Hello,
It is possible in theory to compile Amber11 and AmberTools with ifort and
gcc, but you have to figure out how to properly link gcc-created object
files with ifort-created object files (I've never tried myself). It's
easier with all GNU compilers or all intel compilers because that
compatibility is well-documented. The mix-and-match is a little harder to
set up, but I'm sure with a little google-magic you can get it figured out.
Easiest thing to do would just be install icc OR use just the gnu compilers,
following Paul's advice.
Good luck!
Jason
On Fri, Sep 24, 2010 at 12:46 AM, Paul S. Nerenberg <psn.berkeley.edu>wrote:
> Ah...I see now why you compiled AmberTools with the gnu compilers. In
> that case, you've got two options:
>
> (1) If you have easy access to the intel compilers, you could install
> icc. You would still need to re-build the MPICH2 installation,
> because you need mpicc (and mpif90), but you could also download
> OpenMPI and go from there (see below for more on that). Then you
> would build AT and AMBER from scratch using the intel compilers.
>
> (2) You probably have the full set of gnu compilers (i.e., gcc, g++,
> and gfortran) if you were able to compile the serial versions of AT
> and AMBER. You could download OpenMPI and either build it with the
> configure_openmpi script (since 1.4.2 is the latest version, you need
> to edit the configure_openmpi script to look for that version number)
> or just build it from scratch. Either way make sure that you put the
> path for the OpenMPI installation after the path to the MPICH2
> installation so that "which mpicc" and "which mpif90" are both
> pointing to the OpenMPI installation and *not* MPICH2. Then build
> parallel AMBER with the gnu compilers.
>
> I realize that these are somewhat barebones instructions for how to do
> things, but I think it's the best I can do from afar. Hopefully it's
> enough to get you started and/or you can find a (local) friend to help
> you out...
>
> Best,
>
> Paul
>
>
> On Sep 23, 2010, at 9:03 PM, Sangita Kachhap wrote:
>
> > Thanks for reply
> >
> > which ifort gave following output:
> >
> > /intel/Compiler/11.1/072/bin/intel64/ifort
> >
> >
> > which icc gave following output:
> >
> > /usr/bin/which: no icc in
> > (/home/sangita/intel/Compiler/11.1/072/bin/intel64:/usr/kerberos/
> > bin:/usr/local/bin:/usr/bin:/bin
> > :/usr/X11R6/bin:/home/sangita/bin:/home/sangita/mpich2-install/bin:/
> > home/sangita/amber11/bin
> > :/home/sangita:/usr/local/bin/vmd:/home/sangita/intel/Compiler/
> > 11.1/072/bin/intel64/ifortvars_intel64.sh
> > :/usr/X11R6/lib/libXt.so.6)
> >
> > That means icc is not insatalled in my system.
> > Can I install without icc because I have only ifort or I have to
> > install icc
> > first then start compilation?
> >
> >
> >> Hi Sangita,
> >>
> >> The error you got when you tried building in parallel suggests that
> >> your system has the intel compilers. Moreover, the output of your
> >> "mpif90 -show" command clearly shows that the intel compilers are the
> >> ones linked to your mpich2 installation, *not* the gnu compilers.
> >> You
> >> need to be consistent about which compilers you are using, so I would
> >> suggest doing a "make clean" everywhere, and then configuring/
> >> building
> >> everything from the beginning using the intel compilers. (In
> >> addition
> >> these compilers usually generate faster code than the gnu compilers,
> >> so that's another reason to use them.)
> >>
> >> Just to be sure we know what you've got, can you also type "which
> >> icc"
> >> and "which ifort" (and maybe "icc -v" and "ifort -v" just to be very
> >> thorough)?
> >>
> >> Paul
> >>
> >>
> >> On Sep 23, 2010, at 10:11 AM, Sangita Kachhap wrote:
> >>
> >>>
> >>> Thanks for reply
> >>>
> >>> Then should I configure AmberTool-1.4 with intel compiler? Since I
> >>> have configured
> >>> it with gnu compiler.
> >>>
> >>>> Sangita,
> >>>>
> >>>> You need to type "./configure_openmpi intel" -- or whatever
> >>>> compiler
> >>>> you are using (looks like intel though). The OpenMPI configuration
> >>>> script is much like the AMBER configure script...
> >>>>
> >>>> Best,
> >>>>
> >>>> Paul
> >>>>
> >>>>
> >>>> On Sep 23, 2010, at 9:50 AM, Sangita Kachhap wrote:
> >>>>
> >>>>>
> >>>>> Thanks to quick reply
> >>>>> sorry for incomplete mail
> >>>>>
> >>>>> Error which I have got was :
> >>>>>
> >>>>> [sangita.localhost src]$ make parallel > amber-6.log &
> >>>>> [1] 8628
> >>>>> [sangita.localhost src]$ ifort: command line warning #10006:
> >>>>> ignoring unknown
> >>>>> option '-ffree-form'
> >>>>> : error #5149: Illegal character in statement label field [m]
> >>>>>
> >>>>> : error #5149: Illegal character in statement label field [o]
> >>>>>
> >>>>> : error #5149: Illegal character in statement label field [d]
> >>>>>
> >>>>> : error #5149: Illegal character in statement label field [u]
> >>>>>
> >>>>> : error #5149: Illegal character in statement label field [l]
> >>>>>
> >>>>> : error #5118: First statement in file must not be continued
> >>>>>
> >>>>> _constants.f(223): error #5149: Illegal character in statement
> >>>>> label
> >>>>> field [i]
> >>>>> integer, parameter :: RETIRED_INPUT_OPTION = -10301 ! first 5
> >>>>> digit
> >>>>> palindromic
> >>>>> prime
> >>>>> ^
> >>>>> _constants.f(223): error #5149: Illegal character in statement
> >>>>> label
> >>>>> field [n]
> >>>>> integer, parameter :: RETIRED_INPUT_OPTION = -10301 ! first 5
> >>>>> digit
> >>>>> palindromic
> >>>>> prime
> >>>>> -^
> >>>>> _constants.f(223): error #5149: Illegal character in statement
> >>>>> label
> >>>>> field [t]
> >>>>> integer, parameter :: RETIRED_INPUT_OPTION = -10301 ! first 5
> >>>>> digit
> >>>>> palindromic
> >>>>> prime
> >>>>> --^
> >>>>> _constants.f(223): error #5149: Illegal character in statement
> >>>>> label
> >>>>> field [e]
> >>>>> integer, parameter :: RETIRED_INPUT_OPTION = -10301 ! first 5
> >>>>> digit
> >>>>> palindromic
> >>>>> prime
> >>>>> ---^
> >>>>> _constants.f(223): error #5149: Illegal character in statement
> >>>>> label
> >>>>> field [g]
> >>>>> integer, parameter :: RETIRED_INPUT_OPTION = -10301 ! first 5
> >>>>> digit
> >>>>> palindromic
> >>>>> prime
> >>>>> ----^
> >>>>> _constants.f(224): error #5149: Illegal character in statement
> >>>>> label
> >>>>> field [i]
> >>>>> integer, parameter :: NO_INPUT_VALUE = 12344321 ! from Bob Duke
> >>>>> ^
> >>>>> _constants.f(224): error #5149: Illegal character in statement
> >>>>> label
> >>>>> field [n]
> >>>>> integer, parameter :: NO_INPUT_VALUE = 12344321 ! from Bob Duke
> >>>>> -^
> >>>>> _constants.f(224): error #5149: Illegal character in statement
> >>>>> label
> >>>>> field [t]
> >>>>> integer, parameter :: NO_INPUT_VALUE = 12344321 ! from Bob Duke
> >>>>> --^
> >>>>> _constants.f(224): error #5149: Illegal character in statement
> >>>>> label
> >>>>> field [e]
> >>>>> integer, parameter :: NO_INPUT_VALUE = 12344321 ! from Bob Duke
> >>>>> ---^
> >>>>> _constants.f(224): error #5149: Illegal character in statement
> >>>>> label
> >>>>> field [g]
> >>>>> integer, parameter :: NO_INPUT_VALUE = 12344321 ! from Bob Duke
> >>>>> ----^
> >>>>> _constants.f(226): error #5149: Illegal character in statement
> >>>>> label
> >>>>> field [e]
> >>>>> end module constants
> >>>>> ^
> >>>>> _constants.f(226): error #5149: Illegal character in statement
> >>>>> label
> >>>>> field [n]
> >>>>> end module constants
> >>>>> -^
> >>>>> _constants.f(226): error #5149: Illegal character in statement
> >>>>> label
> >>>>> field [d]
> >>>>> end module constants
> >>>>> --^
> >>>>> _constants.f(226): error #5149: Illegal character in statement
> >>>>> label
> >>>>> field [m]
> >>>>> end module constants
> >>>>> ----^
> >>>>> _constants.f(223): error #5082: Syntax error, found ',' when
> >>>>> expecting one of: (
> >>>>> % : . = =>
> >>>>> integer, parameter :: RETIRED_INPUT_OPTION = -10301 ! first 5
> >>>>> digit
> >>>>> palindromic
> >>>>> prime
> >>>>> -------^
> >>>>> _constants.f(223): error #5082: Syntax error, found '=' when
> >>>>> expecting one of: (
> >>>>> * ) :: , <END-OF-STATEMENT> ; + . - % (/ [ : ] /) . ** / // ...
> >>>>> integer, parameter :: RETIRED_INPUT_OPTION = -10301 ! first 5
> >>>>> digit
> >>>>> palindromic
> >>>>> prime
> >>>>> -------------------------------------------^
> >>>>> _constants.f(227): error #5082: Syntax error, found END-OF-FILE
> >>>>> when
> >>>>> expecting
> >>>>> one of: <END-OF-STATEMENT> ; <LABEL> BLOCK BLOCKDATA PROGRAM
> >>>>> MODULE
> >>>>> TYPE COMPLEX
> >>>>> BYTE ...
> >>>>>
> >>>>> ^
> >>>>> compilation aborted for _constants.f (code 1)
> >>>>> make[1]: *** [constants.o] Error 1
> >>>>> make: *** [parallel] Error 2
> >>>>>
> >>>>>
> >>>>> mpif90 -show gave following output:
> >>>>>
> >>>>> ifort -I/home/sangita/mpich2-install/include
> >>>>> -I/home/sangita/mpich2-install/include -L/home/sangita/mpich2-
> >>>>> install/lib
> >>>>> -L/home/sangita/mpich2-install/lib -lmpichf90 -lmpichf90 -lmpich -
> >>>>> lopa -lpthread
> >>>>> -lrt
> >>>>>
> >>>>>
> >>>>> When I ran configure_openmpi it telles that compiler is not
> >>>>> supported
> >>>>>
> >>>>> [sangita.localhost src]$ ./configure_openmpi
> >>>>> Setting AMBERHOME to /home/sangita/amber11/AmberTools
> >>>>>
> >>>>> Architecture/compiler is not supported
> >>>>>
> >>>>> Usage: ./configure_openmpi [flags] compiler
> >>>>>
> >>>>> where compiler is one of:
> >>>>>
> >>>>> gnu (=gcc/gfortran), intel (=icc/ifort), pgi (=pgcc/pgf90),
> >>>>> solaris (=cc/f90)
> >>>>>
> >>>>> Option flags:
> >>>>> -static Create statically linked executables (not
> >>>>> recommended for
> >>>>> MacOSX)
> >>>>>
> >>>>> Note: See http://www.open-mpi.org/software/ompi/v1.4/ for
> >>>>> information
> >>>>> on how to obtain openmpi.
> >>>>>
> >>>>>
> >>>>>
> >>>>>> On Thu, Sep 23, 2010, Sangita Kachhap wrote:
> >>>>>>>
> >>>>>>> I have installed amber11 in serial using gnu compiler for
> >>>>>>> AmberTool1.4. But when compiling it in parallel using same and
> >>>>>>> mpicc of
> >>>>>>> Mpich2 which I have installed in my home directory it is giving
> >>>>>>> follwing
> >>>>>>> error:
> >>>>>>>
> >>>>>>> compilation aborted for _constants.f (code 1)
> >>>>>>> make[1]: *** [constants.o] Error 1
> >>>>>>
> >>>>>> We need more information about the error messages: there must
> >>>>>> have
> >>>>>> been
> >>>>>> something before the snippet you quote above.
> >>>>>>
> >>>>>> I think the constants.f is the first file that is compiled in a
> >>>>>> parallel
> >>>>>> install, so there is probably something wrong with the compiler.
> >>>>>> Note that
> >>>>>> constants.f is a fortran file, so the key quantity is mpif90, not
> >>>>>> mpicc.
> >>>>>> What does "mpif90 -show" report?
> >>>>>>
> >>>>>> The easiest and recommended path for parallel installations is to
> >>>>>> run the
> >>>>>> "configure_openmpi" script in $AMBERHOME/AmberTools/src, and to
> >>>>>> use
> >>>>>> that
> >>>>>> installation (which will automatically be in your PATH if your
> >>>>>> PATH
> >>>>>> includes
> >>>>>> $AMBERHOME/bin). We know that this is configured correctly, and
> >>>>>> works for
> >>>>>> all the compilers we support.
> >>>>>>
> >>>>>> The case where this won't work is if you have a specialized
> >>>>>> interconnect (say,
> >>>>>> infiniband) which is not supported by generic openmpi. Even
> >>>>>> then,
> >>>>>> I would
> >>>>>> recommend using "configure_openmpi" first, so that you will
> >>>>>> have a
> >>>>>> baseline
> >>>>>> and some experience, before trying to link to an external MPI
> >>>>>> library.
> >>>>>>
> >>>>>> ....hope this helps....dac
> >>>>>>
> >>>>>>
> >>>>>> _______________________________________________
> >>>>>> AMBER mailing list
> >>>>>> AMBER.ambermd.org
> >>>>>> http://lists.ambermd.org/mailman/listinfo/amber
> >>>>>>
> >>>>>>
> >>>>>
> >>>>>
> >>>>> Sangita Kachhap
> >>>>> JRF
> >>>>> BIC,IMTECH
> >>>>> CHANDIGARH
> >>>>>
> >>>>> _______________________________________________
> >>>>> AMBER mailing list
> >>>>> AMBER.ambermd.org
> >>>>> http://lists.ambermd.org/mailman/listinfo/amber
> >>>>
> >>>>
> >>>> _______________________________________________
> >>>> AMBER mailing list
> >>>> AMBER.ambermd.org
> >>>> http://lists.ambermd.org/mailman/listinfo/amber
> >>>>
> >>>>
> >>>
> >>>
> >>> Sangita Kachhap
> >>> JRF
> >>> BIC,IMTECH
> >>> CHANDIGARH
> >>>
> >>> _______________________________________________
> >>> AMBER mailing list
> >>> AMBER.ambermd.org
> >>> http://lists.ambermd.org/mailman/listinfo/amber
> >>
> >>
> >> _______________________________________________
> >> AMBER mailing list
> >> AMBER.ambermd.org
> >> http://lists.ambermd.org/mailman/listinfo/amber
> >>
> >>
> >
> >
> > Sangita Kachhap
> > JRF
> > BIC,IMTECH
> > CHANDIGARH
> >
> > _______________________________________________
> > AMBER mailing list
> > AMBER.ambermd.org
> > http://lists.ambermd.org/mailman/listinfo/amber
>
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
--
Jason M. Swails
Quantum Theory Project,
University of Florida
Ph.D. Graduate Student
352-392-4032
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu Sep 23 2010 - 23:00:04 PDT
