[CP2K:2678] Re: compile the latest version of cp2k with g95

Hou Zhufeng vale... at gmail.com
Wed Apr 21 16:48:33 CEST 2010


Dear axel,

Sorry for spelling your name wrong in my last reply.

After trying many times for the compilation and installment of scalpack,
openmpi, and cp2k with g95, finally I obtained an useable cp2k.popt. The
problem comes from the compilation of BLACS with g95.  After removing
"TRANSCOMM = -DCSameF77" in the Bmake.inc and recompiling BLACS, the problem
is solved.  If "TRANSCOMM = -DCSameF77" is added into "Bmake.inc, it will
lead to a warning message when BI_MPI_F77_to_c_trans_comm.c in BLACS is
compiled.

Cheers

Zh


On Wed, Apr 21, 2010 at 10:56 PM, Hou Zhufeng <vale... at gmail.com> wrote:

> Dear Alex,
>
> Thank you very much for your reply.
>
> In the source codes of cp2k "message_r_passing.F", there is a comment as
> follows:
> "
>     ! requires that your MPI installation matches your Fortran compiler...
> be aware
>     ! e.g. on a 64 bit machine, NAG will require MPI_ADDRESS_KIND=4, and
> gfortran MPI_ADDRESS_KIND=8
> "
> Are the error messages related to the "MPI_ADDRESS_KIND"?  How can we make
> MPI_ADDRESS_KIND=8 when installing openmpi with g95?
> Actually, when the "message_passing.f90" and "dbcsr_message_passing.f90"
> were compiled with openmpi and g95, some warning messages appeared:
> for example:
>     Included at dbcsr_message_passing.f90:1822
>
>     CALL MPI_WIN_CREATE(range, winsize, data_type_size,&
>                         2
> Warning (155): Inconsistent types (COMPLEX(8)/INTEGER(4)) in actual
> argument lists at (1) and (2)
>
>
> Zh
>
>
>
> On Wed, Apr 21, 2010 at 10:19 PM, Axel <akoh... at gmail.com> wrote:
>
>>
>> On Apr 21, 8:43 am, zh <vale... at gmail.com> wrote:
>> > Dear all,
>> >
>> > I tried to compile the latest version (e.g. 2.1.223) of cp2k with g95,
>> > scalapck (0.96), openmpi-1.4.1. The g95 installed on my computer
>> > cluster is the 0.92 version built on Linux x86_64/EMT64 (32 bit
>> > D.I.).  The fftw3, openmpi, and scalapack are compiled with this
>> > version of g95 and installed on my machines.The following arch file
>> > was used for compiling cp2k:
>>
>> as you can see from the stack trace, the error
>> is coming from openmpi and thus unlikely to be
>> compiler dependent.
>>
>> what you see on the XT5 is very likely a different issue.
>>
>> axel.
>>
>>
>>
>>
>> > CC       = cc
>> > CPP      = /usr/bin/cpp
>> > FC       = /work/z-hou/soft/openmpi-g95/bin/mpif90
>> > LD       = /work/z-hou/soft/openmpi-g95/bin/mpif90
>> > AR       = ar -r
>> > DFLAGS   = -D__G95 -D__FFTSG -D__parallel -D__BLACS -D__SCALAPACK -
>> > D__FFTW3
>> > G95_INC  = /work/z-hou/soft/fftw3-g95/include
>> > CPPFLAGS = -C -traditional  $(DFLAGS) -I$(G95_INC)
>> > FCFLAGS  = $(DFLAGS) -I$(G95_INC) -ffree-form -fno-underscoring  -O1 -
>> > ffast-math -cpp -g -fbounds-check -ftrace=full   -funroll-loops
>> > LDFLAGS  = $(FCFLAGS)
>> > G95_LIB = /work/z-hou/scalapack_installer_0.96/lib
>> > LIBS     = $(G95_LIB)/libscalapack.a \
>> >            $(G95_LIB)/blacsC.a \
>> >            $(G95_LIB)/blacsF77.a \
>> >            $(G95_LIB)/blacs.a \
>> >            $(G95_LIB)/libreflapack.a \
>> >            $(G95_LIB)/librefblas.a \
>> >            /work/z-hou/soft/fftw3-g95/lib/libfftw3.a
>> >
>> > OBJECTS_ARCHITECTURE = machine_g95.o
>> >
>> > The binary module of cp2k (i.e., cp2k.popt) is successfully obtained
>> > with the above arch file. But when I run it in parallel manner, the
>> > following errors were met:
>> >
>> > [pcc-hs36:16109] *** Process received signal ***
>> > [pcc-hs36:16109] Signal: Segmentation fault (11)
>> > [pcc-hs36:16109] Signal code:  (128)
>> > [pcc-hs36:16109] Failing at address: (nil)
>> > [pcc-hs36:16109] [ 0] /lib64/libpthread.so.0 [0x382700e7c0]
>> > [pcc-hs36:16109] [ 1] /work/z-hou/soft/openmpi-g95/lib/libmpi.so.0
>> > [0x2b66ad6d1c9b]
>> > [pcc-hs36:16109] [ 2] /work/z-hou/soft/openmpi-g95/lib/libmpi.so.
>> > 0(PMPI_Group_incl+0x172) [0x2b66ad6f87d2]
>> > [pcc-hs36:16109] [ 3] /work/z-hou/bin/cp2k.sopt(BI_TransUserComm+0x52)
>> > [0x34957e2]
>> > [pcc-hs36:16109] [ 4] /work/z-hou/bin/cp2k.sopt(blacs_gridmap+0x10d)
>> > [0x348dc0d]
>> > [pcc-hs36:16109] [ 5] /work/z-hou/bin/cp2k.sopt(blacs_gridinit+0xbb)
>> > [0x348dadb]
>> > [pcc-hs36:16109] [ 6] /work/z-hou/bin/
>> > cp2k.sopt(cp_blacs_calls_MP_cp_blacs_gridinit+0x4c) [0x2cc830c]
>> > [pcc-hs36:16109] [ 7] /work/z-hou/bin/
>> > cp2k.sopt(cp_blacs_env_MP_cp_blacs_env_create+0x542) [0x1b1984a]
>> > [pcc-hs36:16109] [ 8] /work/z-hou/bin/
>> > cp2k.sopt(qs_environment_types_MP_init_qs_env+0x1098) [0xb1ef24]
>> > [pcc-hs36:16109] [ 9] /work/z-hou/bin/
>> > cp2k.sopt(qs_environment_types_MP_qs_env_create+0x122) [0xb21b4d]
>> > [pcc-hs36:16109] [10] /work/z-hou/bin/
>> > cp2k.sopt(qs_main_MP_quickstep_create_force_env+0x12b) [0xcff143]
>> > [pcc-hs36:16109] [11] /work/z-hou/bin/
>> > cp2k.sopt(f77_interface_MP_create_force_env+0x1c15) [0x4f515b]
>> > [pcc-hs36:16109] [12] /work/z-hou/bin/cp2k.sopt(cp2k_runs_MP_cp2k_run
>> > +0x1b83) [0x4afc53]
>> > [pcc-hs36:16109] [13] /work/z-hou/bin/cp2k.sopt(cp2k_runs_MP_run_input
>> > +0x24d) [0x4bab8b]
>> > [pcc-hs36:16109] [14] /work/z-hou/bin/cp2k.sopt(MAIN_+0x154b)
>> > [0x4ad1cb]
>> > [pcc-hs36:16109] [15] /work/z-hou/bin/cp2k.sopt(main+0xe) [0x366f97e]
>> > [pcc-hs36:16109] [16] /lib64/libc.so.6(__libc_start_main+0xf4)
>> > [0x382641d994]
>> > [pcc-hs36:16109] [17] /work/z-hou/bin/cp2k.sopt [0x4abbe9]
>> > [pcc-hs36:16109] *** End of error message ***
>> >
>> --------------------------------------------------------------------------
>> > mpirun noticed that process rank 0 with PID 16109 on node pcc-hs36
>> > exited on signal 11 (Segmentation fault).
>> >
>> --------------------------------------------------------------------------
>> >
>> > From searching the solution of the similar problem on internet, it
>> > seems to be related to how fortran compiler handles the 32 bit and 64
>> > bit in mpi. But I did not find out the exact solution.
>> >
>> > I also tried the 0.92 version of g95 built on Linux x86_64/EMT64 (64
>> > bit D.I.). The same errors were met.  I also suffered this terrible
>> > problem when I used the intel fortran compiler on a same machine and
>> > the pgi fortran compiler on a CARY-XT5 machine to compile the latest
>> > version of cp2k. If anyone can give me some helps or hints to fix this
>> > problem, I will greatly appreciate him.
>> >
>> > --
>> > You received this message because you are subscribed to the Google
>> Groups "cp2k" group.
>> > To post to this group, send email to cp... at googlegroups.com.
>> > To unsubscribe from this group, send email to
>> cp2k+uns... at googlegroups.com <cp2k%2Bun... at googlegroups.com>.
>> > For more options, visit this group athttp://
>> groups.google.com/group/cp2k?hl=en.
>>
>> --
>> You received this message because you are subscribed to the Google Groups
>> "cp2k" group.
>> To post to this group, send email to cp... at googlegroups.com.
>> To unsubscribe from this group, send email to
>> cp2k+uns... at googlegroups.com <cp2k%2Bun... at googlegroups.com>.
>> For more options, visit this group at
>> http://groups.google.com/group/cp2k?hl=en.
>>
>>
>

-- 
You received this message because you are subscribed to the Google Groups "cp2k" group.
To post to this group, send email to cp... at googlegroups.com.
To unsubscribe from this group, send email to cp2k+uns... at googlegroups.com.
For more options, visit this group at http://groups.google.com/group/cp2k?hl=en.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20100421/c72c154f/attachment.html>


More information about the CP2K-user mailing list