Fix a bug in `tovtkm::ConvertSingleType`
There was a bug, where the worklet `ReorderHex` was being explicitly marked as device only (not using VTKM_EXEC) when CUDA was enabled. However, in `BuildSingleTypeVoxelCellSetVisitor` where the worklet is invoked, the Cuda device was explicitly being disabled. This results in undefined behaviour. This path is only trigered when converting an unstructured grid containing all Voxels to vtkm DataSet.
Please register or sign in to comment