abinit compiling os-x 10.8.2/gfortran 4.7/gcc4.7/openmpi  [SOLVED]

option, parallelism,...

Moderators: fgoudreault, mcote

Forum rules
Please have a look at ~abinit/doc/config/build-config.ac in the source package for detailed and up-to-date information about the configuration of Abinit 8 builds.
For a video explanation on how to build Abinit 7.x for Linux, please go to: http://www.youtube.com/watch?v=DppLQ-KQA68.
IMPORTANT: when an answer solves your problem, please check the little green V-like button on its upper-right corner to accept it.
Locked
davydden
Posts: 17
Joined: Fri Feb 15, 2013 6:50 pm

abinit compiling os-x 10.8.2/gfortran 4.7/gcc4.7/openmpi

Post by davydden » Fri Feb 15, 2013 6:58 pm

Dear All,

I'm very new to the Abinit (7.0.5), and trying to compile it on my os-x using macports.
After looking in forums i came up with the following config:

Code: Select all

./configure --prefix=/opt/abinit/7.0.5 --enable-mpi --enable-64bit-flags --with-mpi-prefix=/opt/local --with-linalg-flavor="atlas" --with-linalg-libs="-L/opt/local/lib -llapack -lcblas -lf77blas -latlas"  --enable-optim="yes" CFLAGS_OPTIM="-O2" FCFLAGS_OPTIM="-O2" --enable-fc-wrapper CC=mpicc FC=mpif90 CXX=mpicxx --enable-gui CPP=/opt/local/bin/cpp-mp-4.7 FCCCP=mpicc


after running config, i make
TRUE_CPP = /opt/local/bin/cpp-mp-4.7 -P -std=c99
in /fallbacks/config.mk

Problems i have are related to utility_cpp.f90:

Code: Select all

mpif90 -m64 utility_cpp.f90  -DHAVE_CONFIG_H -ffree-form -m64 -g -ffree-line-length-none -O2 -g -O2 -c -o utility.o  -I. -I.. -I/Users/davydden/Desktop/work/C++/abinit/abinit-7.0.5/fallbacks/exports/include -I/Users/davydden/Desktop/work/C++/abinit/abinit-7.0.5/fallbacks/exports/include
utility_cpp.f90:196.29:

    if(pos<=0) call io_error('utility_string_to_coord: Problem reading string i
                             1
Error: Missing ')' in statement at or before (1)
utility_cpp.f90:206.85:

o_error('utility_string_to_coord: Problem reading string into real number '
                                                                           1         
Error: Syntax error in argument list at (1)
utility_cpp.f90:198.25:

    read(ctemp2,*,err=100,end=100) outvec(1)
                         1
Error: ERR tag label 100 at (1) not defined
utility_cpp.f90:202.25:

    read(ctemp2,*,err=100,end=100) outvec(2)
                         1
Error: ERR tag label 100 at (1) not defined
utility_cpp.f90:204.24:

    read(ctemp,*,err=100,end=100) outvec(3)
                        1
Error: ERR tag label 100 at (1) not defined
ABINIT WRAPPER END (exit code 1)
make[6]: *** [utility.lo] Error 1
make[5]: *** [all-recursive] Error 1
make[4]: *** [all] Error 2
make[3]: *** [stamps/wannier90-build-stamp] Error 2
make[2]: [all-local] Error 2 (ignored)
Checking build of wannier90 fallback
test -e stamps/wannier90-install-stamp
make[2]: *** [all-local] Error 1
make[1]: *** [multi] Error 2
make: *** [mj4] Error 2


if I remove "--enable-fc-wrapper " I got the following error:

Code: Select all

Making all in 01_interfaces_ext
mpif90 -m64 -DHAVE_CONFIG_H -I. -I../.. -I../../src/incs -I../../src/incs -I/Users/davydden/Desktop/work/C++/abinit/abinit-7.0.5/fallbacks/exports/include -I/Users/davydden/Desktop/work/C++/abinit/abinit-7.0.5/fallbacks/exports/include   -ffree-form -J/Users/davydden/Desktop/work/C++/abinit/abinit-7.0.5/src/mods -O2 -m64 -g -ffree-line-length-none   -c -o m_iso_c_bindings.o m_iso_c_bindings.F90
In file included from m_iso_c_bindings.F90:28:0:
../../config.h:642:2: error: #error "HAVE_MPI1 and HAVE_MPI2 are both undefined"
make[5]: *** [m_iso_c_bindings.o] Error 1
make[4]: *** [all-recursive] Error 1
make[3]: *** [all-recursive] Error 1
make[2]: *** [all] Error 2
make[1]: *** [multi] Error 2
make: *** [mj4] Error 2

openmpi lib are in opt/loc/lib (installed by macports), thus "export DYLD_LIBRARY_PATH=/opt/local/lib:$DYLD_LIBRARY_PATH" does not help.

Thank you in advance for your help

Regards,
Denis.

User avatar
Alain_Jacques
Posts: 279
Joined: Sat Aug 15, 2009 9:34 pm
Location: Université catholique de Louvain - Belgium

Re: abinit compiling os-x 10.8.2/gfortran 4.7/gcc4.7/openmpi

Post by Alain_Jacques » Sat Feb 16, 2013 7:48 pm

Hi Denis,

FCCPP=mpicc is probably a bad idea ... cc is not a preprocessor.
I have a very similar environment. In order to avoid the problems related to the default Apple cpp 4.2.1, I have created a directory in my home with softlinks similar to these:

Code: Select all

cpp -> /opt/local/bin/cpp-mp-4.7
libtool -> /opt/local/bin/glibtool
libtoolize -> /opt/local/bin/glibtoolize
m4 -> /opt/local/bin/gm4

and then I

Code: Select all

export PATH=~/links:$PATH
to be sure to call cpp-mp-4.7. Then I have no trouble when compiling abinit-7.0.5 with --with-dft-flavor="libxc+atompaw" flag, mpich2-1.5 and macports gcc 4.7 toolchain, without CPP= , without FCCPP=.

Kind regards,

Alain

davydden
Posts: 17
Joined: Fri Feb 15, 2013 6:50 pm

Re: abinit compiling os-x 10.8.2/gfortran 4.7/gcc4.7/openmpi

Post by davydden » Sat Feb 16, 2013 9:25 pm

Thanks, Alain, for your advice.
I managed to make it compiled, finally.
Probably need to run some tests (hopefully won't have any sigfault stuff ;) )

Below is a detailed instructions i followed (just in case someone will be looking how to compile in os-x with openmpi, gnu compilers 4.7):

1. gcc 4.7,g++4.7,gfortran4.7 are coming from macports. As well as atlas (blas+lapack). Xcode has to be installed in order to install macports.
2. For some reason, openmpi from the macports was not compiled with gcc4.7 but with apple's one (even though i run "port install openmpi+gcc47"),
thus i used self-compiled version of openmpi 1.6. (that could be a reason for some problems i had in the above).
3. I created links in a home folder ~/links
ln -s /opt/local/bin/cpp-mp-4.7 cpp
ln -s /opt/local/bin/glibtool libtool
ln -s /opt/local/bin/glibtoolize libtoolize
ln -s /opt/local/bin/gm4 m4
and added it to the path in /User/xxxx/.profile along with macport's path and openmpi:
export PATH=/opt/openmpi/bin:~/links:/opt/local/bin:/opt/local/sbin:$PATH
export DYLD_LIBRARY_PATH=/opt/openmpi/lib/:$DYLD_LIBRARY_PATH
4. Config i used:
./configure --prefix=/opt/abinit/7.0.5 --enable-mpi --enable-64bit-flags --with-mpi-prefix=/opt/openmpi --with-linalg-flavor="atlas" --with-linalg-libs="-L/opt/local/lib -llapack -lcblas -lf77blas -latlas" --enable-gui --enable-optim="yes" CFLAGS_OPTIM="-O2" FCFLAGS_OPTIM="-O2"
5. make
6.make check (to compile and run tests)

davydden
Posts: 17
Joined: Fri Feb 15, 2013 6:50 pm

Re: abinit compiling os-x 10.8.2/gfortran 4.7/gcc4.7/openmpi

Post by davydden » Tue Feb 19, 2013 11:16 am

I run the "make check" and after quite a long time have the following error:

Code: Select all

[libxc][t41] File ab.files created, starting abinit
[libxc][t41] Finished abinit (Runtime: 4 seconds)
[libxc][t41] Comparing t41.out with reference file
[libxc][t41] Reference file: /Users/.../abinit-7.0.5/tests/libxc/Refs/t41.out
Doing floating point compare of t41.out with /Users/.../abinit-7.0.5/tests/libxc/Refs/t41.out and option -medium
Prepare the short report

[libxc][t999] File t*_WFK* erased

[libxc][t999] File t*_1WF* erased

[libxc][t999] File t*_DEN* erased

[libxc][t999] File t*_SCR* erased

[libxc][t999] File t*_SUS* erased

[libxc][t999] File t*_KSS* erased

[libxc][t999] File t*_xsf* erased
tutorial/tmp-test_Darwin_20130216/report:Case_gw2_2            failed (too many erroneous lines : 17 , accepted 16 )
tutorial/tmp-test_Darwin_20130219/report:Case_gw2_2            failed (too many erroneous lines : 17 , accepted 16 )
/bin/sh: /bin/false: No such file or directory
make[3]: *** [check-local] Error 127
make[2]: *** [check-am] Error 2
make[1]: *** [check-recursive] Error 1
make: *** [check-recursive] Error 1


Not sure how "make check" should finish, but that seems like a (shell) scripting error to me.
What do i do with it?

User avatar
Alain_Jacques
Posts: 279
Joined: Sat Aug 15, 2009 9:34 pm
Location: Université catholique de Louvain - Belgium

Re: abinit compiling os-x 10.8.2/gfortran 4.7/gcc4.7/openmpi

Post by Alain_Jacques » Tue Feb 19, 2013 11:47 am

Hi Denis,

I'm not sure what "make check" does. I would rather prefer that you change to the tests subdirectory in the build directory (where you did the configure, make, ...) and perform a make clean then make tests_acc or make tests_acc4 (with 4 parallel threads). Then do a more */*/report to view a short summary of the passed/failed tests.

You're right, false lives in /bin under Linux, under /usr/bin under OSX. The error message probably comes from an hard coded path in the scripts. It won't do any harm to link /bin/false to /usr/bin/false

I'm more concerned about the failed gw tests. Could you post the corresponding diff files?

Kind regards,

Alain

davydden
Posts: 17
Joined: Fri Feb 15, 2013 6:50 pm

Re: abinit compiling os-x 10.8.2/gfortran 4.7/gcc4.7/openmpi

Post by davydden » Tue Feb 19, 2013 2:54 pm

Dear Alain,

thanks for your kind and prompt reply.
So, following your advise I did links for false/true:
from within /bin: 1) ln -s /usr/bin/false false 2) ln -s /usr/bin/true true

Here are the last lines of output for "make tests_acc4"

Code: Select all

[etsf_io][t22] File ab.files created, starting abinit
[etsf_io][t22] Finished abinit (Runtime: 1 seconds)
[etsf_io][t22] Comparing t22.out with reference file
[etsf_io][t22] Reference file: /Users/davydden/Desktop/work/C++/abinit/abinit-7.0.5/tests/etsf_io/Refs/t22.out
Doing floating point compare of t22.out with /Users/davydden/Desktop/work/C++/abinit/abinit-7.0.5/tests/etsf_io/Refs/t22.out and option
Prepare the short report

[etsf_io][t999] File t*_WFK* erased

[etsf_io][t999] File t*_1WF* erased

[etsf_io][t999] File t*_DEN* erased

[etsf_io][t999] File t*_SCR* erased

[etsf_io][t999] File t*_SUS* erased

[etsf_io][t999] File t*_KSS* erased

[etsf_io][t999] File t*_xsf* erased
make tests_gui dirname=test
timeout="0"  ./Scripts/wrap-standard-tests.sh test gui 
Machine test
Bindir /Users/davydden/Desktop/work/C++/abinit/abinit-7.0.5/src/98_main
Top_testdir /Users/davydden/Desktop/work/C++/abinit/abinit-7.0.5/tests
Testdir  gui
cd gui
Testing the Abinit code on the test Darwin platform
Suite of tests GUI of abinit

Following tests will be run: first to end
Do not create directory, tmp-test_Darwin_20130219 already exists
cd tmp-test_Darwin_20130219

[gui][t01] Starting STATCHK
/bin/true 1>/Users/davydden/Desktop/work/C++/abinit/abinit-7.0.5/tests/gui/tmp-test_Darwin_20130219/t01.out 2>/Users/davydden/Desktop/work/C++/abinit/abinit-7.0.5/tests/gui/tmp-test_Darwin_20130219/t01.err
[gui][t01] Finished STATCHK, status OK
touch test4.end
while test "`ls -1 test*end 2>/dev/null | wc -l`" -lt 4 ; do  sleep 10; done
rm -f test1.end test2.end test3.end test4.end


t01.err is empty, so i guess it's fine.
The next best thing is to run "tests_min4" probably?

User avatar
Alain_Jacques
Posts: 279
Joined: Sat Aug 15, 2009 9:34 pm
Location: Université catholique de Louvain - Belgium

Re: abinit compiling os-x 10.8.2/gfortran 4.7/gcc4.7/openmpi

Post by Alain_Jacques » Tue Feb 19, 2013 3:38 pm

Hi Denis,

make tests_acc4 performs the full test suite aka the ACCuracy suite (not All Catastrophic Cases) with a few exceptions as for example long running inputs from the tutorials. So you can consider that you have a fully working abinit binary if it passes the suite with no failure. "Passed" tests are within the tolerances. Believe me, abinit developers are paranoid about accuracy and consistency of the results. Code contributions that break the tests_acc nightly checks on the test farm are never merged into the public trunk of Abinit and so should never appear in the production releases (not even in the development versions).

After running the make tests_acc4, you should do a more */*/report to display the (long) summary of the different tests sections (v1, v2, ..., gw, ...) and investigate those that are marked "failed". The files to look at are the diff.t??. The t??.err files contain the standard error; they will only contain relevant information in case of a catastrophic failure like a segfault.

Kind regards,

Alain

davydden
Posts: 17
Joined: Fri Feb 15, 2013 6:50 pm

Re: abinit compiling os-x 10.8.2/gfortran 4.7/gcc4.7/openmpi

Post by davydden » Wed Feb 20, 2013 9:38 am

Alain_Jacques wrote:After running the make tests_acc4, you should do a more */*/report to display the (long) summary of the different tests sections (v1, v2, ..., gw, ...) and investigate those that are marked "failed". The files to look at are the diff.t??. The t??.err files contain the standard error; they will only contain relevant information in case of a catastrophic failure like a segfault.


Dear Alain,

I went through most of directories in tests (*/*/report)
and i think everything is either passed or succeeded.

the "report" file in v1 complains about test indices:

Code: Select all

Error : indices of cases do not match in analysed and reference files 
Analysed file  summaryfldiff, line 0, gives Case_01 ;
Reference file /Users/davydden/Desktop/work/C++/abinit/abinit-7.0.5/tests/v1/Input/report.in, line 0, gives Case_00 ;
exit

"summaryfldiff", however, shows "no significant difference has been found" for all tests.
The only thing i notice is "tutorial:Case_gw2_2 failed (too many erroneous lines : 17 , accepted 16 )"
that's what is mentioned in summarydiff:
"Summary Case_gw2_2 : different lines= 17 , max abs_diff= 1.000e-03 (l.381), max rel_diff= 3.610e-03 (l.1320)"
Judging from the diff. does not look to be serious

Code: Select all

2,3c2,3
< .Version 7.0.5 of ABINIT
< .(MPI version, prepared for a i386_darwin12.2.1_gnu4.7 computer)
---
> .Version 7.0.2 of ABINIT
> .(MPI version, prepared for a x86_64_linux_gnu4.7 computer)
17,18c17,18
< .Starting date : Tue 19 Feb 2013.
< - ( at 14h13 )
---
> .Starting date : Sun  7 Oct 2012.
> - ( at 17h16 )
20c20
< - input  file    -> /Users/davydden/Desktop/work/C++/abinit/abinit-7.0.5/tests/tutorial/Input/tgw2_2.in
---
> - input  file    -> /home/gmatteo/archives/702/gmatteo-private/tests/tutorial/Input/tgw2_2.in
181c181
< .Using single precision arithmetic ; gwpc =  4
---
> .Using double precision arithmetic ; gwpc =  8
191,192c191,192
< - pspini: atom type   1  psp file is /Users/davydden/Desktop/work/C++/abinit/abinit-7.0.5/tests/Psps_for_tests/13al.981214.fhi
< - pspatm: opening atomic psp file    /Users/davydden/Desktop/work/C++/abinit/abinit-7.0.5/tests/Psps_for_tests/13al.981214.fhi
---
> - pspini: atom type   1  psp file is /home/gmatteo/archives/702/gmatteo-private/tests/Psps_for_tests/13al.981214.fhi
> - pspatm: opening atomic psp file    /home/gmatteo/archives/702/gmatteo-private/tests/Psps_for_tests/13al.981214.fhi
284c284
< - max sum_G a(n,k,G)a(n",k,G) =  0.240593
---
> - max sum_G a(n,k,G)a(n",k,G) =  0.240595
315c315
<         0.000  -0.000   0.000   0.000   0.000  -0.000   0.000   0.000  -0.000
---
>         0.000   0.000   0.000   0.000   0.000  -0.000  -0.000  -0.000  -0.000
318c318
<         0.000  -0.000   0.000   0.000  -0.000   0.000   0.000  -0.000   0.000
---
>        -0.000   0.000   0.000   0.000  -0.000   0.000   0.000  -0.000   0.000
357c357
<     2   0.000  -7.456   0.017   0.754   0.138   0.754   0.138   0.754   0.138
---
>     2   0.000  -7.457   0.017   0.754   0.138   0.754   0.138   0.754   0.138
381c381
<     2   0.000   6.061  -2.950  -2.197   0.127  -2.197   0.127  -2.197   0.126
---
>     2   0.000   6.060  -2.950  -2.197   0.127  -2.197   0.127  -2.197   0.126
389c389
<     2   0.000  -4.717  -3.528  -0.757  -0.981  -0.757  -0.981  -0.757  -0.981
---
>     2   0.000  -4.718  -3.528  -0.757  -0.981  -0.757  -0.981  -0.757  -0.981
398c398
<        -0.000   0.000   0.000   0.000  -0.000   0.000  -0.000   0.000  -0.000
---
>        -0.000   0.000   0.000   0.000  -0.000   0.000   0.000  -0.000   0.000
403c403
<         0.000  -0.000  -0.000  -0.000  -0.000  -0.000  -0.000  -0.000  -0.000
---
>        -0.000  -0.000  -0.000  -0.000  -0.000  -0.000  -0.000  -0.000  -0.000
406c406
<        -0.000   0.000   0.000  -0.000   0.000  -0.000   0.000   0.000  -0.000
---
>        -0.000  -0.000   0.000  -0.000  -0.000   0.000   0.000   0.000   0.000
414c414
<        -0.000  -0.000   0.000  -0.000  -0.000  -0.000   0.000   0.000   0.000
---
>        -0.000  -0.000   0.000  -0.000  -0.000   0.000   0.000   0.000   0.000
422c422
<        -0.000  -0.000   0.000  -0.000  -0.000   0.000   0.000   0.000  -0.000
---
>        -0.000  -0.000   0.000  -0.000  -0.000   0.000   0.000   0.000   0.000
434c434
<       0.000  -0.000   0.000   0.000   0.000  -0.000   0.000   0.000  -0.000
---
>       0.000  -0.000  -0.000   0.000   0.000  -0.000  -0.000   0.000   0.000
437c437
<       0.000   0.000  -0.000  -0.000  -0.000   0.000  -0.000  -0.000   0.000
---
>       0.000   0.000   0.000  -0.000  -0.000   0.000   0.000  -0.000  -0.000
533c533
<       0.000  -0.000  -0.000  -0.000  -0.000  -0.000  -0.000  -0.000  -0.000
---
>      -0.000  -0.000  -0.000  -0.000  -0.000  -0.000  -0.000  -0.000  -0.000
536c536
<       0.000  -0.000  -0.000  -0.000  -0.000  -0.000  -0.000  -0.000  -0.000
---
>      -0.000  -0.000  -0.000  -0.000  -0.000  -0.000  -0.000  -0.000  -0.000
565c565
<         0.000   0.000  -0.000   0.000  -0.000  -0.000  -0.000  -0.000  -0.000
---
>         0.000   0.000   0.000   0.000   0.000   0.000   0.000  -0.000   0.000
600c600
<         0.154  -5.842   1.595   1.541   1.063   1.541   1.063   1.541   1.063
---
>         0.154  -5.843   1.595   1.541   1.063   1.541   1.063   1.541   1.063
645c645
<        -0.000  -0.000   0.000  -0.000  -0.000   0.000  -0.000   0.000   0.000
---
>         0.000   0.000   0.000   0.000  -0.000  -0.000  -0.000  -0.000   0.000
648c648
<         0.000   0.000   0.000   0.000  -0.000   0.000   0.000   0.000   0.000
---
>        -0.000  -0.000   0.000   0.000  -0.000   0.000  -0.000   0.000  -0.000
653c653
<         0.000  -0.000   0.000   0.000  -0.000  -0.000  -0.000  -0.000  -0.000
---
>         0.000   0.000   0.000   0.000  -0.000  -0.000  -0.000   0.000   0.000
656c656
<         0.000  -0.000   0.000   0.000   0.000   0.000   0.000   0.000   0.000
---
>        -0.000  -0.000   0.000   0.000  -0.000   0.000  -0.000  -0.000  -0.000
661c661
<         0.000  -0.000   0.000   0.000   0.000   0.000   0.000   0.000   0.000
---
>         0.000   0.000   0.000   0.000  -0.000  -0.000  -0.000   0.000   0.000
664c664
<         0.000  -0.000   0.000  -0.000  -0.000   0.000  -0.000   0.000   0.000
---
>        -0.000  -0.000   0.000   0.000  -0.000   0.000  -0.000  -0.000  -0.000
669c669
<         0.000  -0.000   0.000   0.000  -0.000  -0.000  -0.000   0.000   0.000
---
>         0.000   0.000  -0.000   0.000  -0.000   0.000   0.000   0.000   0.000
672c672
<         0.000   0.000   0.000   0.000  -0.000   0.000   0.000   0.000  -0.000
---
>        -0.000  -0.000   0.000   0.000  -0.000   0.000  -0.000  -0.000  -0.000
683c683
<         0.000   0.000  -0.000   0.000  -0.000   0.000  -0.000   0.000   0.000
---
>         0.000   0.000   0.000   0.000  -0.000   0.000   0.000  -0.000   0.000
686c686
<        -0.000  -0.000   0.000  -0.000  -0.000   0.000  -0.000   0.000  -0.000
---
>        -0.000   0.000   0.000  -0.000  -0.000   0.000  -0.000   0.000  -0.000
722c722
<     1   3.486   0.410  -0.014  -0.119   0.236  -0.119   0.236  -0.119   0.236
---
>     1   3.485   0.410  -0.014  -0.119   0.236  -0.119   0.236  -0.119   0.236
763c763
<        -0.000   0.000   0.000  -0.000  -0.000  -0.000  -0.000  -0.000  -0.000
---
>        -0.000   0.000   0.000  -0.000  -0.000   0.000   0.000  -0.000   0.000
766c766
<        -0.000   0.000  -0.000  -0.000   0.000   0.000  -0.000   0.000  -0.000
---
>        -0.000  -0.000   0.000  -0.000  -0.000   0.000  -0.000  -0.000  -0.000
771c771
<         0.000   0.000   0.000   0.000  -0.000  -0.000  -0.000   0.000   0.000
---
>        -0.000   0.000   0.000  -0.000  -0.000  -0.000   0.000   0.000  -0.000
779c779
<        -0.000   0.000   0.000  -0.000  -0.000  -0.000  -0.000  -0.000  -0.000
---
>        -0.000   0.000  -0.000  -0.000  -0.000  -0.000   0.000   0.000  -0.000
782c782
<        -0.000   0.000   0.000  -0.000  -0.000   0.000  -0.000  -0.000  -0.000
---
>        -0.000  -0.000   0.000  -0.000  -0.000   0.000  -0.000  -0.000  -0.000
787c787
<         0.000   0.000   0.000  -0.000   0.000  -0.000   0.000  -0.000  -0.000
---
>        -0.000   0.000  -0.000  -0.000  -0.000  -0.000   0.000   0.000  -0.000
790c790
<        -0.000   0.000   0.000   0.000  -0.000   0.000  -0.000  -0.000  -0.000
---
>        -0.000  -0.000   0.000  -0.000  -0.000   0.000  -0.000  -0.000  -0.000
801c801
<        -0.000   0.000  -0.000  -0.000  -0.000  -0.000  -0.000   0.000  -0.000
---
>         0.000   0.000   0.000   0.000  -0.000   0.000  -0.000   0.000   0.000
881c881
<        -0.000  -0.000  -0.000  -0.000  -0.000  -0.000  -0.000   0.000   0.000
---
>        -0.000   0.000   0.000   0.000  -0.000  -0.000  -0.000   0.000   0.000
884c884
<         0.000   0.000   0.000  -0.000  -0.000   0.000  -0.000  -0.000   0.000
---
>        -0.000  -0.000   0.000  -0.000  -0.000   0.000  -0.000  -0.000  -0.000
889c889
<        -0.000  -0.000   0.000   0.000  -0.000   0.000  -0.000  -0.000  -0.000
---
>        -0.000   0.000   0.000   0.000  -0.000  -0.000  -0.000   0.000   0.000
892c892
<         0.000   0.000   0.000  -0.000  -0.000   0.000  -0.000  -0.000  -0.000
---
>        -0.000  -0.000   0.000  -0.000  -0.000   0.000  -0.000  -0.000  -0.000
897c897
<        -0.000   0.000   0.000   0.000  -0.000   0.000  -0.000   0.000   0.000
---
>        -0.000   0.000  -0.000   0.000  -0.000  -0.000  -0.000   0.000  -0.000
900c900
<        -0.000   0.000   0.000   0.000  -0.000   0.000  -0.000   0.000  -0.000
---
>        -0.000  -0.000   0.000   0.000  -0.000   0.000  -0.000  -0.000  -0.000
905c905
<        -0.000   0.000   0.000   0.000  -0.000  -0.000  -0.000   0.000  -0.000
---
>        -0.000   0.000  -0.000  -0.000  -0.000  -0.000  -0.000   0.000  -0.000
908c908
<        -0.000   0.000   0.000   0.000  -0.000   0.000  -0.000   0.000  -0.000
---
>        -0.000  -0.000   0.000   0.000  -0.000   0.000  -0.000   0.000  -0.000
919c919
<        -0.000   0.000  -0.000   0.000  -0.000   0.000  -0.000   0.000   0.000
---
>         0.000   0.000   0.000  -0.000  -0.000  -0.000  -0.000   0.000   0.000
938c938
<         0.024  -0.373  -0.356   0.161  -0.013   0.161  -0.013   0.142   0.164
---
>         0.024  -0.374  -0.356   0.161  -0.013   0.161  -0.013   0.142   0.164
943c943
<        -1.866  -0.062   0.027   0.007  -0.051   0.007  -0.051  -0.078  -0.087
---
>        -1.867  -0.062   0.027   0.007  -0.051   0.007  -0.051  -0.078  -0.087
999c999
<         0.000   0.000  -0.000   0.000  -0.000  -0.000  -0.000   0.000  -0.000
---
>        -0.000   0.000   0.000  -0.000  -0.000  -0.000  -0.000   0.000  -0.000
1002c1002
<        -0.000  -0.000  -0.000  -0.000  -0.000   0.000  -0.000   0.000  -0.000
---
>        -0.000  -0.000  -0.000  -0.000  -0.000   0.000  -0.000  -0.000  -0.000
1007c1007
<         0.000   0.000  -0.000  -0.000  -0.000  -0.000  -0.000   0.000  -0.000
---
>        -0.000   0.000   0.000  -0.000  -0.000  -0.000  -0.000   0.000  -0.000
1010c1010
<        -0.000  -0.000  -0.000  -0.000  -0.000   0.000  -0.000  -0.000   0.000
---
>        -0.000  -0.000  -0.000  -0.000  -0.000   0.000  -0.000  -0.000  -0.000
1015c1015
<         0.000   0.000  -0.000  -0.000  -0.000  -0.000  -0.000   0.000   0.000
---
>        -0.000   0.000   0.000  -0.000  -0.000  -0.000  -0.000   0.000  -0.000
1018c1018
<        -0.000  -0.000  -0.000   0.000  -0.000   0.000  -0.000  -0.000  -0.000
---
>        -0.000  -0.000   0.000  -0.000  -0.000   0.000  -0.000  -0.000  -0.000
1023c1023
<        -0.000   0.000   0.000   0.000  -0.000   0.000  -0.000  -0.000   0.000
---
>        -0.000   0.000   0.000  -0.000  -0.000  -0.000  -0.000   0.000  -0.000
1026c1026
<        -0.000   0.000  -0.000   0.000  -0.000   0.000  -0.000  -0.000  -0.000
---
>        -0.000  -0.000   0.000   0.000  -0.000   0.000  -0.000  -0.000  -0.000
1037c1037
<        -0.000   0.000   0.000  -0.000  -0.000  -0.000  -0.000  -0.000   0.000
---
>        -0.000   0.000   0.000   0.000  -0.000  -0.000  -0.000  -0.000   0.000
1040c1040
<        -0.000  -0.000   0.000   0.000   0.000   0.000   0.000  -0.000  -0.000
---
>        -0.000   0.000   0.000   0.000   0.000   0.000   0.000  -0.000  -0.000
1044c1044
<     1 -12.625  -0.244  -0.230  -0.205  -0.205  -0.205  -0.205  -0.230  -0.244
---
>     1 -12.626  -0.244  -0.230  -0.205  -0.205  -0.205  -0.205  -0.230  -0.244
1076c1076
<     1   5.368  -0.228   1.016  -0.026  -0.026  -0.026  -0.026   1.016  -0.228
---
>     1   5.367  -0.228   1.016  -0.026  -0.026  -0.026  -0.026   1.016  -0.228
1079c1079
<     2  -0.228  -9.018   0.485   1.415  -0.441   1.415  -0.441  -0.033  -0.251
---
>     2  -0.228  -9.019   0.485   1.415  -0.441   1.415  -0.441  -0.033  -0.251
1095c1095
<     2   0.182   7.464  -0.435  -1.396   0.734  -1.396   0.734  -1.453   1.140
---
>     2   0.182   7.463  -0.435  -1.396   0.734  -1.396   0.734  -1.453   1.140
1103c1103
<     2   0.104   9.199  -1.256  -1.625   0.394  -1.626   0.394  -1.226   1.060
---
>     2   0.104   9.199  -1.256  -1.626   0.394  -1.626   0.394  -1.226   1.060
1117c1117
<        -0.000  -0.000   0.000   0.000   0.000   0.000  -0.000   0.000  -0.000
---
>        -0.000  -0.000   0.000   0.000  -0.000  -0.000  -0.000  -0.000   0.000
1120c1120
<         0.000  -0.000   0.000   0.000   0.000  -0.000   0.000  -0.000   0.000
---
>         0.000   0.000   0.000   0.000   0.000   0.000   0.000  -0.000  -0.000
1125c1125
<         0.000   0.000  -0.000  -0.000  -0.000   0.000   0.000  -0.000  -0.000
---
>        -0.000  -0.000  -0.000   0.000  -0.000  -0.000  -0.000   0.000  -0.000
1128c1128
<        -0.000   0.000   0.000   0.000  -0.000   0.000   0.000  -0.000  -0.000
---
>         0.000  -0.000   0.000   0.000   0.000   0.000   0.000  -0.000   0.000
1133c1133
<        -0.000   0.000  -0.000   0.000   0.000   0.000   0.000   0.000   0.000
---
>        -0.000   0.000  -0.000   0.000  -0.000  -0.000  -0.000   0.000  -0.000
1136c1136
<        -0.000   0.000   0.000   0.000  -0.000   0.000   0.000  -0.000  -0.000
---
>        -0.000  -0.000   0.000   0.000  -0.000   0.000   0.000  -0.000   0.000
1141c1141
<         0.000  -0.000  -0.000  -0.000   0.000   0.000   0.000  -0.000  -0.000
---
>        -0.000   0.000  -0.000   0.000  -0.000  -0.000  -0.000   0.000  -0.000
1144c1144
<         0.000   0.000   0.000   0.000  -0.000  -0.000  -0.000  -0.000  -0.000
---
>        -0.000  -0.000   0.000   0.000  -0.000   0.000   0.000  -0.000   0.000
1155c1155
<         0.000   0.000  -0.000  -0.000  -0.000  -0.000  -0.000   0.000  -0.000
---
>         0.000   0.000  -0.000  -0.000  -0.000   0.000  -0.000   0.000   0.000
1158c1158
<        -0.000  -0.000  -0.000  -0.000  -0.000   0.000  -0.000  -0.000  -0.000
---
>        -0.000   0.000  -0.000  -0.000  -0.000   0.000  -0.000  -0.000  -0.000
1165c1165
<     2  -0.222  -3.959   0.136   0.423   0.189   0.423   0.189   0.286   0.304
---
>     2  -0.222  -3.960   0.136   0.423   0.189   0.423   0.189   0.286   0.304
1243c1243
<         0.000   0.000   0.000   0.000  -0.000   0.000  -0.000   0.000  -0.000
---
>        -0.000   0.000   0.000  -0.000  -0.000  -0.000  -0.000   0.000   0.000
1246c1246
<        -0.000  -0.000  -0.000  -0.000  -0.000  -0.000  -0.000   0.000  -0.000
---
>        -0.000  -0.000  -0.000  -0.000  -0.000   0.000  -0.000  -0.000  -0.000
1251c1251
<        -0.000   0.000  -0.000  -0.000  -0.000   0.000  -0.000   0.000   0.000
---
>        -0.000   0.000   0.000  -0.000  -0.000  -0.000  -0.000   0.000   0.000
1254c1254
<        -0.000   0.000  -0.000  -0.000  -0.000   0.000  -0.000  -0.000  -0.000
---
>        -0.000  -0.000  -0.000  -0.000  -0.000  -0.000  -0.000  -0.000  -0.000
1259c1259
<         0.000   0.000  -0.000  -0.000  -0.000   0.000  -0.000   0.000   0.000
---
>        -0.000   0.000   0.000  -0.000  -0.000  -0.000  -0.000   0.000   0.000
1262c1262
<        -0.000  -0.000  -0.000  -0.000  -0.000  -0.000  -0.000  -0.000  -0.000
---
>        -0.000  -0.000   0.000  -0.000  -0.000  -0.000  -0.000  -0.000  -0.000
1273c1273
<        -0.000   0.000   0.000  -0.000  -0.000  -0.000  -0.000  -0.000  -0.000
---
>         0.000  -0.000   0.000  -0.000   0.000  -0.000  -0.000  -0.000   0.000
1276c1276
<        -0.000  -0.000  -0.000  -0.000  -0.000   0.000   0.000  -0.000  -0.000
---
>         0.000   0.000   0.000  -0.000  -0.000   0.000   0.000  -0.000  -0.000
1280c1280
<     1 -10.180  -0.517  -0.159  -0.243  -0.099  -0.099  -0.243  -0.159  -0.517
---
>     1 -10.181  -0.517  -0.159  -0.243  -0.099  -0.099  -0.243  -0.159  -0.517
1283c1283
<     2  -0.517 -10.180   0.043   0.520   1.347   1.926   0.345   0.318  -0.800
---
>     2  -0.517 -10.181   0.043   0.520   1.347   1.926   0.345   0.318  -0.800
1320c1320
<     1   6.518   0.200   0.464   0.063  -0.138  -0.139   0.063   0.464   0.200
---
>     1   6.518   0.200   0.464   0.063  -0.139  -0.139   0.063   0.464   0.200
1347c1347
<     2   0.022   3.872  -0.665  -0.492   0.115  -0.036  -0.014  -0.320   0.335
---
>     2   0.022   3.871  -0.665  -0.492   0.115  -0.036  -0.014  -0.320   0.335
1353c1353
<         0.000   0.000  -0.000  -0.000  -0.000  -0.000  -0.000  -0.000   0.000
---
>        -0.000  -0.000   0.000  -0.000  -0.000  -0.000  -0.000  -0.000   0.000
1356c1356
<        -0.000  -0.000   0.000  -0.000   0.000  -0.000   0.000  -0.000  -0.000
---
>         0.000  -0.000   0.000   0.000  -0.000   0.000   0.000  -0.000  -0.000
1361c1361
<        -0.000  -0.000  -0.000   0.000   0.000   0.000  -0.000   0.000   0.000
---
>        -0.000  -0.000  -0.000   0.000  -0.000  -0.000  -0.000   0.000   0.000
1364c1364
<         0.000   0.000  -0.000   0.000  -0.000  -0.000   0.000  -0.000  -0.000
---
>         0.000  -0.000   0.000   0.000  -0.000   0.000   0.000  -0.000  -0.000
1369c1369
<         0.000  -0.000  -0.000  -0.000   0.000   0.000  -0.000   0.000  -0.000
---
>        -0.000  -0.000  -0.000  -0.000  -0.000  -0.000  -0.000   0.000   0.000
1372c1372
<         0.000   0.000   0.000   0.000  -0.000  -0.000   0.000  -0.000   0.000
---
>         0.000  -0.000   0.000   0.000  -0.000   0.000   0.000  -0.000   0.000
1377c1377
<        -0.000  -0.000  -0.000   0.000   0.000   0.000   0.000   0.000  -0.000
---
>        -0.000  -0.000  -0.000  -0.000  -0.000  -0.000   0.000   0.000   0.000
1380c1380
<         0.000   0.000  -0.000   0.000  -0.000  -0.000   0.000  -0.000  -0.000
---
>         0.000  -0.000  -0.000   0.000  -0.000   0.000   0.000  -0.000   0.000
1568c1568
< - Proc.   0 individual time (sec): cpu=          4.6  wall=          4.6
---
> - Proc.   0 individual time (sec): cpu=          8.6  wall=         10.1
1574c1574
< +Overall time at end (sec) : cpu=          4.6  wall=          4.6
---
> +Overall time at end (sec) : cpu=          8.6  wall=         10.1


Thanks again for your advises on compiling on OS-X and doing tests.

p/s/ i was trying to upload the diff, but it always complained about extension no matter how i rename the file.

Regards,
Denis.

User avatar
Alain_Jacques
Posts: 279
Joined: Sat Aug 15, 2009 9:34 pm
Location: Université catholique de Louvain - Belgium

Re: abinit compiling os-x 10.8.2/gfortran 4.7/gcc4.7/openmpi

Post by Alain_Jacques » Wed Feb 20, 2013 2:51 pm

gw2_2: a possible explanation is that the reference has been made with double precision GW, yours is simple.

For the allowed extensions, try .log - even for binary file. These safety restrictions are a bit childish.

Alain

davydden
Posts: 17
Joined: Fri Feb 15, 2013 6:50 pm

Re: abinit compiling os-x 10.8.2/gfortran 4.7/gcc4.7/openmpi

Post by davydden » Wed Mar 13, 2013 12:59 am

Dear Alain,

i just noticed that my compilation does not include FFT library,
i can compile Abinit with FFTW3 from macports

Code: Select all

./configure --prefix=/opt/abinit/7.0.5 --enable-mpi --enable-64bit-flags --with-mpi-prefix=/opt/openmpi --with-linalg-flavor="atlas" --with-linalg-libs="-L/opt/local/lib -llapack -lcblas -lf77blas -latlas" --enable-gui --enable-optim="yes" CFLAGS_OPTIM="-O2" FCFLAGS_OPTIM="-O2" \
--with-fft-flavor="fftw3" --with-fft-libs="-L/opt/local/lib -lfftw3" --with-fft-incs="-I/opt/local/include"

, however it is not compiled with OpenMPI.

When i try to use externally compiled FFTW3 (same compilers and openmpi):

Code: Select all

./configure --prefix=/opt/abinit/7.0.5 --enable-mpi --enable-64bit-flags --with-mpi-prefix=/opt/openmpi --with-linalg-flavor="atlas" --with-linalg-libs="-L/opt/local/lib -llapack -lcblas -lf77blas -latlas" --enable-gui --enable-optim="yes" CFLAGS_OPTIM="-O2" FCFLAGS_OPTIM="-O2" \
--with-fft-flavor="fftw3" --with-fft-libs="-L/opt/fftw3/lib -lfftw3f_mpi" --with-fft-incs="-I/opt/fftw3/include"

./configure complains that the provided FFTW3 lib is not working...
"make check" from FFTW3 build directory confirms that the library works fine and all tests are passed.

Could you please write how do you compile Abinit with FFTW3?

Thank you in advance,
Regards,
Denis.

User avatar
Alain_Jacques
Posts: 279
Joined: Sat Aug 15, 2009 9:34 pm
Location: Université catholique de Louvain - Belgium

Re: abinit compiling os-x 10.8.2/gfortran 4.7/gcc4.7/openmpi

Post by Alain_Jacques » Wed Mar 13, 2013 1:09 pm

Hi Denis,

Linking abinit with FFTW3 is definitely a well deserved effort ... it makes a significant speed improvement.

Unfortunately, macports provided fftw3 libraries don't work because the maintainer persists to disable Fortran support. If you peek into the corresponding Portfile, you'll see lines similar to
configure.args \
--enable-threads \
--disable-fortran \
--enable-shared

So no Fortran interface and therefore Abinit configure tests fail.
BTW that's the reason why I don't maintain an Abinit macports entry.

If you have some experience with macports internals, a solution would be to create a local repository by editing /opt/local/etc/macports/sources.conf to feed a modified fftw-3 Portfile with Fortran support.
Another way is to compile your own FFTW3 library. it's quite easy ... IIRW, I use a configure line like:
./configure --prefix=/opt/fftw-3.2.2 --enable-sse2 --enable-shared --enable-threads CC=gcc-mp-4.7 CPP=cpp-mp-4.7 F77=gfortran-mp-4.7
No need for fftw MPI. Replace enable-sse2 with enable-avx if your CPU supports these instructions (and your ar assembler doesn't choke on them). And don't forget to prepend the library path to DYLD_LIBRARY_PATH in order to use your own libs during Abinit configure tests instead of those provided with macports. You may have to play with the install_name_tool OSX utility on the final Abinit binaries to adjust the path to the right libraries.

Kind regards,

Alain

davydden
Posts: 17
Joined: Fri Feb 15, 2013 6:50 pm

Re: abinit compiling os-x 10.8.2/gfortran 4.7/gcc4.7/openmpi

Post by davydden » Wed Mar 13, 2013 4:47 pm

Dear Alain,

Thank you very much for your reply.

Alain_Jacques wrote:Another way is to compile your own FFTW3 library. it's quite easy ... IIRW, I use a configure line like:
./configure --prefix=/opt/fftw-3.2.2 --enable-sse2 --enable-shared --enable-threads CC=gcc-mp-4.7 CPP=cpp-mp-4.7 F77=gfortran-mp-4.7
No need for fftw MPI. Replace enable-sse2 with enable-avx if your CPU supports these instructions (and your ar assembler doesn't choke on them). And don't forget to prepend the library path to DYLD_LIBRARY_PATH in order to use your own libs during Abinit configure tests instead of those provided with macports. You may have to play with the install_name_tool OSX utility on the final Abinit binaries to adjust the path to the right libraries.


this configure worked fine for me, thank you! I successfully compiled Abinit.
I'm actually not sure if one need to add anything in DYLD path (i did not), it seems to me that my abinit is linked against the propper FFTW dylib:

Code: Select all

otool -L /opt/abinit/7.0.5/bin/abinit 
/opt/abinit/7.0.5/bin/abinit:
   /opt/fftw3/lib/libfftw3.3.dylib (compatibility version 7.0.0, current version 7.2.0)
   /opt/openmpi/lib/libmpi_f77.1.dylib (compatibility version 2.0.0, current version 2.3.0)
   /opt/openmpi/lib/libmpi.1.dylib (compatibility version 2.0.0, current version 2.3.0)
   /opt/local/lib/gcc47/libgfortran.3.dylib (compatibility version 4.0.0, current version 4.0.0)
   /usr/lib/libSystem.B.dylib (compatibility version 1.0.0, current version 169.3.0)
   /opt/local/lib/gcc47/libgcc_s.1.dylib (compatibility version 1.0.0, current version 1.0.0)
   /opt/local/lib/gcc47/libquadmath.0.dylib (compatibility version 1.0.0, current version 1.0.0)


These are config options i added:

Code: Select all

--with-fft-flavor="fftw3" --with-fft-libs="-L/opt/fftw3/lib -lfftw3" --with-fft-incs="-I/opt/fftw3/include"


I also wonder why should not i compile FFTW3 with openmpi support and actually provide that version of fftw3 to Abinit?

p/s/
A a side not, i was actually able to compile Abinit with the fftw3 from Macports

Code: Select all

 fftw-3 @3.3.3_0+gcc47 (active)
  fftw-3-single @3.3.3_0+gcc47 (active)

so maybe they changed something.

Code: Select all

--with-fft-flavor="fftw3" --with-fft-libs="-L/opt/local/lib -lfftw3" --with-fft-incs="-I/opt/local/include"


Kind regards,
Denis.

User avatar
Alain_Jacques
Posts: 279
Joined: Sat Aug 15, 2009 9:34 pm
Location: Université catholique de Louvain - Belgium

Re: abinit compiling os-x 10.8.2/gfortran 4.7/gcc4.7/openmpi  [SOLVED]

Post by Alain_Jacques » Wed Mar 13, 2013 6:47 pm

Hi Denis,

I also wonder why should not i compile FFTW3 with openmpi support and actually provide that version of fftw3 to Abinit?

You can always compile FFTW3 with MPI support, no adverse effect. But Abinit won't use this functionality, especially considering that you didn't include the libfftw3_mpi library in the configure flags of Abinit.

A a side not, i was actually able to compile Abinit with the fftw3 from Macports

I don't want to be stubborn but the corresponding part of the Portfile says
variant gcc47 description {compile using gcc47 toolchain} conflicts gcc43 gcc44 gcc45 gcc46 g95 {
227 depends_lib-append port:gcc47
228 configure.compiler macports-gcc-4.7
229 configure.args-delete --disable-fortran
230 if { ${os.arch} == "i386" } {
231 lappend merger_configure_args(ppc) "--disable-fortran"
232 lappend merger_configure_args(ppc64) "--disable-fortran"
233 } else {
234 lappend merger_configure_args(i386) "--disable-fortran"
235 lappend merger_configure_args(x86_64) "--disable-fortran"
236 }
237 }
And so Fortran support gets disabled making it unsuitable for Abinit. And I remember at least calls to the legacy Fortran interface in the FFTW3 wrappers and in the van der Waals correction to XC routine. An easy way to know is to look for dfftw_ symbols in macports libfftw3.dylib with an

Code: Select all

nm /opt/local/lib/libfftw3.dylib |grep dfftw_
; if they are in the code (i.e. T) then it's fine otherwise the final link will fail.

Kind regards,

Alain

sourabh.sinha
Posts: 2
Joined: Sun Mar 31, 2013 1:50 am

Re: abinit compiling os-x 10.8.2/gfortran 4.7/gcc4.7/openmpi

Post by sourabh.sinha » Sun Mar 31, 2013 3:29 am

Hi Dennis, Alain,
First thanks for your post. I was able to build abinit-7.0.5 on my mac OS X 10.8.3 by following your post, it was really helpful. I have couple of questions:

1. While testing by 'make tests_acc4' I got one more error other than 'Case_gw2_2'. That is:
Case_88 failed (too many erroneous lines : 23 , accepted 22 )

First and last few lines of ~/Downloads/abinit-7.0.5/tests/v4/tmp-test_Darwin_20130330/diff.t88 shows,

Code: Select all

2,3c2,3
< .Version 7.0.5 of ABINIT
< .(MPI version, prepared for a i386_darwin12.3.0_gnu4.7 computer)
---
> .Version 7.0.2 of ABINIT
> .(MPI version, prepared for a x86_64_linux_gnu4.7 computer)
17,18c17,18
< .Starting date : Sat 30 Mar 2013.
< - ( at 17h 0 )
---
> .Starting date : Sun  7 Oct 2012.
> - ( at 16h 8 )
20c20
< - input  file    -> /Users/sourabh/Downloads/abinit-7.0.5/tests/v4/Input/t88.in
---
> - input  file    -> /home/gmatteo/archives/702/gmatteo-private/tests/v4/Input/t88.in
110c110
< -   accesswff0 =  0 , fftalg0 =312 , wfoptalg0 =  0
---
> -   accesswff0 =  0 , fftalg0 =112 , wfoptalg0 =  0
341,342c341,342
< - pspini: atom type   1  psp file is /Users/sourabh/Downloads/abinit-7.0.5/tests/Psps_for_tests/14si.pspnc
< - pspatm: opening atomic psp file    /Users/sourabh/Downloads/abinit-7.0.5/tests/Psps_for_tests/14si.pspnc
---
> - pspini: atom type   1  psp file is /home/gmatteo/archives/702/gmatteo-private/tests/Psps_for_tests/14si.pspnc
> - pspatm: opening atomic psp file    /home/gmatteo/archives/702/gmatteo-private/tests/Psps_for_tests/14si.pspnc
379c379
<  ETOT  4  -8.8581677488816    -1.450E-05 2.626E-06 5.054E-03
---
>  ETOT  4  -8.8581677488815    -1.450E-05 2.626E-06 5.054E-03
381c381

.
.
.
.
.

<     1    -3.421   -10.699    -2.156     7.242   -15.917     5.991     1.000     0.000    -9.926    -0.527    -2.683    -2.683
---
>     1    -3.421   -10.699    -2.156     7.242   -15.917     5.990     1.000     0.000    -9.927    -0.528    -2.684    -2.685
1868c1868
<  New Fermi energy :        2.900605E-01 Ha ,    7.892949E+00 eV
---
>  New Fermi energy :        2.900605E-01 Ha ,    7.892948E+00 eV
1883c1883
< .  kptgw no:  3; Maximum DeltaE = (  -0.722   0.002) for band index:  1
---
> .  kptgw no:  3; Maximum DeltaE = (  -0.723   0.002) for band index:  1
1885c1885
< .  kptgw no:  5; Maximum DeltaE = (  -0.527   0.044) for band index:  1
---
> .  kptgw no:  5; Maximum DeltaE = (  -0.528   0.045) for band index:  1
1929,1930c1929,1930
<             fcart1     1.0248607038E-31 -1.0248607038E-31  1.0248607038E-31
<                       -1.0248607038E-31  1.0248607038E-31 -1.0248607038E-31
---
>             fcart1     0.0000000000E+00  1.0248607038E-31  0.0000000000E+00
>                        0.0000000000E+00 -1.0248607038E-31  0.0000000000E+00
2168c2168
< - Proc.   0 individual time (sec): cpu=         24.0  wall=         24.0
---
> - Proc.   0 individual time (sec): cpu=         23.6  wall=         23.6
2173,2174c2173,2174
< .Delivered   26 WARNINGs and  25 COMMENTs to log file.
< +Overall time at end (sec) : cpu=         24.0  wall=         24.0
---
> .Delivered   22 WARNINGs and  25 COMMENTs to log file.
> +Overall time at end (sec) : cpu=         23.6  wall=         23.6
(END)


Any insight why this error, how to resolve or is it serious?

2. This is regarding building with fftw3 library with openmpi. I was checking the release_notes (http://www.abinit.org/downloads/source- ... notes.html) of the current most version (7.2) and it says:

* Now FFTW3 wrappers can be used with MPI (and MPI+opemMP)
Introduction of fftalg 314
LOBPCG algorithm can now use fftalg=31X (e.g. 312 or 314) (FFTW3+MPI)
in addition to fftalg=40X
By F. Dahm and M. Torrent

So my second question is whether that solves the problem Alain mentioned for previous version (7.0.5) and if so, how do I first compile FFTW3 with openmpi support and next what will the abinit configure script look like. My current script is exactly same as suggested here, i.e.,

Code: Select all

./configure --prefix=/opt/UserPkg/Abinit/7.0.5 --enable-mpi --enable-64bit-flags --with-mpi-prefix=/opt/UserLib/openmpi-1.6.4 --with-linalg-flavor="atlas" --with-linalg-libs="-L/opt/local/lib -llapack -lcblas -lf77blas -latlas" --enable-gui --enable-optim="yes" CFLAGS_OPTIM="-O2" FCFLAGS_OPTIM="-O2" --with-fft-flavor="fftw3" --with-fft-libs="-L/opt/UserLib/fftw-3.3.3/lib -lfftw3" --with-fft-incs="-I/opt/UserLib/fftw-3.3.3/include"


Thanks,
Sourabh

User avatar
jbeuken
Posts: 365
Joined: Tue Aug 18, 2009 9:24 pm
Contact:

Re: abinit compiling os-x 10.8.2/gfortran 4.7/gcc4.7/openmpi

Post by jbeuken » Fri Apr 05, 2013 7:03 pm

Hi,

* Now FFTW3 wrappers can be used with MPI (and MPI+opemMP)
Introduction of fftalg 314
LOBPCG algorithm can now use fftalg=31X (e.g. 312 or 314) (FFTW3+MPI)
in addition to fftalg=40X
By F. Dahm and M. Torrent

So my second question is whether that solves the problem Alain mentioned for previous version (7.0.5) and if so, how do I first compile FFTW3 with openmpi support and next what will the abinit configure script look like. My current script is exactly same as suggested here, i.e.,
CODE: SELECT ALL
./configure --prefix=/opt/UserPkg/Abinit/7.0.5 --enable-mpi --enable-64bit-flags --with-mpi-prefix=/opt/UserLib/openmpi-1.6.4 --with-linalg-flavor="atlas" --with-linalg-libs="-L/opt/local/lib -llapack -lcblas -lf77blas -latlas" --enable-gui --enable-optim="yes" CFLAGS_OPTIM="-O2" FCFLAGS_OPTIM="-O2" --with-fft-flavor="fftw3" --with-fft-libs="-L/opt/UserLib/fftw-3.3.3/lib -lfftw3" --with-fft-incs="-I/opt/UserLib/fftw-3.3.3/include"


there is a bot in our abinit test farm which tests the "fftalg=314" ( yet experimental )

I compile the fftw3 3.3.3 with all "supports" : single and float + omp + mpi

Code: Select all

/usr/local/fftw3/lib/libfftw3.a
/usr/local/fftw3/lib/libfftw3f.a
/usr/local/fftw3/lib/libfftw3f_mpi.a
/usr/local/fftw3/lib/libfftw3f_omp.a
/usr/local/fftw3/lib/libfftw3_mpi.a
/usr/local/fftw3/lib/libfftw3_omp.a


this is my build.ac file

Code: Select all

enable_gw_dpc = yes
enable_mpi = yes
enable_mpi_io = yes
enable_openmp = yes
with_mpi_prefix = /usr/local/openmpi
with_dft_flavor = atompaw+bigdft+libxc+wannier90
with_trio_flavor = netcdf+etsf_io+fox
with_linalg_flavor = netlib
with_linalg_incs = -I/usr/local/include
with_linalg_libs = -L/usr/local/lib64 -llapack -lblas
with_fft_flavor = fftw3-mpi
with_fft_incs = -I/usr/local/fftw3/include/
with_fft_libs = -L/usr/local/fftw3/lib -lfftw3_mpi -lfftw3_omp -lfftw3 -lfftw3f


run this cmd : ./configure --with-config-file=./build.ac

concerning the tests,
all sequential succeed
for paral/mpiio, we must replace all fftalg parameter in Input/*.in files by fftalg=314 excluding theses tests :

tests/paral/Input/tK.in
tests/paral/Input/tAB.in
tests/paral/Input/tAF.in
tests/mpiio/Input/tAI.in


Hope these help,

jmb
------
Jean-Michel Beuken
Computer Scientist

Locked