This PR switches to calling the SummaryState constructor which is
aware of the value of undefined UDQs (OPM/opm-common#4052) directly.
While here, also sort headers, split some long lines, and prefer
initialisation lists to constructor body assignments.
This commits adds cmake functionality that can
hipify the cuistl framework to support AMD GPUs.
Some tests have been written as HIP does not mirror
CUDA exactly.
CONVERT_CUDA_TO_HIP is the new CMAKE argument.
CMAKE version is increased to include HIP
as a language (3.21 required).
A macro is added to create a layer of indirection
that will make only cuistl files that have been
changed rehipified.
Some BDA stuff is extracted to make sure CUDA
is not accidentally included.
This commit includes the fraction of pore-volume whose CNV targets
are violated as a new per-iteration quantity in the INFOITER file
(--output-extra-convergence-info=iteration), with the column header
"CnvErrPvFrac". We collect the values which are already calculated
in
BlackoilModel<>::getReservoirConvergence()
and store these as a pair of numerator and denominator in the
ConvergenceReport class. Note that we need both the numerator and
the denominator in order to aggregate contributions from multiple
ranks.
While here, also make a few more objects 'const' and calculate
column widths directly instead of the maximum number of characters
in writeConvergenceHeader().
if we use a separate instance the actnum arrays get out of sync. in
particular the two cells with PORV == 0 will not be marked as inactive
in the actnum array of the separate instance. this leads to a segfault
in lookUpData as the field props will have those cells filtered, while the grid
still has the cells active.
Make some changes to Georgs original code:
dynamically allocated arrays with std::vectors instead
Implement new class structure handling what
MPI communication implementation to use
create extra scopes to avoid reuse of index variable i
Update related tests:
Update test_cuowneroverlapcopy to account for new
class strucutre
Also remove line that invalidates the MPI tests for multiple processes
In the test methods of the test suite we construct a new logger that
logs to a local stringstream. This logger will be stored in a static
map and hence live longer that the test method and the stringstream.
Some methods use Opm::Log::MessageType::Note, others
Opm::Log::MessageType::Warning for adding the loggging backend . There
is one map per MessageType and sometimes (e.g. previous method used Note
and this one uses Warning) we will log to the old logger with dangling
references.
This problem materialized in a segmentation fault on ppc64el architecture.
The initial use case is calculating the phase-filled pore-volume
weighted average of the fluid mass densities per PVT region. This
value goes into calculating depth-corrected per-cell phase pressure
values such as the BPPO and BPPG summary vectors.
This class manages a single linear array which separately tracks the
averages' numerators and denominators as running sums per region and
region set. We pick this data structure to simplify the cross-rank
reduction needed in MPI parallel runs. Client code is expected to
add individual per-cell and per-phase contributions using the
addCell() member function and then call the accumulateParallel()
member to affect the cross-rank reduction. The averages will then
be available through the fieldValue() and value() member functions.
As a further view towards the initial use case, we track two
different types of average per phase--one for the phase-filled
volume and one for the pore-volume filled volume. The latter is the
average we would get for the case of the phase saturation being one
throughout the region. This alternative value is the fallback
option for the case of the phase saturation being identically zero
throughout the region.
We accomplish that by passing the module version as a string to the
constructors of LogOutputHelper and EclGenericOutputBlackoilModel
instead of calling moduleVersionName() in LogOutputHelper. That way
moduleVersionName is not needed by libopmsimulators anymore and
compilation works again for people requesting shared libraries via
CMake's BUILD_SHARED_LIBS variable.
preconditioner. Uses graph coloring to exploit
parallelism in upper and triangular solves when
computing a diagonal approximate inverse of a
sparse matrix. Supports blocksizes up to 3.
This commit ensures that we have backing support for region IDs up
to and including NTFIP (TABDIMS(5), REGDIMS(1)). The existing setup
would fail (segmentation violation) in the case of a summary vector
of the form
ROFT
36 31 /
/
when the maximum FIPNUM value was 30. We nevertheless support
maximum FIPNUM values exceeding NTFIP.
We add a new optional parameter to the EclInterRegionFlowMap
constructor. The parameter allows client code to specifiy a
"minimum maximum" region ID that all ranks must support. This value
will be enforced during parallel aggregation.
There is no need to restrict PRT file flow reports to those wells
which are active on the current rank since the SummaryState object
holds information for all wells active at the current report step.
This commit splits the production, injection, and cumulative
*Report_() functions into three logically distinct parts, one for
outputting the report header (begin*Report_()), one for outputting a
single report record (output*ReportRecord_()), and one for ending
the report (end*Report_()). This simplifies the logic of the
*Record_() member functions since they no longer need to infer the
context which is already available in the caller and can perform a
more narrow task than before.
With this separation we're also able to remove the dashed lines
which would previously separate each report record, thereby creating
PRT file report sheets which have a more expected layout.
Moreover, as an aid to future maintenance, we also factor out common
code for the well and group cases in each *Record_() function.
Finally, fix a unit conversion problem in the report values for
cumulative gas quantities. The sheet header states that we should
be outputting values in 'MM' prefixed units, but we were only
scaling the gas values by a factor of 1000 instead of 1000*1000. In
other words, the injection and production gas values in the
cumulative sheet were off by a factor of 1000.
The StreamLog::addMessageUnconditionally() member function will end
each message with a newline (std::endl) so we should not add such
newlines ourselves. The extra newline characters produce spurious
blank lines in the report sheets, e.g., for the "PRODUCTION REPORT".
This commit removes the last newline character from each report
request, thus deferring that responsibility to OpmLog::note()
instead. Doing so, however, means we have take a little more care
with the first line of each report lest we create report sheets
which are smushed together.
Implement graphcoloring to expose rows in level sets that that can be
executed in parallel during the sparse triangular solves.
Add copy of A matrix that is reordered to ensure continuous memory reads
when traversing the matrix in level set order.
TODO: add number of threads available as constructor argument in DILU
Implement calls to cuBlas, cuSparse and implement necessary
CUDA kernels to perform a single iteration of the jacobi preconditioner.
Add tests that verify new kernels and the preconditioner in its totality.
The preconditioner is verified on 2x2 and 3x3 blocks, which as of now
are the only supported sizes. 1x1 are not supported because cuSparse
does not support it.
This commit adds a parallel calculation object derived from the serial
PAvgCalculator class. This parallel version is aware of MPI
communicators and knows how to aggregate contributions from wells that
might be distributed across ranks.
We also add a wrapper class, ParallelWBPCalculation, which knows how to
exchange information from PAvgCalculatorCollection objects on different
ranks and, especially, how to properly prune inactive cells/connections.
This commit adds a new container class,
ParallelPAvgDynamicSourceData
which inherits from PAvgDynamicSourceData and provides a parallel
view of source contributions. Member function
collectLocalSources
will call the user-provided source term evaluation function for each
source location in its purview--typically those locations owned by
the current MPI rank. Those values will be distributed to other MPI
ranks through member function synchroniseSources which will fill the
base class' 'src_' data member, and become available to clients
through read-only item spans.
