abinit 7.8.2 tests [libxc][t03] and [v67mbpt][t28] failed

weitong · Post by **weitong** » Wed Jul 30, 2014 1:09 am

Following is my build.ac.

enable_64bit_flags="yes"
prefix="/usr/local/abinit7.8.2_mpich3intel13"
CC="mpicc"
CXX="mpicxx"
FC="mpif90"
enable_mpi="yes"
enable_mpi_io="yes"
with_mpi_prefix="/usr/local/mpich3_intel13"
with_trio_flavor="netcdf+etsf_io+fox"
with_fft_flavor="fftw3"
with_fft_libs="-L/opt/intel/mkl/lib/intel64 -Wl,--start-group  -lmkl_intel_lp64 -lmkl_sequential -lmkl_core -Wl,--end-group"
with_linalg_flavor="mkl"
with_linalg_libs="-L/opt/intel/mkl/lib/intel64 -Wl,--start-group  -lmkl_intel_lp64 -lmkl_sequential -lmkl_core -Wl,--end-group"
with_dft_flavor="atompaw+bigdft+libxc+wannier90"
enable_gw_dpc="yes"

Run tests suit, it gives the results:

Code: Select all

Suite        failed  passed  succeeded  skipped  disabled  run_etime  tot_etime
atompaw           0       1          1        0         0      17.11      17.59
bigdft            0       3         19        0         0     509.55     512.61
built-in          0       0          7        0         0      11.55      12.24
etsf_io           0       0          7        0         0      15.86      16.85
fast              0       0         11        0         0      42.58      45.42
fox               0       1          1        0         0      90.13      90.40
gpu               0       0          0        4         0       0.00       0.00
libxc             1       4         16        0         0     227.47     235.15
mpiio             0       0          1       13         0       2.75       2.88
paral             0       5         16       68         0     487.02     495.17
seq               0       0          0       18         0       0.00       0.01
tutoparal         0       0          1        1         0       0.48       0.53
tutoplugs         0       4          0        0         0      19.72      20.48
tutorespfn        0       8         14        0         0    1184.58    1201.21
tutorial          0       7         41        0         0     887.37     902.54
unitary           0       0         21       13         0      89.71      90.69
v1                0       2         74        0         0     232.01     245.96
v2                0      12         67        0         0     288.21     305.81
v3                0      12         68        0         0     450.59     479.16
v4                0      17         45        0         0     531.12     555.42
v5                0      17         57        0         0    1065.16    1105.31
v6                0      12         48        0         0     779.16     809.91
v67mbpt           1       4         12        0         0     308.22     314.31
v7                0      13         16        0         0     644.34     659.04
vdwxc             0       0          1        0         0      13.60      13.69
wannier90         0       7          0        0         0      38.17      39.20

The suite_report.html gives

Code: Select all

Results of test [libxc][t03][np=1]
MPI nprocs = 1, run_etime = 4.93 s, tot_etime = 5.02 s
stdin, stdout, stderr

File_to_test    Status    fld_output    fld_options    txt_diff    html_diff
t03.out   failed   t03.out.fldiff   ['-medium']   t03.out.diff   

t03.out failed: absolute error 0.2 > 0.11

Code: Select all

Results of test [v67mbpt][t28][np=1]
MPI nprocs = 1, run_etime = 4.43 s, tot_etime = 4.63 s
stdin, stdout, stderr

File_to_test    Status    fld_output    fld_options    txt_diff    html_diff
t28.out   failed   t28.out.fldiff   ['-medium']   t28.out.diff   

t28.out failed: relative error 0.9148 > 0.5

Jordan · Post by **Jordan** » Wed Jul 30, 2014 3:45 pm

Hi,

According to the fact that you have only 2 failed tests, and that the error are quite close to the tolerence criteria, you can assume your abinit works.

Numerical error are du to the way cpu actually do the calculation and how the code was compiled. So you may give the operating system you use and the cpu on you computer and the version of each compiler/library you use as well as the dft_flavors ( in particular libxc ).

Cheers,

Jordan

weitong · Post by **weitong** » Sat Aug 02, 2014 9:03 pm

Thanks.
My OS is CentOS 6.5- 64bit. Compiler is mpicc and mpif90 for MPICH version 3.1, icc and ifort version 13.1.3.192 Build 20130607 (gcc version 4.4.7 compatibility).
dft_flavor="atompaw+bigdft+libxc+wannier90", atompaw-3.1.0.2, bigdft-1.7.0.93, bigdft-abi-1.0.4, libxc-2.0.3.

Jordan wrote:Hi,

According to the fact that you have only 2 failed tests, and that the error are quite close to the tolerence criteria, you can assume your abinit works.

Numerical error are du to the way cpu actually do the calculation and how the code was compiled. So you may give the operating system you use and the cpu on you computer and the version of each compiler/library you use as well as the dft_flavors ( in particular libxc ).

Cheers,

Jordan

Jordan · Post by **Jordan** » Mon Aug 04, 2014 3:39 pm

Abinit is tested on CentOs 6.5 with intel 13.1.3 compiler and openmpi 1.6.5 MKL11.0 and FFTW3 (from MKL) and libxc.

Our config file is very likely yours :

Code: Select all

enable_mpi="yes"
enable_mpi_io="yes"
with_mpi_prefix="/usr/local/openmpi_intel13"
with_trio_flavor="netcdf+etsf_io+fox"
with_fft_flavor="fftw3"
with_fft_libs="-L/opt/intel/mkl/lib/intel64 -Wl,--start-group  -lmkl_intel_lp64 -lmkl_sequential -lmkl_core -Wl,--end-group"
with_linalg_flavor="mkl"
with_linalg_libs="-L/opt/intel/mkl/lib/intel64 -Wl,--start-group  -lmkl_intel_lp64 -lmkl_sequential -lmkl_core -Wl,--end-group"
with_dft_flavor="atompaw+bigdft+libxc+wannier90"
enable_clib="yes"
enable_gw_dpc="yes"
enable_memory_profiling="no"
enable_openmp="no"
enable_maintainer_checks="no"

The CPU is an Intel Xeon. What do you have ? You should post the t03.out and t28.out so we can investigate.

Cheers.

PS : As I said, I would not worry about those 2 errors (especially if you don't want to use MBPT nor functionnal such as BLYP BP OLYP (thoses tested in t03).

weitong · Post by **weitong** » Fri Aug 08, 2014 3:15 pm

Thanks.
My CPU is 2x Xeon E5-2690. Attached please find 2 files, t03.out (libxc) and t28.out (v67mbpt).
Meanwhile, I am going to try openmpi.
---Just tested: with openmpi 1.6 +intel13., it shows the same error.
WT

Jordan · Post by **Jordan** » Mon Aug 18, 2014 5:08 pm

I've had a look to your tests and compared them to the references.
To me, there is no significant difference. For the libxc test, note that none of the dataset is converged and thus you can not expect the first 6 iterations to be exactely the same. But when convergency is reached the results should be the same.
The main difference is the FFT used in your case (312) when the references were built with (112). The default value has recently changed. That coul be the explanation.

Conclusion : your installation is ready for production.

Cheers

Jordan

ABINIT Discussion Forums

abinit 7.8.2 tests [libxc][t03] and [v67mbpt][t28] failed [SOLVED]

abinit 7.8.2 tests [libxc][t03] and [v67mbpt][t28] failed

Re: abinit 7.8.2 tests [libxc][t03] and [v67mbpt][t28] faile

Re: abinit 7.8.2 tests [libxc][t03] and [v67mbpt][t28] faile

Re: abinit 7.8.2 tests [libxc][t03] and [v67mbpt][t28] faile

Re: abinit 7.8.2 tests [libxc][t03] and [v67mbpt][t28] faile

Re: abinit 7.8.2 tests [libxc][t03] and [v67mbpt][t28] faile [SOLVED]