OpenVDB
1.2.0
|
#include <tbb/parallel_for.h>
#include <tbb/parallel_reduce.h>
#include <openvdb/Types.h>
#include <openvdb/Grid.h>
Go to the source code of this file.
Namespaces | |
openvdb | |
openvdb::v1_2_0 | |
openvdb::v1_2_0::tools | |
openvdb::v1_2_0::tools::valxform | |
Constant Groups | |
openvdb | |
openvdb::v1_2_0 | |
openvdb::v1_2_0::tools | |
openvdb::v1_2_0::tools::valxform | |
Functions | |
template<typename IterT , typename XformOp > | |
void | foreach (const IterT &iter, XformOp &op, bool threaded=true, bool shareOp=true) |
template<typename IterT , typename XformOp > | |
void | foreach (const IterT &iter, const XformOp &op, bool threaded=true, bool shareOp=true) |
template<typename InIterT , typename OutGridT , typename XformOp > | |
void | transformValues (const InIterT &inIter, OutGridT &outGrid, XformOp &op, bool threaded=true, bool shareOp=true) |
template<typename InIterT , typename OutGridT , typename XformOp > | |
void | transformValues (const InIterT &inIter, OutGridT &outGrid, const XformOp &op, bool threaded=true, bool shareOp=true) |
template<typename IterT , typename XformOp > | |
void | accumulate (const IterT &iter, XformOp &op, bool threaded=true) |
tools::foreach() and tools::transformValues() transform the values in a grid by iterating over the grid with a user-supplied iterator and applying a user-supplied functor at each step of the iteration. With tools::foreach(), the transformation is done in-place on the input grid, whereas with tools::transformValues(), transformed values are written to an output grid (which can, for example, have a different value type than the input grid). Both functions can optionally transform multiple values of the grid in parallel.
tools::accumulate() can be used to accumulate the results of applying a functor at each step of a grid iteration. (The functor is responsible for storing and updating intermediate results.) When the iteration is done serially the behavior is the same as with tools::foreach(), but when multiple values are processed in parallel, an additional step is performed: when any two threads finish processing, op.join(otherOp)
is called on one thread's functor to allow it to coalesce its intermediate result with the other thread's.