cub/block/block_raking_layout.cuh

File members: cub/block/block_raking_layout.cuh

/******************************************************************************
 * Copyright (c) 2011, Duane Merrill.  All rights reserved.
 * Copyright (c) 2011-2018, 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.
 *
 ******************************************************************************/

#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>

CUB_NAMESPACE_BEGIN

template <typename T, int BLOCK_THREADS, int LEGACY_PTX_ARCH = 0>
struct BlockRakingLayout
{
  //---------------------------------------------------------------------
  // Constants and type definitions
  //---------------------------------------------------------------------

  enum
  {
    SHARED_ELEMENTS = BLOCK_THREADS,

    MAX_RAKING_THREADS = CUB_MIN(BLOCK_THREADS, CUB_WARP_THREADS(0)),

    SEGMENT_LENGTH = (SHARED_ELEMENTS + MAX_RAKING_THREADS - 1) / MAX_RAKING_THREADS,

    RAKING_THREADS = (SHARED_ELEMENTS + SEGMENT_LENGTH - 1) / SEGMENT_LENGTH,

    HAS_CONFLICTS = (CUB_SMEM_BANKS(0) % SEGMENT_LENGTH == 0),

    CONFLICT_DEGREE = (HAS_CONFLICTS) ? (MAX_RAKING_THREADS * SEGMENT_LENGTH) / CUB_SMEM_BANKS(0) : 1,

    USE_SEGMENT_PADDING = ((SEGMENT_LENGTH & 1) == 0) && (SEGMENT_LENGTH > 2),

    GRID_ELEMENTS = RAKING_THREADS * (SEGMENT_LENGTH + USE_SEGMENT_PADDING),

    UNGUARDED = (SHARED_ELEMENTS % RAKING_THREADS == 0),
  };

  struct __align__(16) _TempStorage
  {
    T buff[BlockRakingLayout::GRID_ELEMENTS];
  };

  struct TempStorage : Uninitialized<_TempStorage>
  {};

  static _CCCL_DEVICE _CCCL_FORCEINLINE T* PlacementPtr(TempStorage& temp_storage, unsigned int linear_tid)
  {
    // Offset for partial
    unsigned int offset = linear_tid;

    // Add in one padding element for every segment
    if (USE_SEGMENT_PADDING > 0)
    {
      offset += offset / SEGMENT_LENGTH;
    }

    // Incorporating a block of padding partials every shared memory segment
    return temp_storage.Alias().buff + offset;
  }

  static _CCCL_DEVICE _CCCL_FORCEINLINE T* RakingPtr(TempStorage& temp_storage, unsigned int linear_tid)
  {
    return temp_storage.Alias().buff + (linear_tid * (SEGMENT_LENGTH + USE_SEGMENT_PADDING));
  }
};

CUB_NAMESPACE_END