4c365b8c03
Add complete CCCL CUB header tree (1394 files) to cccl_preload/include/: - cub/device/ — DeviceRadixSort, DeviceScan, DeviceHistogram, DeviceReduce, DeviceSelect - cub/agent/ — all agent implementations (sort, scan, reduce, histogram, etc) - cub/block/ — BlockScan, BlockReduce, BlockExchange, BlockLoad, BlockStore, etc - cub/warp/ — WarpScan, WarpReduce, WarpExchange, WarpMergeSort - cub/thread/ — thread-level operators - thrust/ — sort_by_key, iterator utilities - cuda/ — execution, stream, memory_resource, functional New kernel: cccl_moe_sort_scatter.cu - Uses CUB DeviceRadixSort::SortPairs to sort (expert_id, token_idx) pairs - O(n) radix sort replaces O(n log n) torch.argsort in MoE prefill path - Boundary detection + fill for expert offsets/sizes - Compiled against CCCL upstream headers (not corex CUB) to avoid BI-V100 bugs Previously only 288 CCCL headers (CachingDeviceAllocator only). Now 1394 headers — full CUB device-level algorithm stack available for all future kernels.
197 lines
7.5 KiB
C++
197 lines
7.5 KiB
C++
// SPDX-FileCopyrightText: Copyright (c) 2008-2013, NVIDIA Corporation. All rights reserved.
|
|
// SPDX-License-Identifier: Apache-2.0
|
|
|
|
/*! \file generate.h
|
|
* \brief Fills a range with values "generated" from a function of no arguments
|
|
*/
|
|
|
|
#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 transformations
|
|
* \{
|
|
*/
|
|
|
|
/*! \p generate assigns the result of invoking \p gen, a function object that takes no arguments,
|
|
* to each element in the range <tt>[first,last)</tt>.
|
|
*
|
|
* The algorithm's execution is parallelized as determined by \p exec.
|
|
*
|
|
* \param exec The execution policy to use for parallelization.
|
|
* \param first The first element in the range of interest.
|
|
* \param last The last element in the range of interest.
|
|
* \param gen A function argument, taking no parameters, used to generate values to assign to
|
|
* elements in the range <tt>[first,last)</tt>.
|
|
*
|
|
* \tparam DerivedPolicy The name of the derived execution policy.
|
|
* \tparam ForwardIterator is a model of <a href="https://en.cppreference.com/w/cpp/iterator/forward_iterator">Forward
|
|
* Iterator</a>, and \p ForwardIterator is mutable.
|
|
* \tparam Generator A nullary function with a return type convertible to \p ForwardIterator's \c value_type.
|
|
*
|
|
* The following code snippet demonstrates how to fill a \c host_vector with random numbers,
|
|
* using the standard C library function \c rand using the \p thrust::host execution policy for parallelization:
|
|
*
|
|
* \code
|
|
* #include <thrust/generate.h>
|
|
* #include <thrust/host_vector.h>
|
|
* #include <thrust/execution_policy.h>
|
|
* #include <cstdlib>
|
|
* ...
|
|
* thrust::host_vector<int> v(10);
|
|
* srand(13);
|
|
* thrust::generate(thrust::host, v.begin(), v.end(), rand);
|
|
*
|
|
* // the elements of v are now pseudo-random numbers
|
|
* \endcode
|
|
*
|
|
* \see generate_n
|
|
* \see https://en.cppreference.com/w/cpp/algorithm/generate
|
|
*
|
|
* \verbatim embed:rst:leading-asterisk
|
|
* .. versionadded:: 2.2.0
|
|
* \endverbatim
|
|
*/
|
|
template <typename DerivedPolicy, typename ForwardIterator, typename Generator>
|
|
_CCCL_HOST_DEVICE void
|
|
generate(const thrust::detail::execution_policy_base<DerivedPolicy>& exec,
|
|
ForwardIterator first,
|
|
ForwardIterator last,
|
|
Generator gen);
|
|
|
|
/*! \p generate assigns the result of invoking \p gen, a function object that takes no arguments,
|
|
* to each element in the range <tt>[first,last)</tt>.
|
|
*
|
|
* \param first The first element in the range of interest.
|
|
* \param last The last element in the range of interest.
|
|
* \param gen A function argument, taking no parameters, used to generate values to assign to
|
|
* elements in the range <tt>[first,last)</tt>.
|
|
*
|
|
* \tparam ForwardIterator is a model of <a href="https://en.cppreference.com/w/cpp/iterator/forward_iterator">Forward
|
|
* Iterator</a>, and \p ForwardIterator is mutable.
|
|
* \tparam Generator A nullary function with a return type convertible to \p ForwardIterator's \c value_type.
|
|
*
|
|
* The following code snippet demonstrates how to fill a \c host_vector with random numbers,
|
|
* using the standard C library function \c rand.
|
|
*
|
|
* \code
|
|
* #include <thrust/generate.h>
|
|
* #include <thrust/host_vector.h>
|
|
* #include <thrust/execution_policy.h>
|
|
* #include <cstdlib>
|
|
* ...
|
|
* thrust::host_vector<int> v(10);
|
|
* srand(13);
|
|
* thrust::generate(v.begin(), v.end(), rand);
|
|
*
|
|
* // the elements of v are now pseudo-random numbers
|
|
* \endcode
|
|
*
|
|
* \see generate_n
|
|
* \see https://en.cppreference.com/w/cpp/algorithm/generate
|
|
*
|
|
* \verbatim embed:rst:leading-asterisk
|
|
* .. versionadded:: 2.2.0
|
|
* \endverbatim
|
|
*/
|
|
template <typename ForwardIterator, typename Generator>
|
|
void generate(ForwardIterator first, ForwardIterator last, Generator gen);
|
|
|
|
/*! \p generate_n assigns the result of invoking \p gen, a function object that takes no arguments,
|
|
* to each element in the range <tt>[first,first + n)</tt>. The return value is <tt>first + n</tt>.
|
|
*
|
|
* The algorithm's execution is parallelized as determined by \p exec.
|
|
*
|
|
* \param exec The execution policy to use for parallelization.
|
|
* \param first The first element in the range of interest.
|
|
* \param n The size of the range of interest.
|
|
* \param gen A function argument, taking no parameters, used to generate values to assign to
|
|
* elements in the range <tt>[first,first + n)</tt>.
|
|
*
|
|
* \tparam DerivedPolicy The name of the derived execution policy.
|
|
* \tparam OutputIterator is a model of <a href="https://en.cppreference.com/w/cpp/iterator/output_iterator">Output
|
|
* Iterator</a>. \tparam Size is an integral type (either signed or unsigned).
|
|
* \tparam Generator A nullary function with a return type convertible to \p ForwardIterator's \c value_type.
|
|
*
|
|
* The following code snippet demonstrates how to fill a \c host_vector with random numbers,
|
|
* using the standard C library function \c rand using the \p thrust::host execution policy for parallelization:
|
|
*
|
|
* \code
|
|
* #include <thrust/generate.h>
|
|
* #include <thrust/host_vector.h>
|
|
* #include <thrust/execution_policy.h>
|
|
* #include <cstdlib>
|
|
* ...
|
|
* thrust::host_vector<int> v(10);
|
|
* srand(13);
|
|
* thrust::generate_n(thrust::host, v.begin(), 10, rand);
|
|
*
|
|
* // the elements of v are now pseudo-random numbers
|
|
* \endcode
|
|
*
|
|
* \see generate
|
|
* \see https://en.cppreference.com/w/cpp/algorithm/generate
|
|
*
|
|
* \verbatim embed:rst:leading-asterisk
|
|
* .. versionadded:: 2.2.0
|
|
* \endverbatim
|
|
*/
|
|
template <typename DerivedPolicy, typename OutputIterator, typename Size, typename Generator>
|
|
_CCCL_HOST_DEVICE OutputIterator generate_n(
|
|
const thrust::detail::execution_policy_base<DerivedPolicy>& exec, OutputIterator first, Size n, Generator gen);
|
|
|
|
/*! \p generate_n assigns the result of invoking \p gen, a function object that takes no arguments,
|
|
* to each element in the range <tt>[first,first + n)</tt>. The return value is <tt>first + n</tt>.
|
|
*
|
|
* \param first The first element in the range of interest.
|
|
* \param n The size of the range of interest.
|
|
* \param gen A function argument, taking no parameters, used to generate values to assign to
|
|
* elements in the range <tt>[first,first + n)</tt>.
|
|
*
|
|
* \tparam OutputIterator is a model of <a href="https://en.cppreference.com/w/cpp/iterator/output_iterator">Output
|
|
* Iterator</a>. \tparam Size is an integral type (either signed or unsigned).
|
|
* \tparam Generator A nullary function with a return type convertible to \p ForwardIterator's \c value_type.
|
|
*
|
|
* The following code snippet demonstrates how to fill a \c host_vector with random numbers,
|
|
* using the standard C library function \c rand.
|
|
*
|
|
* \code
|
|
* #include <thrust/generate.h>
|
|
* #include <thrust/host_vector.h>
|
|
* #include <stdlib.h>
|
|
* ...
|
|
* thrust::host_vector<int> v(10);
|
|
* srand(13);
|
|
* thrust::generate_n(v.begin(), 10, rand);
|
|
*
|
|
* // the elements of v are now pseudo-random numbers
|
|
* \endcode
|
|
*
|
|
* \see generate
|
|
* \see https://en.cppreference.com/w/cpp/algorithm/generate
|
|
*
|
|
* \verbatim embed:rst:leading-asterisk
|
|
* .. versionadded:: 2.2.0
|
|
* \endverbatim
|
|
*/
|
|
template <typename OutputIterator, typename Size, typename Generator>
|
|
OutputIterator generate_n(OutputIterator first, Size n, Generator gen);
|
|
|
|
/*! \} // end transformations
|
|
*/
|
|
|
|
THRUST_NAMESPACE_END
|
|
|
|
#include <thrust/detail/generate.inl>
|