Closed stvdwtt closed 2 years ago
I'm looking into the integration test error and will push a fix when I have a chance. I think it might be just that I changed the test input file to be a larger problem and to include localization, but that messed with the expected integration test result.
Still times out
Odd -- I'll take a closer look at it this afternoon.
Still times out but it goes further. We have two choices:
-j2
or -j4
otherwise it will take too long to run the CI.-E integration_3d
to https://github.com/adamantine-sim/adamantine/blob/master/ci/jenkins_config#L74 should do the trick.What do you think?
I was able to trim the DA integration test runtime by a factor of two (at least on my machine), so hopefully that won't be an issue anymore. I also added the "-j 4" flag to the ctest command as you suggested. If we think there could still be some issues I can split up the integration tests, but for now I was going for the lightest touch that would work.
(And even though I had a commit that upped the timeout to 3h, I backed it down to 2h in the latest commit -- it just seems unreasonable to have tests that run for that long.)
This removes the last dense matrix operations in
DataAssimilator
by adding support for localization. As expected, this permits much larger meshes (at least 8x, I haven't tried larger yet), depending on the localization cutoff length.