mirror of https://github.com/abinit/abinit.git
36 lines
1.4 KiB
Plaintext
36 lines
1.4 KiB
Plaintext
In this directory, one has gathered (perhaps temporarily)
|
|
the tests that could NOT be executed in parallel, or
|
|
should be executed in parallel independently of the present seq only runs.
|
|
They have been taken from different test directories
|
|
at the time of the v5.9.0 merge.
|
|
|
|
Please, see the README file in the tests directory (above this one) to get help.
|
|
|
|
|
|
For the description of tests, see the corresponding original directories.
|
|
|
|
The reasons are :
|
|
|
|
1) Not parallelised capabilities of ABINIT
|
|
v2#81,82, v3#03,04,05 -> Berry phase calculation, with positive value of nberry
|
|
v4#90 -> CASINO output
|
|
|
|
2) Numerically incorrect cases, when MPI plugin is used
|
|
(to be corrected as soon as possible)
|
|
v4#55 -> finite electric field
|
|
v4#78,80 -> finite electric field phonon calculations,
|
|
v5#101 ->TFvW preconditioner
|
|
v5#112,113 -> Berry phase in PAW
|
|
|
|
|
|
Added one more test, v6.15.0 (XG120521), calculation with red_efieldbar (finite electric field),
|
|
because the seq and paral runs differ. Actualy linked with v4#55 behaviour ...
|
|
v6#121
|
|
Actually, I do not understand how some other berryopt 4 calculation actually
|
|
give essentially the same result (v5#23, v6#43, tutorespfn#ffield_6), or a difference
|
|
that is acceptable (v6#111).
|
|
|
|
Added five more tests (v6#122-126), v6.15.1 (XG120616), similar reason than for the test v6#121.
|
|
|
|
Added v7#70, v7.9.3 (XG1400901, similar reason than for the test v6#121.
|