(Help!) Compiling NWChem6.1 with ifort,icc,mpich2, and mkl library
Clicked A Few Times
8:07:57 PM PDT - Thu, Mar 15th 2012
Hello everyone

These days I'm struggling to install NWChem6.1 on my lab's cluster PC.

I tried to find informations from internet and this forum. I could understand some part of them but not everything, and now I think I've got stuck.
(I'm not so familiar with linux system and commands)

Would you like to help me, please?

I briefly wrote about things I did below, but actually
I tried many times to compile NWChem with a little bit different bash scripts.
I think, however, below description would be enough to explain what I did.


This is the script that I wrote on "~/.bashrc" file.

export LARGE_FILES=TRUE
export USE_NOFSCHECK=TRUE
export NWCHEM_TOP=/opt/nwchem-6.1 # where I unzipped the tar file
export NWCHEM_TARGET=LINUX64
export NWCHEM_MODULES=all

export BLASOPT="-L/opt/intel/Compiler/11.1/059/mkl/lib/em64t -lmkl_intel_ilp64 -lmkl_sequential -lmkl_core" # I found my cluster has these libraries in this directory
export FC=ifort
export CC=icc

export TCGRSH=/usr/bin/ssh
export USE_MPI=y
export USE_MPIF=y
export USE_MPIF4=y
export MPI_LOC=/opt/mpi/intel/mpich2-1.2.1p1
export LIBMPI="-L/opt/mpi/intel/mpich2-1.2.1p1/lib/ -lmpich -lopa" # I found my cluster has these libraries in this directory
export MPI_LIB=$MPI_LOC/lib
export MPI_INCLUDE=$MPI_LOC/include


Then I did
   ] $ source ~/.bashrc

   ] $ cd /opt/nwchem-6.1/src
src] $ make nwchem_config # I got about one page of messages after typing this.
src] $ make >& make.log #Then I waited. More than 5 or 10 minutes were needed.


After that, I checked the "make.log" file. (contains about 13,000 lines)
At the end of that file, I found these error messages.

tce_residual_t1.F(72): error #6404: This name does not have a type, and must have an explicit type. [DDOT]
             residual = ddot(size,dbl_mb(k_r1),1,dbl_mb(k_r1),1)
^
tce_residual_t1.F(176): error #6404: This name does not have a type, and must have an explicit type. [DDOT]
             residual = ddot(size,dbl_mb(k_r1),1,dbl_mb(k_r1),1)
^
compilation aborted for tce_residual_t1.F (code 1)make[2]: *** [/opt/nwchem-6.1/lib/LINUX64/libtce.a(tce_residual_t1.o)] Error 1
make[2]: *** Waiting for unfinished jobs....make[1]: *** [optimized] Error 2
make[1]: *** Waiting for unfinished jobs....make: *** [libraries] Error 1


I first checked 72nd and 176th line of the "tce_residual_t1.F"
and found "residual = ddot(......)" script as wrotten above.

I tried google search about this and found that this error may come from "util.fh" file.
(http://www.emsl.pnl.gov/docs/nwchem/nwchem-support/2009/11/0026.Re:_NWCHEM_error_compiling...)

Then I manually added "double precision ddot" and "external ddot" on "util.fh" file and saved.


After that, I did "make" again.

src] $ make realclean
src] $ make nwchem_config
src] $ make FC=ifort >& make.log


After about 10 minutes again, I checked the make.log file.
And I found the same error message!
So I checked the util.fh file again, and found that they don't have two "ddot" lines that I added!


Now I have no idea what to do.

Actually I'm not sure whether I wrote the bash script appropriately.
(Is the grammer correct? Did I missed some scripts that should be included? Did I missed some environment of my cluster PC?)

I feel sorry to ask this kind of lengthy question.
I would be really grateful if you could give me some advice...

Forum Vet
12:13:41 PM PDT - Fri, Mar 16th 2012
please send the output of "cat $NWCHEM_TOP/src/util/util.fh"
Yjleedaniel,
Could you please post the output of the command

cat $NWCHEM_TOP/src/util/util.fh

I have some theories that might explain your failure, but I need to see what's in your util.fh to be sure.

Cheers, Edo

Clicked A Few Times
9:04:10 PM PDT - Sun, Mar 18th 2012
Thanks a lot, Edoapra!
Thank you, Edoapra! I feel very thankful for your concern about this problem.

Now I can run NWChem. (I got help from one student in other lab) I changed the NWChem scripts in .bashrc file.

This is the new scripts that I wrote.



  ### (same scripts as before) ###

  export LARGE_FILES=TRUE

  export USE_NOFSCHECK=TRUE

  export NWCHEM_TARGET=LINUX64

  export NWCHEM_MODULES=all

  export TCGRSH=/usr/bin/ssh

  export USE_MPI=y

  export USE_MPIF=y

  export USE_MPIF4=y

  export MPI_LIB=$MPI_LOC/lib

  export MPI_INCLUDE=$MPI_LOC/include

  

  ### (changed) ###

  export NWCHEM_TOP=/opt/nwchem/nwchem-6.1   # I changed the path a little for general site running.

  export NWCHEM_EXEC=/opt/nwchem/nwchem-6.1-general/bin/nwchem

  export BLASOPT="-L/opt/intel/Compiler/11.1/059/mkl/lib/em64t -lmkl_intel_ilp64 -lmkl_sequential -lmkl_core -lmkl_lapack"      # "mkl_lapack" library was added

  export MPI_LOC=/opt/mpi/intel/openmpi-1.4.1   # changed from mpich2 to openmpi

  export LIBMPI="-L/opt/mpi/intel/openmpi-1.4.1/lib -lmpi -lopen-pal -lopen-rte -lmpi_f90 -lmpi_f77"    # libraries in openmpi folder

  export FC=$MPI_LOC/bin/mpif90     #here I wrote path for compilers in openmpi folder

  export F77=$MPI_LOC/bin/mpif77

  export CC=$MPI_LOC/bin/mpicc



I checked the util.fh file after I succeeded. I found that it has "ddot scripts"!!

This is util.fh scripts that I checked after running NWChem.



  #include "printlevels.fh"

  c

  C$Id: util.fh 19707 2010-10-29 17:59:36Z d3y133 $

     logical util_print

     double precision ddot

     double precision util_cpusec

     double precision util_wallsec

     double precision util_random

     integer util_batch_job_time_remaining

     integer util_time_remaining

     logical util_test_time_remaining

     logical util_nwchemrc_get

     external util_print

     external ddot

     external util_cpusec

     external util_wallsec

     external util_random

     external util_batch_job_time_remaining

     external util_time_remaining

     external util_test_time_remaining

     external util_nwchemrc_get

  c

     logical util_xyz_seek

     external util_xyz_seek

     logical util_xyz_nframes

     external util_xyz_nframes

  c

     logical util_get_io_unit

     external util_get_io_unit

  c

     integer nw_max_path_len   ! Maximum path len -> posix standard is what?

     parameter (nw_max_path_len = 255)

  c




I see the difference between two .bashrc script but still don't understand what made the difference.
(I mean, I couldn't run NWChem before but now I can. Also util.fh has ddot scripts.)
Would you(Edoapra? anyone?) like to explain the difference between two scripts?

Anyway, I'd like to express my thanks again. (Thank you!!)

Forum Vet
10:28:52 AM PDT - Mon, Mar 19th 2012
Yjleedaniel,
Could you please post the complete compile line that you can in both cases? This should be something like
ifort .... or mpif90 ...
?
Thanks

Clicked A Few Times
5:16:22 AM PDT - Thu, Mar 22nd 2012
Thanks again, Edoapra.
Sorry for late-

Actually, I tried to post about the description that I did, but I couldn't. When I click the submit button after writing the post,
an error message like "The specified URL cannot be found" comes out.

Now I'm going to describe what I did briefly. Still I don't know the reason but it seems when I write a long post, the error message comes out.

I did what the INSTALL manual file says. (I mean, the install file in nwchem-6.1 folder)

Clicked A Few Times
5:35:08 AM PDT - Thu, Mar 22nd 2012
Sorry for this!
I'm trying to post but I can't!
I'm sorry for above post (only one sentence description?!), but when I try to post some more lines, the error message comes out!

Forum Vet
3:23:27 PM PDT - Fri, Mar 30th 2012
We have found that certain combinations of characters are considered operations/actions and those get blocked. Please send the requested data to edoardo.apra@pnnl.gov.

Bert

Quote:Yjleedaniel Mar 22nd 12:35 pm
I'm trying to post but I can't!
I'm sorry for above post (only one sentence description?!), but when I try to post some more lines, the error message comes out!


Forum >> NWChem's corner >> Compiling NWChem