[solved] Abinit 6.8.2. Testsuite v4 case48/49 fail

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
nandaloo
Posts: 12
Joined: Thu Sep 15, 2011 4:48 am

[solved] Abinit 6.8.2. Testsuite v4 case48/49 fail

Post by nandaloo » Fri Sep 30, 2011 8:58 am

Hello

I got one question about the built-in testing suite: Why do I need "make" for running tests? Shouldn't it just be calling the abinit installation with some input parameters instead of building code? Or does it in fact not build code at all?

Then, I've installed three 'flavours' of ABINIT 6.8.2 on my ubuntu 10.04 64bit machine:
- sequential only
- mpi (OpenMPI 1.4.2) enabled
- mpi (OpenMPI 1.4.2) + CUDA (SDK 4.0) enabled

"../configure" and "make mj4" works fine, however, running the tests with "make tests_min4" produces the exact same error (and numbers) with all three flavours:
all fast tests succeed, tests in v1-v6 succeed, but in v4:

Code: Select all

Case_48            failed (too large absolute error : 12.6707953000341 , accepted 2.0e-6 )
Case_49            failed (too large absolute error : 17.4036964809396 , accepted 3.0e-6 )

The parallel tests that are done with tests_min4 (make tests_paral paral_host=test paral_mode=seq) succeed as well, at least I think. However, I cannot find any report file for it.
The output on the screen just was, without any warning or error:

Code: Select all

Parallel tests set A
Parallel tests set B
 ... etc ...
Parallel tests set AB


I use gcc 4.4.3 if that is of any interest.

Can anyone give me a hint how to fix that? Or, is this error important for me at all if I'm not planning to do any calculations that use the functionality which was used for this particular test case?

Thank you a lot for your help

Philipp
Last edited by nandaloo on Fri Oct 07, 2011 6:27 am, edited 1 time in total.
Top
nandaloo
Posts: 12
Joined: Thu Sep 15, 2011 4:48 am

Re: Abinit 6.8.2. Testsuite v4 case48/49 fail

Post by nandaloo » Fri Sep 30, 2011 11:09 am

Also, if I do

Code: Select all

make tests_in

I get this error for all installations:
echo "Running built-in test in_wannier90"
Running built-in test in_wannier90
timeout="0" /usr/bin/perl ../../tests/Scripts/run-basic-tests.pl built-in in_wannier90 /home/cudauser/ABINIT_SOURCE/nMPInCUDA/..
STOP wannier90 error: examine the output/error file for details

Status file, with repetition rate 49, status number 295

Level abinit : call driver
Level driver : call gstateimg
Level gstateimg : enter
Level gstate : call scfcv
Level scfcv : call outscfcv
istep = 3

[edit:] In "testin_wannier90o_w90.wout" it says this:
Time to get kmesh 0.020 (sec)
ERROR: unitarity of initial U
nkp= 1
i= 1 j= 1
[u_matrix.transpose(u_matrix)]_ij= 0.936050 0.000000
Exiting.......
Error in unitarity of initial U in overlap_project

I just downloaded the source code and built it again - same result.

Maybe it's about the plugins? Is it better to manually install and link them?
Thanks a lot for your suggestions!

Philipp
Top
nandaloo
Posts: 12
Joined: Thu Sep 15, 2011 4:48 am

Re: Abinit 6.8.2. Testsuite v4 case48/49 fail

Post by nandaloo » Fri Oct 07, 2011 5:18 am

Compiling Wannier90 with LAPACK and ATLAS (and NOT standard BLAS from the ubuntu repo) and linking against these library when compiling Abinit solved the problem.
Top
Locked

Return to “Configuring and compiling ABINIT”