Martin Diehl
ebb2e3ca85
avoid hardcoding (requires/ensures systematic naming)
2023-12-30 16:02:00 +01:00
Martin Diehl
936328ae8b
re-enabled support for older CMake
2023-12-29 08:32:52 +01:00
Martin Diehl
e4e0840479
simplified
2023-12-28 09:28:04 +01:00
Martin Diehl
af7f73f379
CMake knows compiler flags
2023-12-27 15:07:04 +01:00
Martin Diehl
bdb30118fb
starting to support new LLVM compiler
...
does not work yet, but good to have for testing once in a while.
Current version is 17, require minimum 19 to make clear that it does not
work
2023-12-21 06:30:46 +01:00
Martin Diehl
d1807b1860
libfyaml 0.9 has a list of libraries
2023-10-20 17:30:28 +02:00
Martin Diehl
57e02bb991
PETSc 3.20
2023-10-09 20:10:05 +02:00
Martin Diehl
7d1433875d
PETSc 3.19 is out
2023-04-13 20:13:01 +02:00
Martin Diehl
2147ac3a12
base framework for unit testing in Fortran
2023-02-24 22:27:56 +01:00
Martin Diehl
42329a4ca4
ensure recent compiler
2022-12-17 14:52:41 +01:00
Martin Diehl
0ca4d42678
PETSc 3.18.0 compatibility
...
CHKERR macros are now longer, should be replaced with PetscCall in the
future
2022-10-02 12:48:54 +02:00
Martin Diehl
3b57934d6e
Marc precision handling
2022-06-27 08:37:41 +00:00
Martin Diehl
58fab14c6c
PETSc 3.17.1 adjustments
2022-05-06 12:34:32 +02:00
Martin Diehl
c806885cb5
include libfyaml headers from non-standard locations
...
the CMake manual promises more than it holds.
Also adding some diagnostics+fixes for zlib, we still assume that it
exists (true on most systems, HDF5 also requires it normally)
see also https://stackoverflow.com/questions/21593
2022-04-13 07:58:31 +02:00
Martin Diehl
723252ef15
libfyaml backend
...
currently it simply converts the whole file content to flow mode.
The nice thing is: It can convert EVERYTHING to flow mode. libfyaml is
the only library that passes the whole YAML test suite
2022-04-12 23:24:07 +02:00
Martin Diehl
affa7baa22
search for libfyaml
...
no consequences so far
2022-04-12 21:02:09 +02:00
Martin Diehl
16f5a27e08
detecting HDF5 and FFTW
...
some PETSc installations are not aware of them, so it might make sense
to become independent of PETSc. But PETSc-defined variables should take
precedence
2022-04-09 21:10:15 +02:00
Martin Diehl
8d44c55141
support for PETSc 3.17.0
...
DCHKERRQ has been replaced by PetscCall
2022-04-03 17:56:17 +02:00
Martin Diehl
f36db86b3c
better debug for gfortran, removing options not needed
2022-02-05 18:24:24 +01:00
Martin Diehl
771e8acdb9
revert change
...
need to find PETSc 'twice' to ensure version range ...
2022-01-12 07:46:25 +01:00
Martin Diehl
eea8aa94af
not used
2022-01-10 18:50:28 +01:00
Martin Diehl
3ae0452440
more robust build
...
${PETSC_DIR} and ${PETSC_ARCH} are not always substituted in
${PETSC_DIR}/${PETSC_ARCH}/lib/petsc/conf/petscvariables
Therefore, get location from pkgconfig (<XXX>_LIBRARY_DIRS) and use
PETSC_EXTERNAL_LIB_BASIC instead of PETSC_WITH_EXTERNAL_LIB to get the
linker commands for the external libraries.
Also simplified regex/cleaning
2022-01-10 12:01:48 +01:00
Martin Diehl
a9293db171
new LLVM-based Intel compilers work
...
quick test shows significant performance improvement
2021-12-25 18:36:35 +01:00
Martin Diehl
d086162350
consistent style
...
there is no space before opening brackets in the CMake doc
2021-10-15 22:16:47 +02:00
Martin Diehl
dd7549a381
seems to work without further changes
2021-10-06 07:50:42 +02:00
Martin Diehl
caf9a29b73
useful information, especially in logs from bug reports
2021-08-19 07:10:00 +02:00
Martin Diehl
49ccbf3eb8
still not perfect:
...
https://stackoverflow.com/questions/32411963
2021-08-18 22:20:30 +02:00
Martin Diehl
8e75e87ad9
Merge branch 'MPI_F08' into polishing-for-beta
2021-07-09 11:32:32 +02:00
Martin Diehl
4c3ff4bef4
consistent capitalization
2021-07-08 15:05:01 +02:00
Martin Diehl
249365979f
for spack, following DealII
...
https://spack.readthedocs.io/en/latest/build_systems/cmakepackage.html
'RelWithDebInfo' is not easy to remember
2021-07-06 06:49:25 +02:00
Martin Diehl
f0a75d8d62
consistent capitalization
...
Capitals for pre processor macros to avoid clashes
2021-07-04 17:43:19 +02:00
Martin Diehl
67352d9ec7
location depends on PETSC_ARCH (if set)
2021-05-19 19:22:06 +02:00
Martin Diehl
f6684a0692
easier to read
2021-05-19 18:53:25 +02:00
Martin Diehl
3cf2ea1749
simplified
2021-05-19 17:31:59 +02:00
Martin Diehl
d4f439ae8b
read directly from file
...
detour via dummy makefile not needed
2021-05-19 17:19:11 +02:00
Martin Diehl
15c498f1ed
polishing
2021-05-19 17:02:11 +02:00
Martin Diehl
3a5ecc8be5
simplified and in correct order
2021-05-19 14:26:17 +02:00
Martin Diehl
49d51e196c
no hope for GPI
...
flang will hopefully work in the near future
2021-05-19 13:42:14 +02:00
Martin Diehl
80b6aaec77
inclusion of rules not needed
2021-05-19 13:31:18 +02:00
Martin Diehl
7250fbb7c6
DAMASK is compatible with PETSc 3.15
...
still need to update the test server (and DAMASK_interface.f90)
2021-04-25 08:06:12 +02:00
Martin Diehl
e8b3e0f3ee
fail as early as possible
2021-01-04 10:01:24 +01:00
Martin Diehl
35f9861818
Fortran standard is 2018
...
will not work for older compilers
2020-12-18 15:19:04 +01:00
Martin Diehl
9ce406ed11
wrapper for zlib library
...
assume that zlib is available for linking. This is usually the case
(zlib is a stable standard library), but there is currently no check
that gives a nice error message or the possibility to use zlib from a
non-standard location
2020-09-06 17:49:00 +02:00
Martin Diehl
59c5fbc5fe
position independent code required for Fedora 32
2020-04-30 08:05:42 +02:00
Martin Diehl
d99c05860e
FEM is now Mesh
2020-04-28 10:35:43 +02:00
Martin Diehl
540428aab7
works also for ifort
2020-03-29 19:06:17 +02:00
Martin Diehl
9690f170e1
4 space indentation
2020-01-29 22:44:42 +01:00
Martin Diehl
7e9ce183a4
better readable
2020-01-26 15:10:27 +01:00
Martin Diehl
bd5f963260
polishing
2020-01-04 16:54:07 +01:00
Martin Diehl
cba7114dff
there is also an FEM solver that operates on regular grids
2020-01-02 21:42:28 +01:00