_DIR} \
>> -DCMAKE_INSTALL_PREFIX=${INSTALL_DIR} \
>> -DFFTWF_INCLUDE_DIR=$FFTWSINGLE/include \
>> -DFFTWF_LIBRARY=$FFTWSINGLE/lib/libfftw3f.a \
>> -DCMAKE_PREFIX_PATH="/global/software/boost/gcc" \
>> -DCMAKE_PREFIX_PATH="$FFTWSINGLE" \
>> -DGMX_BUILD_MDRUN_ONLY=ON \
>> -DGMX_MPI=on \
>> -DGMX_SIMD="SSE4.1"
>>
>>
>>
>>
>>
>> Here the output:
>>
>> -- Configuring done
>> -- Generating done
>> -- Build files have been written to:
>> /home/@USER/Install_Software/gromacs/builds/gromacs_50rc1_intel1214_sharedfftw3_openmp_sse41_mpi_single
>> Scanning dependencies of target git-version-info
>> Scanning dependencies of target mdrun_objlib
>> [ 0%] Generating git version information
>> [ 1%] [ 1%] [ 1%] Building CXX object
>> src/programs/CMakeFiles/mdrun_objlib.dir/mdrun/md.cpp.o
>> Building C object src/programs/CMakeFiles/mdrun_objlib.dir/mdrun/membed.c.o
>> Building CXX object
>> src/programs/CMakeFiles/mdrun_objlib.dir/mdrun/repl_ex.cpp.o
>> warning #13000: could not open message catalog file: diagsc.cat
>> warning #13000: could not open message catalog file: diagsc.cat
>> warning #13000: could not open message catalog file: diagsc.cat
>> [ 1%] Built target git-version-info
>> [ 1%] Building CXX object
>> src/programs/CMakeFiles/mdrun_objlib.dir/mdrun/runner.cpp.o
>> warning #13000: could not open message catalog file: diagsc.cat
>> [ 1%] Building CXX object
>> src/programs/CMakeFiles/mdrun_objlib.dir/mdrun/mdrun.cpp.o
>> [ 1%] [ 1%] Building NVCC (Device) object
>> src/gromacs/CMakeFiles/libgromacs.dir/gmxlib/cuda_tools/./libgromacs_generated_cudautils.cu.o
Re: [gmx-users] Installing gromacs 5.0.2 single precision with CUDA and MPI ??
Thanks Mark!
It works using the GNU compiler instead of the intel compiler.
Now everything works.
Soren
From: gromacs.org_gmx-users-boun...@maillist.sys.kth.se
[gromacs.org_gmx-users-boun...@maillist.sys.kth.se] on behalf of Mark Abraham
[mark.j.abra...@gmail.com]
Sent: Friday, November 21, 2014 2:22 AM
To: Discussion list for GROMACS users
Subject: Re: [gmx-users] Installing gromacs 5.0.2 single precision with CUDA
and MPI ??
On Fri, Nov 21, 2014 at 2:32 AM, Soren Wacker wrote:
> I tried both, with the same result. I can compile gromacs 5.0.2 as non-MPI
> version using cuda 5.5 and intel 12.1.4.
> But, it is not working together with MPI 1.6.5 nor 1.6.2
>
> Is there anything else to consider when compiling with cuda and mpi?
>
Not on the Gromacs end. If your tools are set up correctly then things will
work. First, get a test CUDA+MPI program to work - consult your sysadmins.
You will likely get better performance and have fewer troubles with gcc,
anyway.
Mark
The error messages do not tell me much:
>
> icc: command line warning #10006: ignoring unknown option '-dumpspecs'
> icc: command line warning #10006: ignoring unknown option '-dumpspecs'
> icc: command line error: no files specified; for help type "icc -help"
> icc: command line error: no files specified; for help type "icc -help"
> CMake Error at cuda_tools_generated_cudautils.cu.o.cmake:206 (message):
> Error generating
>
> $HOME/Install_Software/gromacs-5.0.2/builds/gromacs_5p0p2_intel1214_sharedfftw3_openmp_sse41_mpi_single/src/gromacs/gmxlib/cuda_tools/CMakeFiles/cuda_tools.dir//./cuda_tools_generated_cudautils.cu.o
>
>
> Thanks
> Soren
>
>
> From: gromacs.org_gmx-users-boun...@maillist.sys.kth.se [
> gromacs.org_gmx-users-boun...@maillist.sys.kth.se] on behalf of Justin
> Lemkul [jalem...@vt.edu]
> Sent: Wednesday, November 19, 2014 6:08 PM
> To: gmx-us...@gromacs.org
> Subject: Re: [gmx-users] Installing gromacs 5.0.2 single precision with
> CUDA and MPI ??
>
> On 11/19/14 8:01 PM, Soren Wacker wrote:
> > Which gromacs version would you recommend then??
>
> Are you installing 5.0.2, as the subject line states, or 5.0-rc1, as your
> install directory name implies? If you're not actually installing 5.0.2,
> that's
> the one you should be installing.
>
> -Justin
>
> > Soren
> >
> >
> >
> > From: gromacs.org_gmx-users-boun...@maillist.sys.kth.se [
> gromacs.org_gmx-users-boun...@maillist.sys.kth.se] on behalf of Mark
> Abraham [mark.j.abra...@gmail.com]
> > Sent: Wednesday, November 19, 2014 4:43 PM
> > To: Discussion list for GROMACS users
> > Subject: Re: [gmx-users] Installing gromacs 5.0.2 single precision with
> CUDA and MPI ??
> >
> > On Thu, Nov 20, 2014 at 12:08 AM, Soren Wacker
> wrote:
> >
> >> Hi,
> >>
> >> another question:
> >> Is it possible to install gromacs 5.0.2 with CUDA and MPI?
> >>
> >
> > Yes, but you should use a compiler version that is supported by your CUDA
> > version, and not try to install release candidate versions Gromacs. ;-)
> >
> > Mark
> >
> >
> >> Since, when I try to generate mdrun_mpi everything seems to work fine
> until
> >>> make install
> >>
> >> then I get some cryptic error messages.
> >>
> >> Some details first:
> >> --
> >> Intel(R) Xeon(R) CPU X5550 @ 2.67GHz
> >> # Set up environment to work with the Intel 12.1.4 compilers
> >>
> >> openmpi/1.6.5
> >> intel/2011-sp1.10.319 (intel 12.1.4 or 12.1.0 compiler)
> >> cuda/6.0(also tried cuda 5.5)
> >>
> >>> mpicc -v
> >> icc version 12.1.4 (gcc version 4.4.7 compatibility)
> >>
> >> input commands:
> >> CC=mpicc \
> >> CXX=mpiCC \ (also tried mpicxx)
> >> cmake ${SOURCE_DIR} \
> >> -DCMAKE_INSTALL_PREFIX=${INSTALL_DIR} \
> >> -DFFTWF_INCLUDE_DIR=$FFTWSINGLE/include \
> >> -DFFTWF_LIBRARY=$FFTWSINGLE/lib/libfftw3f.a \
> >> -DCMAKE_PREFIX_PATH="/global/software/boost/gcc" \
> >> -DCMAKE_PREFIX_PATH="$FFTWSINGLE" \
> >> -DGMX_BUILD_MDRUN_ONLY=ON \
> >> -DGMX_MPI=on \
> >> -DGMX_SIMD="SSE4.1"
> >>
> >>
> >>
> >>
> >>
> >> Here the output:
> >>
> >> -- Configuring done
> >> -- Generating done
> >> -- Build files have been written to:
> >>
[gmx-users] mdrun_mpi with cmake
Hi, I would like to generate specifically mdrun with mpi support. with configure/make it was possible to use configure ... make mdrun make install mdrun How would you do this with cmake? best regards Soren -- Gromacs Users mailing list * Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before posting! * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists * For (un)subscribe requests visit https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or send a mail to gmx-users-requ...@gromacs.org.
Re: [gmx-users] mdrun_mpi with cmake
Thanks! That means you compile everything with cmake make #and only at the last step the build is limited to mdrun? make-mdrun Do you already know how you will turn on an 'mdrun-only build' ? best Soren -- Gromacs Users mailing list * Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before posting! * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists * For (un)subscribe requests visit https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or send a mail to gmx-users-requ...@gromacs.org.
Re: [gmx-users] mdrun_mpi with cmake with GMX_BUILD_MDRUN_ONLY=ON
Ok, actually, both solutions generated mdrun_mpi. Would you recommend to set a suffix for the libraries as well? @Justin: I've got > Manually-specified variables were not used by the project: > >GMX_BUILD_MDRUN_ONLY Does this mean the variable is not used at all, or just not here? Or should I set it as environment variable??? Is this the solution for gromacs 5? I am compiling 4.6.5. cmake command was: cmake .. -DFFTWF_INCLUDE_DIR=/home/swacker/Programme/fftw3/single/include -DFFTWF_LIBRARY=~/Programme/fftw3/single/lib/libfftw3f.a -DCMAKE_INSTALL_PREFIX=~/Programme/gromacs/gromacs-4.6.5-gcc-fftw3-single -DGMX_MPI=ON -DGMX_BINARY_SUFFIX=_mpi -DGMX_DEFAULT_SUFFIX=OFF -DGMX_BUILD_MDRUN_ONLY=ON From: gromacs.org_gmx-users-boun...@maillist.sys.kth.se [gromacs.org_gmx-users-boun...@maillist.sys.kth.se] on behalf of Justin Lemkul [jalem...@vt.edu] Sent: Thursday, May 22, 2014 4:38 PM To: gmx-us...@gromacs.org Subject: Re: [gmx-users] mdrun_mpi with cmake On 5/22/14, 6:36 PM, Soren Wacker wrote: > Thanks! > > That means you compile everything with > > cmake > make > #and only at the last step the build is limited to mdrun? > make-mdrun > > Do you already know how you will turn on an 'mdrun-only build' ? > No, use cmake -DGMX_BUILD_MDRUN_ONLY=ON -Justin -- == Justin A. Lemkul, Ph.D. Ruth L. Kirschstein NRSA Postdoctoral Fellow Department of Pharmaceutical Sciences School of Pharmacy Health Sciences Facility II, Room 601 University of Maryland, Baltimore 20 Penn St. Baltimore, MD 21201 jalem...@outerbanks.umaryland.edu | (410) 706-7441 http://mackerell.umaryland.edu/~jalemkul == -- Gromacs Users mailing list * Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before posting! * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists * For (un)subscribe requests visit https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or send a mail to gmx-users-requ...@gromacs.org. -- Gromacs Users mailing list * Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before posting! * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists * For (un)subscribe requests visit https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or send a mail to gmx-users-requ...@gromacs.org.
Re: [gmx-users] mdrun_mpi with cmake with GMX_BUILD_MDRUN_ONLY=ON
...and apparently everything gets compiled. -- Gromacs Users mailing list * Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before posting! * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists * For (un)subscribe requests visit https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or send a mail to gmx-users-requ...@gromacs.org.
Re: [gmx-users] mdrun_mpi with cmake with GMX_BUILD_MDRUN_ONLY=ON
Thanks! -- Gromacs Users mailing list * Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before posting! * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists * For (un)subscribe requests visit https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or send a mail to gmx-users-requ...@gromacs.org.
Re: [gmx-users] PRODRG
Hi, I think, you should avoid the PRODRG server. If you use it, take care with the validation of you parameters. Especially, the charges. There are different tools which can generate topologies and pdb files. To could use RDkit to generate coordinates save it as SDF convert it to PDB with e.g. OpenBabel. Other useful tools would you may consider are MarvinSketch and Antechamber (Ambertools). Which conformer to use depends on which conformer/s you are interested in. best Soren -- Dr. Soeren Wacker Centre For Molecular Simulation, BI # 547 The University of Calgary 2500 University Drive (NW) Calgary, Alberta Canada, T2N 1N4 Tel.: ++1 (403) 210 7860 Email: swac...@ucalgary.ca From: gromacs.org_gmx-users-boun...@maillist.sys.kth.se [gromacs.org_gmx-users-boun...@maillist.sys.kth.se] on behalf of Negar Parvizi [negar.parv...@yahoo.com] Sent: Monday, May 26, 2014 11:00 AM To: gromacs.org_gmx-users@maillist.sys.kth.se Subject: [gmx-users] PRODRG Dear all users, I am using Protein-Ligand Complex tutorial (Justin Lemkul tutorial). For making .pdb file of ligand, we should use PRODRG(draw molecule with JME Editor). Here is my question that when i draw my ligand in JME Editor, which conformer should i select ? because different conformer gives different PRODRG output files. Thanks in advance, N.Parvizi -- Gromacs Users mailing list * Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before posting! * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists * For (un)subscribe requests visit https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or send a mail to gmx-users-requ...@gromacs.org. -- Gromacs Users mailing list * Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before posting! * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists * For (un)subscribe requests visit https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or send a mail to gmx-users-requ...@gromacs.org.
[gmx-users] tunepme does not start
Hi, I try to use mdrun with -tunepme. When I start it at a clusters login node everything works fine. However, when I sent it to a cluster it seems -tunepme gets ignored. mdrun is started and runs the tpr without tuning the number of PME nodes. I attach an example script and a log file. Any idea why this might happen? best regards Soren -- Dr. Soeren Wacker Centre For Molecular Simulation, BI # 547 The University of Calgary 2500 University Drive (NW) Calgary, Alberta Canada, T2N 1N4 Tel.: ++1 (403) 210 7860 Email: swac...@ucalgary.ca From: gromacs.org_gmx-users-boun...@maillist.sys.kth.se [gromacs.org_gmx-users-boun...@maillist.sys.kth.se] on behalf of Chetan Mahajan [chetanv...@gmail.com] Sent: Tuesday, June 17, 2014 3:20 PM To: gmx-us...@gromacs.org Subject: Re: [gmx-users] peculiar water behavior Thanks, Justin. That makes it clear. I have one small question: I do not know how does position restraint algorithm work with other things in gromacs, but doesn't it generate a lot of stress due to slab moving from center to near box boundary despite of position restraints, when -com option is used for refcoord-scaling? If there is a lot of stress, would this stress affect simulation results? Thanks and Regards, Chetan On Tue, Jun 17, 2014 at 10:01 AM, Justin Lemkul wrote: > > > On 6/17/14, 10:32 AM, Chetan Mahajan wrote: > >> Any ideas : >> >> 1. why slab is moving despite of position constraints? >> >> > Restraints are biasing potentials; they don't prevent movement. Why is it > necessary that your slab not move? > > > 2. That too, only with "com" option of refcoord scaling...with "all" >> option >> slab does not move... >> >> > Using "all" means the restraint reference for each atom is the original > position in the input coordinates, with respect to every atom individually. > It is more stringent, though it can often be unstable. The "com" option > means the restraint potential is calculated with respect to the COM of the > group of restrained atoms; it is a bit more forgiving, since the COM might > only move slightly, even if there are relatively larger changes in the > individual atomic positions. > > -Justin > > Thanks >> >> >> On Sun, Jun 15, 2014 at 7:02 PM, Chetan Mahajan >> wrote: >> >> The pics should be seen in the reverse order. Due to sorting by name, >>> "After equilibration" pic occurs first. >>> >>> >>> On Sun, Jun 15, 2014 at 6:54 PM, Chetan Mahajan >>> wrote: >>> >>> Thanks. No, same things do not happen when it's a NVT simulation or no mixed potentials. Also, with mixed potentials in NPT simulation, this things of slab moving in one direction (thus following water movement) happens only with -com option of refcoord-scaling, *not with -all option for refcoord-scaling. * I have a crystal slab with equal amount of water on both the sides. During NPT equilibration (with refcoord-scaling option of -com, not -all), some water on side side of the slab enters opposite side due to periodic boundary conditions I have (say water on the right side of the slab, crosses boundary of the box and enter the left side of the slab due to periodic boundaries. Even crystal slab follows this water and move same distance in the same direction of water movement. Following link shares pictures of the solvated slab before and after equilibration: https://www.dropbox.com/sh/tjq80jgnff1y7ur/AAAzhnv8uwl-UMQ0TYx8PuPAa I feel, this is not an error, but it is odd. Due to periodicity we have, output of any calculations will remain the same. Again, with -all option for refcoord scaling, this movement of water and slab DOES NOT occur. Thanks a lot! regards Chetan On Sat, Jun 14, 2014 at 4:40 AM, Mark Abraham >>> > wrote: Hi, > > Repeating myself - simplify things! Do sane things happen with NVT? > With > no > mixed potentials? > > Your words aren't making sense to me - you'll do much better with > pictures, > making sure you turn on your visualizer's option to show the periodic > box. > > Mark > > > On Fri, Jun 13, 2014 at 8:45 PM, Chetan Mahajan > wrote: > > Hi Justin, >> >> I, actually, do not want to preserve any voids. Initial void that I >> mentioned is supposed to be filled by water during equilibration and >> it >> does so. However, problem is that some or major quantity of water >> > moves out > >> from one side of the box to enter from the opposite side (obeying >> > periodic > >> boundary conditions) and crystal slab moving following water movement. >> > Thus > >> although initially there is almost equal amount of water on both sides >> > of > >> the slab, we have more water on one side of the slab than the
