Failing regression tests for data layout optimization builds
Some regression tests fail when using a DLO-enabled build: prepace_long, region_vs_gradient_based_heterogeneities-serial and parallel.
I suspect they fail because of SV accesses. (For example, when adjust_MIIF_variables
is used, the arguments to sv_put
should be adjusted to account for the vector size (using IIF->get_type()->dlo_vector_size()
for example).
Some of the tests fail validation, or crash in a different way. The problem is also most likely tied to the manipulation of state variables in a different part of the code.
As I will soon stop working on openCARP I am not able to fix those bugs for now, so I'm leaving this issue so it isn't forgotten about.