Closed MelReyCG closed 2 months ago
Attention: Patch coverage is 76.37795%
with 60 lines
in your changes missing coverage. Please review.
Project coverage is 56.00%. Comparing base (
0cbc589
) to head (9c93b4d
).
:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.
@CusiniM
It seems that we are trying to only write a file for rank 0 but I don't see when you gather all stencil data there? This would only work for small cases.... In any case, why not using the timehistory and writing in a parallel format instead? @wrtobin opinions?
Besides debugging I don't really see why someone would want to write out transmissibilities.
This proposal is not completely finished, but it now takes into account all connections within all ranks. The CellToCellDataCollection
now can be used for bigger cases.
To explain why we want to output this data, the goal is to compare the transmissibility data from other simulator(s), in order to understand the differences between the simulations.
@MelReyCG Can you resolve conflicts?
@MelReyCG Can you merge develop into this?
Just leaving a note here, no CI step is legitimately failing here. The CUDA builds were working but crash now because of a CI bug since I updated my branch.
This PR needs one more review.
This PR is after the #3247.
This PR aims to add a new component: the
StencilOutput
, that allow to add aCellToCellOutput
task node in the xml to outputConnectionData
(global id pairs, transmissibilities).For now, by using the
writeCSV="1"
attribute, theCellToCellOutput
will output this data in a CSV. Here is an exemple:(aligned here for readability)