/home/docs/checkouts/readthedocs.org/user_builds/advanced-micro-devices-composable-kernel/checkouts/develop/include/ck_tile/ops/reduce/pipeline/reduce2d_shape.hpp Source File

/home/docs/checkouts/readthedocs.org/user_builds/advanced-micro-devices-composable-kernel/checkouts/develop/include/ck_tile/ops/reduce/pipeline/reduce2d_shape.hpp Source File#

Composable Kernel: /home/docs/checkouts/readthedocs.org/user_builds/advanced-micro-devices-composable-kernel/checkouts/develop/include/ck_tile/ops/reduce/pipeline/reduce2d_shape.hpp Source File

Go to the documentation of this file.

 // SPDX-License-Identifier: MIT
 // Copyright (c) 2018-2025, Advanced Micro Devices, Inc. All rights reserved.
  
 #pragma once
  
 #include "ck_tile/core.hpp"
  
 namespace ck_tile {
  
 template <typename BlockWarps, // num warps along seq<M, N>
           typename BlockTile,  // block size, seq<M, N>
           typename WarpTile,   // warp size, seq<M, N>
           typename ThreadTile> // contiguous pixels(vector size) along seq<M, N>
 struct Reduce2dShape
 {
     static constexpr index_t Block_M = BlockTile::at(number<0>{});
     static constexpr index_t Block_N = BlockTile::at(number<1>{});
  
     static constexpr index_t Warp_M = WarpTile::at(number<0>{});
     static constexpr index_t Warp_N = WarpTile::at(number<1>{});
  
     static constexpr index_t ThreadTile_M = ThreadTile::at(number<0>{});
     static constexpr index_t ThreadTile_N = ThreadTile::at(number<1>{});
  
     static constexpr index_t WarpPerBlock_M = BlockWarps::at(number<0>{});
     static constexpr index_t WarpPerBlock_N = BlockWarps::at(number<1>{});
  
     static constexpr index_t RepeatInWarp =
         Warp_M * Warp_N / ThreadTile_M / ThreadTile_N / ck_tile::get_warp_size();
     static constexpr index_t RepeatInWarp_M =
         (Warp_M / ThreadTile_M > Warp_N / ThreadTile_N) ? RepeatInWarp : 1;
     static constexpr index_t RepeatInWarp_N =
         (Warp_M / ThreadTile_M > Warp_N / ThreadTile_N) ? 1 : RepeatInWarp;
  
     static constexpr index_t ThreadPerWarp_M = Warp_M / ThreadTile_M / RepeatInWarp_M;
     static constexpr index_t ThreadPerWarp_N = Warp_N / ThreadTile_N / RepeatInWarp_N;
  
     static constexpr index_t Repeat_M = Block_M * RepeatInWarp_M / (WarpPerBlock_M * Warp_M);
     static constexpr index_t Repeat_N = Block_N * RepeatInWarp_N / (WarpPerBlock_N * Warp_N);
  
     static constexpr index_t BlockSize =
         ck_tile::get_warp_size() * reduce_on_sequence(BlockWarps{}, multiplies{}, number<1>{});
 };
 } // namespace ck_tile