Closed kubagalecki closed 9 months ago
Attention: 1243 lines
in your changes are missing coverage. Please review.
Comparison is base (
85ab8ef
) 44.45% compared to head (4dbf028
) 36.63%. Report is 4 commits behind head on unstable.
Files | Patch % | Lines |
---|---|---|
src/ArbLattice.cpp | 0.00% | 431 Missing :warning: |
src/LatticeBase.cpp.Rt | 23.83% | 131 Missing :warning: |
src/toArb.cpp | 0.00% | 126 Missing :warning: |
src/vtkLattice.cpp | 24.17% | 69 Missing :warning: |
src/vtuOutput.cpp | 0.00% | 61 Missing :warning: |
src/ArbLatticeLauncher.hpp.Rt | 0.00% | 53 Missing :warning: |
src/Handlers/acUSAdjoint.cpp | 6.12% | 46 Missing :warning: |
src/CartLattice.cpp.Rt | 39.68% | 38 Missing :warning: |
src/main.cpp | 68.42% | 30 Missing :warning: |
src/Handlers/acRunAction.cpp | 0.00% | 27 Missing :warning: |
... and 31 more |
:exclamation: Your organization needs to install the Codecov GitHub app to enable full functionality.
:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.
@llaniewski As seen in https://github.com/CFD-GO/TCLB/actions/runs/6926854681/job/18839682950, there was a bug in determining unique directions. I've now fixed it, but it exposed a limitation of HIP: the maximum number of unique offset directions is 256, e.g. d3q343 would not compile. This is a result of a low limit of fold expression expansions in clang, see minimum example. There is a way to work around this by writing more R template code, but the question is do we realistically need to support models with q > 256?
There is a way to work around this by writing more R template code
There is also a compiler flag -fbracket-depth=N
to raise the limit
toArb
- export Cartesian case as arbitrary