[Pw_forum] Errors in test-suite qe-6.0 and qe-6.1 how bad ?

[Paolo Giannozzi]

Tests for CP are not very significant: they test energies after a few
MD steps, but even tiny differences quickly lead to different MD
trajectories. The numbers you get are rather close to the numbers in
tests, so likely nothing to worry about

Paolo

On Thu, Mar 23, 2017 at 5:35 PM, Thibaut VERY <thibaut.very at idris.fr> wrote:
> Hello,
>
> I installed qe-6.0 and 6.1 on an IBM machine using power8 cores.
> The binaries are linked to ESSL.
>
> When running the test-suite, I got errors that are almost the same for both versions.
> Some seems reasonable but other much less.
> The tests are done in parallel with 4 MPI tasks by setting the variable QE_USE_MPI=1.
> Here are some examples:
> cp_h2o - h2o-mt-blyp-1.in: **FAILED**.
> s1
>     ERROR: absolute error 4.99e-04 greater than 1.00e-05. (Test: -0.46198575.  Benchmark: -0.46148672.)
> cp_h2o - h2o-mt-blyp-4.in: **FAILED**.
> t1
>     ERROR: absolute error 1.92e-03 greater than 1.00e-05. (Test: 43.78585.  Benchmark: 43.78393.)
>
> cp_si - si-vbc-lda-3.in: **FAILED**.
> t1
>     ERROR: absolute error 3.69e-01 greater than 1.00e-05. (Test: 2527.82355.  Benchmark: 2528.19283.)
>
> All done. ERROR: only 10 out of 17 tests passed.
>
> Is it something expected or should I worry?
> In the second case, do you have some ideas how to correct it?
>
> Best regards
> --
> Thibaut Very
>
> thibaut.very at idris.fr
> http://www.idris.fr
> IDRIS - Bat.506 - BP 167 - 91403 Orsay Cedex - France
>
>
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://pwscf.org/mailman/listinfo/pw_forum



-- 
Paolo Giannozzi, Dip. Scienze Matematiche Informatiche e Fisiche,
Univ. Udine, via delle Scienze 208, 33100 Udine, Italy
Phone +39-0432-558216, fax +39-0432-558222