cub/thread/thread_operators.cuh

File members: cub/thread/thread_operators.cuh

/***********************************************************************************************************************
 * Copyright (c) 2011, Duane Merrill.  All rights reserved.
 * Copyright (c) 2011-2025, NVIDIA CORPORATION.  All rights reserved.
 *
 * Redistribution and use in source and binary forms, with or without modification, are permitted provided that the
 * following conditions are met:
 *     * Redistributions of source code must retain the above copyright
 *       notice, this list of conditions and the following disclaimer.
 *     * Redistributions in binary form must reproduce the above copyright
 *       notice, this list of conditions and the following disclaimer in the
 *       documentation and/or other materials provided with the distribution.
 *     * Neither the name of the NVIDIA CORPORATION nor the
 *       names of its contributors may be used to endorse or promote products
 *       derived from this software without specific prior written permission.
 *
 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES,
 * INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
 * DISCLAIMED. IN NO EVENT SHALL NVIDIA CORPORATION BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY,
 * OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
 * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 *
 **********************************************************************************************************************/

/******************************************************************************
 * Simple functor operators
 ******************************************************************************/

#pragma once

#include <cub/config.cuh>

#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 <cub/util_type.cuh>

#include <cuda/functional> // cuda::maximum, cuda::minimum
#include <cuda/std/cstdint> // cuda::std::uint32_t
#include <cuda/std/type_traits> // is_same_v

CUB_NAMESPACE_BEGIN

// TODO(bgruber): deprecate in C++17 with a note: "replace by decltype(cuda::std::not_fn(EqualityOp{}))"
template <typename EqualityOp>
struct InequalityWrapper
{
  EqualityOp op;

  _CCCL_HOST_DEVICE _CCCL_FORCEINLINE InequalityWrapper(EqualityOp op)
      : op(op)
  {}

  template <typename T, typename U>
  _CCCL_HOST_DEVICE _CCCL_FORCEINLINE bool operator()(T&& t, U&& u)
  {
    return !op(_CUDA_VSTD::forward<T>(t), _CUDA_VSTD::forward<U>(u));
  }
};

struct ArgMax
{
  template <typename T, typename OffsetT>
  _CCCL_HOST_DEVICE _CCCL_FORCEINLINE KeyValuePair<OffsetT, T>
  operator()(const KeyValuePair<OffsetT, T>& a, const KeyValuePair<OffsetT, T>& b) const
  {
    // Mooch BUG (device reduce argmax gk110 3.2 million random fp32)
    // return ((b.value > a.value) ||
    //         ((a.value == b.value) && (b.key < a.key)))
    //      ? b : a;

    if ((b.value > a.value) || ((a.value == b.value) && (b.key < a.key)))
    {
      return b;
    }

    return a;
  }
};

struct ArgMin
{
  template <typename T, typename OffsetT>
  _CCCL_HOST_DEVICE _CCCL_FORCEINLINE KeyValuePair<OffsetT, T>
  operator()(const KeyValuePair<OffsetT, T>& a, const KeyValuePair<OffsetT, T>& b) const
  {
    // Mooch BUG (device reduce argmax gk110 3.2 million random fp32)
    // return ((b.value < a.value) ||
    //         ((a.value == b.value) && (b.key < a.key)))
    //      ? b : a;

    if ((b.value < a.value) || ((a.value == b.value) && (b.key < a.key)))
    {
      return b;
    }

    return a;
  }
};

namespace detail
{

struct arg_max
{
  template <typename T, typename OffsetT>
  _CCCL_HOST_DEVICE _CCCL_FORCEINLINE ::cuda::std::pair<OffsetT, T>
  operator()(const ::cuda::std::pair<OffsetT, T>& a, const ::cuda::std::pair<OffsetT, T>& b) const
  {
    if ((b.second > a.second) || ((a.second == b.second) && (b.first < a.first)))
    {
      return b;
    }

    return a;
  }
};

struct arg_min
{
  template <typename T, typename OffsetT>
  _CCCL_HOST_DEVICE _CCCL_FORCEINLINE ::cuda::std::pair<OffsetT, T>
  operator()(const ::cuda::std::pair<OffsetT, T>& a, const ::cuda::std::pair<OffsetT, T>& b) const
  {
    if ((b.second < a.second) || ((a.second == b.second) && (b.first < a.first)))
    {
      return b;
    }

    return a;
  }
};

template <typename ScanOpT>
struct ScanBySegmentOp
{
  ScanOpT op;

  _CCCL_HOST_DEVICE _CCCL_FORCEINLINE ScanBySegmentOp() {}

  _CCCL_HOST_DEVICE _CCCL_FORCEINLINE ScanBySegmentOp(ScanOpT op)
      : op(op)
  {}

  template <typename KeyValuePairT>
  _CCCL_HOST_DEVICE _CCCL_FORCEINLINE KeyValuePairT operator()(const KeyValuePairT& first, const KeyValuePairT& second)
  {
    KeyValuePairT retval;
    retval.key = first.key | second.key;
#ifdef _NVHPC_CUDA // WAR bug on nvc++
    if (second.key)
    {
      retval.value = second.value;
    }
    else
    {
      // If second.value isn't copied into a temporary here, nvc++ will
      // crash while compiling the TestScanByKeyWithLargeTypes test in
      // thrust/testing/scan_by_key.cu:
      auto v2      = second.value;
      retval.value = op(first.value, v2);
    }
#else // not nvc++:
    // if (second.key) {
    //   The second partial reduction spans a segment reset, so it's value
    //   aggregate becomes the running aggregate
    // else {
    //   The second partial reduction does not span a reset, so accumulate both
    //   into the running aggregate
    // }
    retval.value = (second.key) ? second.value : op(first.value, second.value);
#endif
    return retval;
  }
};

template <class OpT>
struct basic_binary_op_t
{
  static constexpr bool value = false;
};

template <typename T>
struct basic_binary_op_t<_CUDA_VSTD::plus<T>>
{
  static constexpr bool value = true;
};

template <typename T>
struct basic_binary_op_t<::cuda::minimum<T>>
{
  static constexpr bool value = true;
};

template <typename T>
struct basic_binary_op_t<::cuda::maximum<T>>
{
  static constexpr bool value = true;
};
} // namespace detail

template <typename B>
struct CastOp
{
  template <typename A>
  _CCCL_HOST_DEVICE _CCCL_FORCEINLINE B operator()(A&& a) const
  {
    return (B) a;
  }
};

template <typename ScanOp>
class SwizzleScanOp
{
private:
  ScanOp scan_op;

public:
  _CCCL_HOST_DEVICE _CCCL_FORCEINLINE SwizzleScanOp(ScanOp scan_op)
      : scan_op(scan_op)
  {}

  template <typename T>
  _CCCL_HOST_DEVICE _CCCL_FORCEINLINE T operator()(const T& a, const T& b)
  {
    T _a(a);
    T _b(b);

    return scan_op(_b, _a);
  }
};

template <typename ReductionOpT>
struct ReduceBySegmentOp
{
  ReductionOpT op;

  _CCCL_HOST_DEVICE _CCCL_FORCEINLINE ReduceBySegmentOp() {}

  _CCCL_HOST_DEVICE _CCCL_FORCEINLINE ReduceBySegmentOp(ReductionOpT op)
      : op(op)
  {}

  template <typename KeyValuePairT>
  _CCCL_HOST_DEVICE _CCCL_FORCEINLINE KeyValuePairT operator()(const KeyValuePairT& first, const KeyValuePairT& second)
  {
    KeyValuePairT retval;
    retval.key = first.key + second.key;
#ifdef _NVHPC_CUDA // WAR bug on nvc++
    if (second.key)
    {
      retval.value = second.value;
    }
    else
    {
      // If second.value isn't copied into a temporary here, nvc++ will
      // crash while compiling the TestScanByKeyWithLargeTypes test in
      // thrust/testing/scan_by_key.cu:
      auto v2      = second.value;
      retval.value = op(first.value, v2);
    }
#else // not nvc++:
    // if (second.key) {
    //   The second partial reduction spans a segment reset, so it's value
    //   aggregate becomes the running aggregate
    // else {
    //   The second partial reduction does not span a reset, so accumulate both
    //   into the running aggregate
    // }
    retval.value = (second.key) ? second.value : op(first.value, second.value);
#endif
    return retval;
  }
};

template <typename ReductionOpT>
struct ReduceByKeyOp
{
  ReductionOpT op;

  _CCCL_HOST_DEVICE _CCCL_FORCEINLINE ReduceByKeyOp() {}

  _CCCL_HOST_DEVICE _CCCL_FORCEINLINE ReduceByKeyOp(ReductionOpT op)
      : op(op)
  {}

  template <typename KeyValuePairT>
  _CCCL_HOST_DEVICE _CCCL_FORCEINLINE KeyValuePairT operator()(const KeyValuePairT& first, const KeyValuePairT& second)
  {
    KeyValuePairT retval = second;

    if (first.key == second.key)
    {
      retval.value = op(first.value, retval.value);
    }

    return retval;
  }
};

#ifndef _CCCL_DOXYGEN_INVOKED // Do not document

//----------------------------------------------------------------------------------------------------------------------
// Predefined operators

namespace detail
{

//----------------------------------------------------------------------------------------------------------------------
// Predefined operators

template <typename, typename = void>
inline constexpr bool is_cuda_std_plus_v = false;

template <typename T>
inline constexpr bool is_cuda_std_plus_v<_CUDA_VSTD::plus<T>, void> = true;

template <typename T>
inline constexpr bool is_cuda_std_plus_v<_CUDA_VSTD::plus<T>, T> = true;

template <typename T>
inline constexpr bool is_cuda_std_plus_v<_CUDA_VSTD::plus<>, T> = true;

template <>
inline constexpr bool is_cuda_std_plus_v<_CUDA_VSTD::plus<>, void> = true;

template <typename, typename = void>
inline constexpr bool is_cuda_std_mul_v = false;

template <typename T>
inline constexpr bool is_cuda_std_mul_v<_CUDA_VSTD::multiplies<T>, void> = true;

template <typename T>
inline constexpr bool is_cuda_std_mul_v<_CUDA_VSTD::multiplies<T>, T> = true;

template <typename T>
inline constexpr bool is_cuda_std_mul_v<_CUDA_VSTD::multiplies<>, T> = true;

template <>
inline constexpr bool is_cuda_std_mul_v<_CUDA_VSTD::multiplies<>, void> = true;

template <typename, typename = void>
inline constexpr bool is_cuda_maximum_v = false;

template <typename T>
inline constexpr bool is_cuda_maximum_v<::cuda::maximum<T>, void> = true;

template <typename T>
inline constexpr bool is_cuda_maximum_v<::cuda::maximum<T>, T> = true;

template <typename T>
inline constexpr bool is_cuda_maximum_v<::cuda::maximum<>, T> = true;

template <>
inline constexpr bool is_cuda_maximum_v<::cuda::maximum<>, void> = true;

template <typename, typename = void>
inline constexpr bool is_cuda_minimum_v = false;

template <typename T>
inline constexpr bool is_cuda_minimum_v<::cuda::minimum<T>, void> = true;

template <typename T>
inline constexpr bool is_cuda_minimum_v<::cuda::minimum<T>, T> = true;

template <typename T>
inline constexpr bool is_cuda_minimum_v<::cuda::minimum<>, T> = true;

template <>
inline constexpr bool is_cuda_minimum_v<::cuda::minimum<>, void> = true;

template <typename, typename = void>
inline constexpr bool is_cuda_std_bit_and_v = false;

template <typename T>
inline constexpr bool is_cuda_std_bit_and_v<_CUDA_VSTD::bit_and<T>, void> = true;

template <typename T>
inline constexpr bool is_cuda_std_bit_and_v<_CUDA_VSTD::bit_and<T>, T> = true;

template <typename T>
inline constexpr bool is_cuda_std_bit_and_v<_CUDA_VSTD::bit_and<>, T> = true;

template <>
inline constexpr bool is_cuda_std_bit_and_v<_CUDA_VSTD::bit_and<>, void> = true;

template <typename, typename = void>
inline constexpr bool is_cuda_std_bit_or_v = false;

template <typename T>
inline constexpr bool is_cuda_std_bit_or_v<_CUDA_VSTD::bit_or<T>, void> = true;

template <typename T>
inline constexpr bool is_cuda_std_bit_or_v<_CUDA_VSTD::bit_or<T>, T> = true;

template <typename T>
inline constexpr bool is_cuda_std_bit_or_v<_CUDA_VSTD::bit_or<>, T> = true;

template <>
inline constexpr bool is_cuda_std_bit_or_v<_CUDA_VSTD::bit_or<>, void> = true;

template <typename, typename = void>
inline constexpr bool is_cuda_std_bit_xor_v = false;

template <typename T>
inline constexpr bool is_cuda_std_bit_xor_v<_CUDA_VSTD::bit_xor<T>, void> = true;

template <typename T>
inline constexpr bool is_cuda_std_bit_xor_v<_CUDA_VSTD::bit_xor<T>, T> = true;

template <typename T>
inline constexpr bool is_cuda_std_bit_xor_v<_CUDA_VSTD::bit_xor<>, T> = true;

template <>
inline constexpr bool is_cuda_std_bit_xor_v<_CUDA_VSTD::bit_xor<>, void> = true;

template <typename, typename = void>
inline constexpr bool is_cuda_std_logical_and_v = false;

template <>
inline constexpr bool is_cuda_std_logical_and_v<_CUDA_VSTD::logical_and<bool>, void> = true;

template <>
inline constexpr bool is_cuda_std_logical_and_v<_CUDA_VSTD::logical_and<bool>, bool> = true;

template <>
inline constexpr bool is_cuda_std_logical_and_v<_CUDA_VSTD::logical_and<>, bool> = true;

template <>
inline constexpr bool is_cuda_std_logical_and_v<_CUDA_VSTD::logical_and<>, void> = true;

template <typename, typename = void>
inline constexpr bool is_cuda_std_logical_or_v = false;

template <>
inline constexpr bool is_cuda_std_logical_or_v<_CUDA_VSTD::logical_or<bool>, void> = true;

template <>
inline constexpr bool is_cuda_std_logical_or_v<_CUDA_VSTD::logical_or<bool>, bool> = true;

template <>
inline constexpr bool is_cuda_std_logical_or_v<_CUDA_VSTD::logical_or<>, bool> = true;

template <>
inline constexpr bool is_cuda_std_logical_or_v<_CUDA_VSTD::logical_or<>, void> = true;

template <typename Op, typename T = void>
inline constexpr bool is_cuda_minimum_maximum_v = is_cuda_maximum_v<Op, T> || is_cuda_minimum_v<Op, T>;

template <typename Op, typename T = void>
inline constexpr bool is_cuda_std_plus_mul_v = is_cuda_std_plus_v<Op, T> || is_cuda_std_mul_v<Op, T>;

template <typename Op, typename T = void>
inline constexpr bool is_cuda_std_bitwise_v =
  is_cuda_std_bit_and_v<Op, T> || is_cuda_std_bit_or_v<Op, T> || is_cuda_std_bit_xor_v<Op, T>;

template <typename Op, typename T = void>
inline constexpr bool is_simd_enabled_cuda_operator =
  is_cuda_minimum_maximum_v<Op, T> || //
  is_cuda_std_plus_mul_v<Op, T> || //
  is_cuda_std_bitwise_v<Op, T>;

//----------------------------------------------------------------------------------------------------------------------
// Generalize Operator

template <typename Op, typename>
struct GeneralizeOperator
{
  using type = Op;
};

template <typename T>
struct GeneralizeOperator<_CUDA_VSTD::plus<T>, T>
{
  using type = _CUDA_VSTD::plus<>;
};

template <typename T>
struct GeneralizeOperator<_CUDA_VSTD::bit_and<T>, T>
{
  using type = _CUDA_VSTD::bit_and<>;
};

template <typename T>
struct GeneralizeOperator<_CUDA_VSTD::bit_or<T>, T>
{
  using type = _CUDA_VSTD::bit_or<>;
};

template <typename T>
struct GeneralizeOperator<_CUDA_VSTD::bit_xor<T>, T>
{
  using type = _CUDA_VSTD::bit_xor<>;
};

template <typename T>
struct GeneralizeOperator<::cuda::maximum<T>, T>
{
  using type = ::cuda::maximum<>;
};

template <typename T>
struct GeneralizeOperator<::cuda::minimum<T>, T>
{
  using type = ::cuda::minimum<>;
};

template <typename Op, typename T>
using generalize_operator_t = typename GeneralizeOperator<Op, T>::type;

//----------------------------------------------------------------------------------------------------------------------
// Identity

template <typename Op, typename T>
inline constexpr T identity_v;

template <typename T>
inline constexpr T identity_v<::cuda::minimum<>, T> = _CUDA_VSTD::numeric_limits<T>::max();

template <typename T>
inline constexpr T identity_v<::cuda::minimum<T>, T> = _CUDA_VSTD::numeric_limits<T>::max();

template <typename T>
inline constexpr T identity_v<::cuda::maximum<>, T> = _CUDA_VSTD::numeric_limits<T>::min();

template <typename T>
inline constexpr T identity_v<::cuda::maximum<T>, T> = _CUDA_VSTD::numeric_limits<T>::min();

} // namespace detail

#endif // !_CCCL_DOXYGEN_INVOKED

CUB_NAMESPACE_END