-
Notifications
You must be signed in to change notification settings - Fork 90
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
landIce_FO_GIS_AdjointSensitivity_Thickness failing again on several platforms #1065
Comments
I'm on it. My fix apparently only fixed it for GPU builds. |
No problem. Thank you @mcarlson801! |
Alright, I think I've got to the bottom of this. I put a bunch of print statements into my kokkos implementation of Response_GLFlux and the original implementation. The changes that would have introduced this behavior happened in March of 2021 (#671) and the "correct" sensitivity for this test was added to the test input in October of 2021 (9688916). @mperego @bartgol @jewatkins Is it possible that the original sensitivity test value is incorrect for this test? |
@mcarlson801 I think your guess on why x/y have no derivs is correct. If you don't pass a valid DataLayout, I think the MDField is not allocated. // Size of sacado scalar dimension
KOKKOS_FORCEINLINE_FUNCTION constexpr unsigned dimension_scalar() const
{ return m_fad_size.value+1; } So if the FAD length was not set, the scalar has length 1. At this point, I would trust your version. |
@mperego What do you think, should we update the sensitivity test value for this test? |
@mcarlson801, I didn't have a chance yet to take a look at that. I'll have a look at it sometime this week. Sorry for the delay |
@mcarlson801 Max, sorry it took me so long to get back to this. I think what you have is correct. Thanks for fixing this! |
@mcarlson801 did you ever make a PR for this fix? |
Posting it now |
Looks like this is clean now, closing. |
Please see: https://sems-cdash-son.sandia.gov/cdash/test/6161385 . It looks like there is a failed comparison.
@mperego @bartgol @mcarlson801
The text was updated successfully, but these errors were encountered: