In this case matrix_ is a nullptr and noGhostMat_ is the optimized
matrix to use. Before this commit we experienced a segmentation fault
as the nullptr was dereferenced and passed to the solver.
Previously, one had to call a seperate binary called
flow_blackoil_dunecpr. Unforntunately, that was only built if users
requested that the tests (`-DBUILD_TESTING=ON`) and the flow
variants (`-DBUILD_FLOW_VARIANTS=ON`) should be built. In addition
it would use a slightly different nonlinear solver implementation.
With this commit flow can be asked to either use
CPR (`--use-cpr=true --matrix-add-well-contributions=true`) or to use
the flexible solvers bx setting the `--linear-solver-configuration`
option. In all other cases the usual solver implementations are still
used.
Note that the flexible solvers still need
`--matrix-add-well-contributions=true` and hence cannot cope with
multi-segment wells.
The default was 0 for flow and 3 for flow_blackoil_dunecpr. Now we
always use 3. This is safe as it will not be used for any other
solvers than the flexible ones.
as there already is a specialization for an arbitrary Block
and that one leads to compile errors because we redefine it
in OPM (and miss some other specializations for Opm:MatrixBlock,
e.g. isNumber).
At least on Debian 10 the standard c++ compiler is g++-8,
but CUDA only supports g++-7 and our test for CUDA in cmake
did send an error in that case/combination which is quite
annoying.
The reason was that check_language(CUDA) did not honor
the CMAKE_CUDA_FLAGS variable and always used the default g++7,
but enable_language(CUDA) did honor it. As we do set the underlying
host compiler the fromer reported that CUDA is available while the
latter marked the CUDA compiler as broken.
With this commit we work around this by setting the environment
variable ENV{CUDAHOSTCXX} which nvcc will use. Hence now we only try
to enable CUDA if it is compatible with the C++ compiler
If no LinearSolverConfiguration is specified but UseCpr is true we
use cpr_quasiimpess. If the user does not specify the maximum
iteration number we use 20 for cpr.
to prevent throwing. The get methods will throw if called without a
default value. This quite unfortunate and not very user friendly, as
there are many properties and the throwing will happen during the
linear solve and result in time step chopping.
This commit should prevent such throws and allow users to provide
jsdon files omitting some options.
The previous was "file" which resulted in runtime errors
like
Program threw an exception: No linear-solver-configuration-json-file given for linear-solver-configuration=file
Previously, it got passed the weights only needed for CPR.
Additionally those were passed with the parameter tree to the
update method and constructor.
Now the CPR constructor gets a function to use for recalculating
the weights and the property is not changed. Unfortunately this
means that the preconditioner creators of the factory get another
parameter.
That the local ids were consecutive and starting from 0 was just
a coincidence and they should never be used to access linear systems
or vectors. This commit fixes this by using the correct mappers instead.
Note the we removed some computations from the constructor of
ISTLSolverEbosCpr as it inherits from ISTLSolverEbos and the operations
already happnen in constructor of the base class.
The InitialStateComputer::temperature_ array, previously used,
was of cartesian size, but used as if it only contained the
values of the cells of a region.
With this commit InitialStateComputer::temperature_ is a compressed local
array and we explicitly copy out only the region values when computing
RS/RV.
Closes#2423.
There are field properties that can usually be queried even if they
are not explicitly specified in the input
file (e.g. PVTNUM). Unfortunately, the ParallelEclipseState cannot
forsee which of these will be queried at startup and after the
loadbalancing only the master process is able to auto creates
these (easily). Hence this commit uses a fall-back if an unstored
keyword is queried. In this case we use get_global-* to auto create
the keyword and use functions of the cartesian mapper to extract the
relevant values on the process.
Of course this temporarily wastes space and we might want to resort to
a more memory savy approach later.
With the arrival of compressed field properties there is no need
to extract the global arrays just to compress them manually
afterwards. This change should remove commununication and
synchronization points.
There is not reason to it as a vector of cartesian size as we access
its entries by the compressed element index. This should save space
and speedup the lookup.
They are attached to the cells as well and are now distributed
during CpGrid::loadBalance. Due to this change we also rename
FieldPropsDataHandle to PropsCentroidsDataHandle.
The created data handle for the communication could in theory be used
with other DUNE grids, too. In reality we will need to merge with the
handle that ALUGrid already uses to communicate the cartesian indices.
This PR gets rid of using the get_global_(double|int) method in
ParallelEclipseState and reduces the amount of boilerplate code there.
It seems like the VERSION_GREATER_EQUAL operator for boolean
expressions was introduced after CMake 3.6 and hence the current
check whether to activate CUDA or not is broken in version 3.6 and
below.
This PR fixes this by using VERSION_GREATER.
Closes#2375.
We experienced weired linker errors when using host compiler version for
compilation that were not supported by the nvcc used to compile the
cuda code:
```
[ 15%] Linking CXX executable bin/test_timer
/usr/bin/ld: /home/mblatt/src/dune/opm-2.6/opm-common/opm-seq/lib/libopmcommon.a(Parser.cpp.o): in function `Opm::(anonymous namespace)::file& std::vector<Opm::(anonymous namespace)::file, std::allocator<Opm::(anonymous namespace)::file> >::emplace_back<std::filesystem::__cxx11::path&, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >&>(std::filesystem::__cxx11::path&, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >&)':
Parser.cpp:(.text+0x1096): undefined reference to `std::filesystem::__cxx11::path::_M_split_cmpts()'
/usr/bin/ld: Parser.cpp:(.text+0x10ad): undefined reference to `std::filesystem::__cxx11::path::_M_split_cmpts()'
/usr/bin/ld: /home/mblatt/src/dune/opm-2.6/opm-common/opm-seq/lib/libopmcommon.a(Parser.cpp.o): in function `Opm::(anonymous namespace)::ParserState::loadFile(std::filesystem::__cxx11::path const&)':
Parser.cpp:(.text+0x23a1): undefined reference to `std::filesystem::canonical(std::filesystem::__cxx11::path const&)'
/usr/bin/ld: Parser.cpp:(.text+0x24e0): undefined reference to
`std::filesystem::__cxx11::path::_M_split_cmpts()'
```
The reason turned out to be that the library path was build up by
paths of the old (g++-7) compiler used by nvcc and the actual (newer) compiler
g++-8. This completely messed up the linker paths for CMake.
To detect this situation already when running cmake we have resorted
to first setting the CMAKE_CUDA_FLAGS to force cmake to make nvcc use
the host compiler and to activate CUDA (if available) before calling
`find_package(CUDA)`. If the host compiler is not supported CMake will
error out during `enable(CUDA)`
Note that we still use (deprecated) FindCUDA later to determine the
libraries to link to.
The users has the option to either deactivate CUDA by setting
`-DCMAKE_DISABLE_FIND_PACKAGE_CUDA=ON` or to use a compiler supported
by nvcc (setting `-DCMAKE_CXX_COMPILER=compiler`).
Additionally we do not try to activate CUDA the CMake version is <
3.8. Please note that previously CMake would have errored out here
anyway since we used the unsupported `enable_language(CUDA)` even in
this case.
Closes#2363.
This fixes name clashes with DUNE's own factory that is introduced
in DUNE 2.7. Hence it closes issue #2266.
BTW: Dune's factory has more template parameters than ours.
... instead of a reference This is needed as we only want to read the
full deck and construct the EclipseGrid only on the master process
with rank 0. Hence all other ranks will pass a nullptr to this
function. This will be possible with this move from a reference to a
pointer.
Markus Blatt