Markus Holzer authored 4 years ago and Dominik Thoennes committed 4 years ago

also clean up some old leftovers nearby

How could this have been working until now?
We should not call `cudaFree` on uninitialized memory.

We have to free the old pointers before resize. This also prevents
memory leakage in case the vector gets shrinked.

Apparently the previous code relied on the fact that `resize` somehow
initialized the memory to null pointers or that `cudaFree` just ignores
pointers it didn't allocate.

staggered boundary generation: direction index should point to fluid cell when inner_or_boundary=False

In that case, the index vector contains the boundary cells, so the direction should point to the fluid cell.
When inner_or_boundary=True, the index vector contains the fluid cells and the direction points to the boundary cell.

Added warning if no SIMD instruction could be determined, adapted field layouts in all tests and benchmarks

This new function does the actual generation using pystencils and creates a target library that can be used as a dependency as usual. One should not list the python script itself anymore in add_executable.
Convention: Use the AppName as a prefix for the generated lib!

Example Usage:

waLBerla_generate_target_from_python( NAME AppNameGeneratedLib FILE GenerateKernel.py OUT_FILES Kernel.h Kernel.cpp)

waLBerla_add_executable ( NAME AppNameGeneratedLibTest DEPENDS GeneratedLib)

Previously, anything that wasn’t velocity, src, src_tmp, pdfs, pdfs_tmp resulted in invalid code

This was a regression introduced in a1614152