You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Follow-up on #5068. Run all (if not, a subset of) tests in single-precision, compare against existing double-precision checksum files with a larger tolerance.
The text was updated successfully, but these errors were encountered:
To generalize the analysis scripts, we could change the APIs of the tolerance argument to the checksum API, and pass a dictionary (double and single as keys) for the tolerances. Or have two arguments, one double tolerance one float (with a dictionary, we keep it open to have other types later, but I do not think this is timely.)
Inside the checksum API, we could do a test for WarpX config used or type of a data file entry read to determine if something was a SP or DP run.
Follow-up on #5068. Run all (if not, a subset of) tests in single-precision, compare against existing double-precision checksum files with a larger tolerance.
The text was updated successfully, but these errors were encountered: