Describe the bug
DaCe was/will be updated with PR #372. In that process, we updated parts of the schedule tree bridge inside DaCe which determines how the generated strees/SDFGs look on the NDSL side. This change broke experimental NDSL-side optimization passes.
In this issue, we keep track of the few tests that we skipped during the DaCe v2 update. We'd like to fix those in a follow-up PR (after #372). This issue merely serves as a reminder to do so and not forget.
To Reproduce
Unskip the tests which are annotated with this issue (e.g. search for the issue number or url of this issue). Then, run the tests and see them break.
Expected behavior
Tests are expected to pass tough they might need adaption to the new SDFG format (e.g. control flow graphs, in particular LoopRegions).
System Environment
Nothing in particular. Can be reproduced with the CI.
Additional context
Describe the bug
DaCe was/will be updated with PR #372. In that process, we updated parts of the schedule tree bridge inside DaCe which determines how the generated strees/SDFGs look on the NDSL side. This change broke experimental NDSL-side optimization passes.
In this issue, we keep track of the few tests that we skipped during the DaCe v2 update. We'd like to fix those in a follow-up PR (after #372). This issue merely serves as a reminder to do so and not forget.
To Reproduce
Unskip the tests which are annotated with this issue (e.g. search for the issue number or url of this issue). Then, run the tests and see them break.
Expected behavior
Tests are expected to pass tough they might need adaption to the new SDFG format (e.g. control flow graphs, in particular
LoopRegions).System Environment
Nothing in particular. Can be reproduced with the CI.
Additional context