loops.h File Reference

#include "pxr/pxr.h"
#include "pxr/base/work/api.h"
#include "pxr/base/work/dispatcher.h"
#include "pxr/base/work/impl.h"
#include "pxr/base/work/threadLimits.h"
#include "pxr/base/tf/errorTransport.h"
#include "pxr/base/tf/mallocTag.h"
#include <algorithm>

Include dependency graph for loops.h:

This graph shows which files directly or indirectly include this file:

Go to the source code of this file.

Classes
class	Work_LoopsTaskWrapper< Fn >
class	Work_MallocTagsLoopsTaskWrapper< Fn >
class	Work_LoopsForEachTaskWrapper< Fn >
class	Work_MallocTagsLoopsForEachTaskWrapper< Fn >

Typedefs
using	Work_ErrorTransports = tbb::concurrent_vector< TfErrorTransport >

Functions
template<typename Fn >
void	WorkSerialForN (size_t n, Fn &&fn)
template<typename Fn >
void	WorkParallelForN (size_t n, Fn &&callback, size_t grainSize)
template<typename Fn >
void	WorkParallelForN (size_t n, Fn &&callback)
template<typename RangeType , typename Fn >
void	WorkParallelForTBBRange (const RangeType &range, Fn &&callback)
template<typename InputIterator , typename Fn >
void	WorkParallelForEach (InputIterator first, InputIterator last, Fn &&fn)

Typedef Documentation

using Work_ErrorTransports = tbb::concurrent_vector<TfErrorTransport>

Definition at line 24 of file loops.h.

Function Documentation

template<typename InputIterator , typename Fn >

void WorkParallelForEach ( InputIterator  first, InputIterator  last, Fn &&  fn  )

inline

WorkParallelForEach(Iterator first, Iterator last, CallbackType callback)

Callback must be of the form:

void LoopCallback(T elem);

where the type T is deduced from the type of the InputIterator template argument.

Definition at line 315 of file loops.h.

template<typename Fn >

void WorkParallelForN	(	size_t	n,
		Fn &&	callback,
		size_t	grainSize
	)

WorkParallelForN(size_t n, CallbackType callback, size_t grainSize = 1)

Runs callback in parallel over the range 0 to n.

Callback must be of the form:

void LoopCallback(size_t begin, size_t end);

grainSize specifies a minimum amount of work to be done per-thread. There is overhead to launching a thread (or task) and a typical guideline is that you want to have at least 10,000 instructions to count for the overhead of launching a thread.

Definition at line 168 of file loops.h.

template<typename Fn >

void WorkParallelForN	(	size_t	n,
		Fn &&	callback
	)

WorkParallelForN(size_t n, CallbackType callback, size_t grainSize = 1)

Runs callback in parallel over the range 0 to n.

Callback must be of the form:

void LoopCallback(size_t begin, size_t end);

Definition at line 210 of file loops.h.

template<typename RangeType , typename Fn >

void WorkParallelForTBBRange	(	const RangeType &	range,
		Fn &&	callback
	)

WorkParallelForTBBRange(const RangeType &r, Fn &&callback)

Runs callback in parallel over a RangeType that adheres to TBB's splittable range requirements: https://oneapi-spec.uxlfoundation.org/specifications/oneapi/latest/elements/onetbb/source/named_requirements/algorithms/range

Callback must be of the form:

void LoopCallback(RangeType range);

Definition at line 230 of file loops.h.

template<typename Fn >

void WorkSerialForN	(	size_t	n,
		Fn &&	fn
	)

WorkSerialForN(size_t n, CallbackType callback)

A serial version of WorkParallelForN as a drop in replacement to selectively turn off multithreading for a single parallel loop for easier debugging.

Callback must be of the form:

void LoopCallback(size_t begin, size_t end);

Definition at line 146 of file loops.h.