CCCL (CUDA C++ Core Libraries) provides: - CUB: device/block/warp-level GPU primitives (reduce, scan, sort, topk) - Thrust: high-level parallel algorithms (transform_reduce, sort, scan) - libcudacxx: CUDA C++ standard library (atomics, barriers, memory) - cudax: experimental features (memory resources, allocators) - Tuning policies: per-SM hardware-specific algorithm parameters Competition optimization vectors mapped to CCCL: - Output TPS (83% weight): warp_reduce, block_reduce, device_topk - Input TPS (14% weight): device_scan, block_load, prefetch - Cache TPS (3% weight): prefix caching strategy patterns - Memory (0.9 util): pooled/cached/buddy allocators Source: https://github.com/NVIDIA/cccl (shallow clone, HEAD only) License: Apache-2.0
186 lines
6.4 KiB
C++
186 lines
6.4 KiB
C++
// SPDX-FileCopyrightText: Copyright (c) 2008-2020, NVIDIA Corporation. All rights reserved.
|
|
// SPDX-License-Identifier: Apache-2.0
|
|
|
|
/*! \file shuffle.h
|
|
* \brief Reorders range by a uniform random permutation
|
|
*/
|
|
|
|
#pragma once
|
|
|
|
#include <thrust/detail/config.h>
|
|
|
|
#if defined(_CCCL_IMPLICIT_SYSTEM_HEADER_GCC)
|
|
# pragma GCC system_header
|
|
#elif defined(_CCCL_IMPLICIT_SYSTEM_HEADER_CLANG)
|
|
# pragma clang system_header
|
|
#elif defined(_CCCL_IMPLICIT_SYSTEM_HEADER_MSVC)
|
|
# pragma system_header
|
|
#endif // no system header
|
|
|
|
#include <thrust/detail/execution_policy.h>
|
|
|
|
THRUST_NAMESPACE_BEGIN
|
|
|
|
/*! \addtogroup reordering
|
|
* \ingroup algorithms
|
|
*
|
|
* \addtogroup shuffling
|
|
* \ingroup reordering
|
|
* \{
|
|
*/
|
|
|
|
/*! \p shuffle reorders the elements <tt>[first, last)</tt> by a uniform pseudorandom permutation, defined by
|
|
* random engine \p g.
|
|
*
|
|
* The algorithm's execution is parallelized as determined by \p exec.
|
|
*
|
|
* \param exec The execution policy to use for parallelization.
|
|
* \param first The beginning of the sequence to shuffle.
|
|
* \param last The end of the sequence to shuffle.
|
|
* \param g A UniformRandomBitGenerator
|
|
*
|
|
* \tparam DerivedPolicy The name of the derived execution policy.
|
|
* \tparam RandomIterator is a random access iterator
|
|
* \tparam URBG is a uniform random bit generator
|
|
*
|
|
* The following code snippet demonstrates how to use \p shuffle to create a random permutation
|
|
* using the \p thrust::host execution policy for parallelization:
|
|
*
|
|
* \code
|
|
* #include <thrust/shuffle.h>
|
|
* #include <thrust/random.h>
|
|
* #include <thrust/execution_policy.h>
|
|
* int A[] = {1, 2, 3, 4, 5, 6, 7, 8, 9, 10};
|
|
* const int N = sizeof(A)/sizeof(int);
|
|
* thrust::default_random_engine g;
|
|
* thrust::shuffle(thrust::host, A, A + N, g);
|
|
* // A is now {6, 5, 8, 7, 2, 1, 4, 3, 10, 9}
|
|
* \endcode
|
|
*
|
|
* \see \p shuffle_copy
|
|
*
|
|
* \verbatim embed:rst:leading-asterisk
|
|
* .. versionadded:: 2.2.0
|
|
* \endverbatim
|
|
*/
|
|
template <typename DerivedPolicy, typename RandomIterator, typename URBG>
|
|
_CCCL_HOST_DEVICE void shuffle(
|
|
const thrust::detail::execution_policy_base<DerivedPolicy>& exec, RandomIterator first, RandomIterator last, URBG&& g);
|
|
|
|
/*! \p shuffle reorders the elements <tt>[first, last)</tt> by a uniform pseudorandom permutation, defined by
|
|
* random engine \p g.
|
|
*
|
|
* \param first The beginning of the sequence to shuffle.
|
|
* \param last The end of the sequence to shuffle.
|
|
* \param g A UniformRandomBitGenerator
|
|
*
|
|
* \tparam RandomIterator is a random access iterator
|
|
* \tparam URBG is a uniform random bit generator
|
|
*
|
|
* The following code snippet demonstrates how to use \p shuffle to create a random permutation.
|
|
*
|
|
* \code
|
|
* #include <thrust/shuffle.h>
|
|
* #include <thrust/random.h>
|
|
* int A[] = {1, 2, 3, 4, 5, 6, 7, 8, 9, 10};
|
|
* const int N = sizeof(A)/sizeof(int);
|
|
* thrust::default_random_engine g;
|
|
* thrust::shuffle(A, A + N, g);
|
|
* // A is now {6, 5, 8, 7, 2, 1, 4, 3, 10, 9}
|
|
* \endcode
|
|
*
|
|
* \see \p shuffle_copy
|
|
*
|
|
* \verbatim embed:rst:leading-asterisk
|
|
* .. versionadded:: 2.2.0
|
|
* \endverbatim
|
|
*/
|
|
template <typename RandomIterator, typename URBG>
|
|
_CCCL_HOST_DEVICE void shuffle(RandomIterator first, RandomIterator last, URBG&& g);
|
|
|
|
/*! shuffle_copy differs from shuffle only in that the reordered sequence is written to different output sequences,
|
|
rather than in place.
|
|
* \p shuffle_copy reorders the elements <tt>[first, last)</tt> by a uniform pseudorandom permutation, defined by
|
|
* random engine \p g.
|
|
*
|
|
* The algorithm's execution is parallelized as determined by \p exec.
|
|
|
|
* \param exec The execution policy to use for parallelization.
|
|
* \param first The beginning of the sequence to shuffle.
|
|
* \param last The end of the sequence to shuffle.
|
|
* \param result Destination of shuffled sequence
|
|
* \param g A UniformRandomBitGenerator
|
|
*
|
|
* \tparam DerivedPolicy The name of the derived execution policy.
|
|
* \tparam RandomIterator is a random access iterator
|
|
* \tparam OutputIterator is a model of <a href="https://en.cppreference.com/w/cpp/iterator/output_iterator">Output
|
|
Iterator</a>.
|
|
* \tparam URBG is a uniform random bit generator
|
|
*
|
|
* The following code snippet demonstrates how to use \p shuffle_copy to create a random permutation.
|
|
*
|
|
* \code
|
|
* #include <thrust/shuffle.h>
|
|
* #include <thrust/random.h>
|
|
* #include <thrust/execution_policy.h>
|
|
* int A[] = {1, 2, 3, 4, 5, 6, 7, 8, 9, 10};
|
|
* int result[10];
|
|
* const int N = sizeof(A)/sizeof(int);
|
|
* thrust::default_random_engine g;
|
|
* thrust::shuffle_copy(thrust::host, A, A + N, result, g);
|
|
* // result is now {6, 5, 8, 7, 2, 1, 4, 3, 10, 9}
|
|
* \endcode
|
|
*
|
|
* \see \p shuffle
|
|
*
|
|
* \verbatim embed:rst:leading-asterisk
|
|
* .. versionadded:: 2.2.0
|
|
* \endverbatim
|
|
*/
|
|
template <typename DerivedPolicy, typename RandomIterator, typename OutputIterator, typename URBG>
|
|
_CCCL_HOST_DEVICE void shuffle_copy(
|
|
const thrust::detail::execution_policy_base<DerivedPolicy>& exec,
|
|
RandomIterator first,
|
|
RandomIterator last,
|
|
OutputIterator result,
|
|
URBG&& g);
|
|
|
|
/*! shuffle_copy differs from shuffle only in that the reordered sequence is written to different output sequences,
|
|
*rather than in place. \p shuffle_copy reorders the elements <tt>[first, last)</tt> by a uniform pseudorandom
|
|
*permutation, defined by random engine \p g.
|
|
*
|
|
* \param first The beginning of the sequence to shuffle.
|
|
* \param last The end of the sequence to shuffle.
|
|
* \param result Destination of shuffled sequence
|
|
* \param g A UniformRandomBitGenerator
|
|
*
|
|
* \tparam RandomIterator is a random access iterator
|
|
* \tparam OutputIterator is a model of <a href="https://en.cppreference.com/w/cpp/iterator/output_iterator">Output
|
|
*Iterator</a>. \tparam URBG is a uniform random bit generator
|
|
*
|
|
* The following code snippet demonstrates how to use \p shuffle_copy to create a random permutation.
|
|
*
|
|
* \code
|
|
* #include <thrust/shuffle.h>
|
|
* #include <thrust/random.h>
|
|
* int A[] = {1, 2, 3, 4, 5, 6, 7, 8, 9, 10};
|
|
* int result[10];
|
|
* const int N = sizeof(A)/sizeof(int);
|
|
* thrust::default_random_engine g;
|
|
* thrust::shuffle_copy(A, A + N, result, g);
|
|
* // result is now {6, 5, 8, 7, 2, 1, 4, 3, 10, 9}
|
|
* \endcode
|
|
*
|
|
* \see \p shuffle
|
|
*
|
|
* \verbatim embed:rst:leading-asterisk
|
|
* .. versionadded:: 2.2.0
|
|
* \endverbatim
|
|
*/
|
|
template <typename RandomIterator, typename OutputIterator, typename URBG>
|
|
_CCCL_HOST_DEVICE void shuffle_copy(RandomIterator first, RandomIterator last, OutputIterator result, URBG&& g);
|
|
|
|
THRUST_NAMESPACE_END
|
|
|
|
#include <thrust/detail/shuffle.inl>
|