building NWChem with Intel 14.0 and OpenMPI 1.6.5


Just Got Here
Hello, I am trying to build NWChem with Intel 14.0 and OpenMPI 1.6.5 with the following configuration:

cd Nwchem-6.5.revision26243-src.2014-09-10
export NWCHEM_TOP=`pwd`
export NWCHEM_TARGET=LINUX64
export NWCHEM_MODULES=all

module load openmpi/1.6.5/intel
module load intel/14.0

export USE_MPI=y
export USE_MPIF=y
export MPI_LIB=/opt/openmpi/1.6.5/intel/13.1/lib
export MPI_INCLUDE=/opt/openmpi/1.6.5/intel/13.1/include
export LIBMPI="-L$MPI_LIB -lmpi_f90 -lmpi_f77 -lmpi -ldl -lm -lnuma -lrt -lnsl -lutil -pthread"

export FC=ifort
export CC=icc
export ARMCI_NETWORK=OPENIB
export IB_HOME=/usr
export IB_INCLUDE=/usr/include/infiniband/
export IB_LIB=/usr/lib64
export IB_LIB_NAME="-libumad -libverbs -lpthread"

export MKLROOT=/opt/intel/composer_xe_2013_sp1.2.144/mkl/lib/intel64
export BLASOPT="-Wl,--start-group $MKLROOT/libmkl_intel_lp64.a $MKLROOT/libmkl_sequential.a $MKLROOT/libmkl_core.a -Wl,--end-group -lpthread -lm"

However, I keep getting this error which I am unable to debug because I don't know what command is being executed:

Parallel-mpi.F(124): (col. 7) remark: LOOP WAS VECTORIZED
compilation aborted for Parallel-mpi.F (code 1)
make[4]: *** [/data/home/eew918/Nwchem-6.5.revision26243-src.2014-09-10/lib/LINUX64/libnwpwlib.a(Parallel-mpi.o)] Error 1
make[3]: *** [optimized] Error 2
make[2]: *** [subdirs] Error 1
make[1]: *** [subdirs] Error 1
make: *** [libraries] Error 1

Any help will be greatly appreciated.

Thanks in advance,
Wadud.

Forum Vet
Wadud
What release of Intel 14.0 are you using (e.g. output of "ifort -V")?
We have not seen the error you are reporting on Intel 14.0 ourselves.

A potential fix for your crash might be to turn off vectorization for the loop at line 124 of Parallel-mpi.F, e.g.

!DIR$ NOVECTOR
      do i=0,np-1
        int_mb(procNd(1)+i) = i
      end do

Cheers, Edo

Just Got Here
Hello Edo,

Thanks for your reply. Below is the version of my Intel compiler:

[eew918@frontend1.apocrita Nwchem-6.5.revision26243-src.2014-09-10]$ ifort -V
Intel(R) Fortran Intel(R) 64 Compiler XE for applications running on Intel(R) 64, Version 14.0.2.144 Build 20140120
Copyright (C) 1985-2014 Intel Corporation. All rights reserved.

Regards,

Forum Vet
Hi Wadud
I have found a computer with ifort 14.0.2.144, but I still did not manage to reproduce your failure.
Could you please do the following and post the complete output (including the full compilation line that should be similar to
ifort -c -i8 -align -vec-report6 -fimf-arch-consistency=true -O3 ... )
cd $NWCHEM_TOP/src/nwpw/nwpwlib/Parallel
make FC=ifort (or make FC=ftn if you are on a Cray)

Clicked A Few Times
The problem seems to be that Parallel_send_values, Parallel_send_ivalues, and Parallel_send_characters do not define mpierr as type integer. It seems one hits this problem if _not_ using MPIF4 (I happened to have just run into this problem while trying to build with an OpenMPI build that support i8).

Gabe

Forum Vet
Gabe
Thank you very much for nailing down this issue.
A patch is available at
http://nwchemgit.github.io/images/Parallelmpi.patch.gz


Forum >> NWChem's corner >> Compiling NWChem