docs/default__epilogue__simt_8h_source.html
| | CUTLASS
CUDA Templates for Linear Algebra Subroutines and Solvers |
default_epilogue_simt.h
[Go to the documentation of this file.](default epilogue simt_8h.html)
1 /***************************************************************************************************
2 * Copyright (c) 2017-2019, NVIDIA CORPORATION. All rights reserved.
3 *
4 * Redistribution and use in source and binary forms, with or without modification, are permitted
5 * provided that the following conditions are met:
6 * * Redistributions of source code must retain the above copyright notice, this list of
7 * conditions and the following disclaimer.
8 * * Redistributions in binary form must reproduce the above copyright notice, this list of
9 * conditions and the following disclaimer in the documentation and/or other materials
10 * provided with the distribution.
11 * * Neither the name of the NVIDIA CORPORATION nor the names of its contributors may be used
12 * to endorse or promote products derived from this software without specific prior written
13 * permission.
14 *
15 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR
16 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND
17 * FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL NVIDIA CORPORATION BE LIABLE
18 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
19 * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
20 * OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
21 * STRICT LIABILITY, OR TOR (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
22 * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
23 *
24 **************************************************************************************************/
33 #pragma once
34
35 #include "cutlass/cutlass.h"
36 #include "cutlass/numeric_types.h"
37 #include "cutlass/array.h"
38
39 #include "cutlass/gemm/gemm.h"
40
41 #include "cutlass/epilogue/thread/linear_combination.h"
42 #include "cutlass/epilogue/thread/conversion_op.h"
43 #include "cutlass/epilogue/thread/reduction_op.h"
44
45 #include "[cutlass/transform/threadblock/regular_tile_iterator_pitch_linear.h](regular tile iterator pitch linear_8h.html)"
46
47 #include "[cutlass/epilogue/warp/fragment_iterator_simt.h](fragment iterator simt_8h.html)"
48 #include "[cutlass/epilogue/warp/tile_iterator_simt.h](tile iterator simt_8h.html)"
49 #include "[cutlass/epilogue/threadblock/default_thread_map_simt.h](default thread map__simt_8h.html)"
50
51 #include "[cutlass/epilogue/threadblock/predicated_tile_iterator.h](epilogue_2threadblock_2predicated tile iterator_8h.html)"
52 #include "[cutlass/epilogue/threadblock/shared_load_iterator.h](shared load iterator_8h.html)"
53 #include "cutlass/epilogue/threadblock/epilogue.h"
54
56
57 namespace cutlass {
58 namespace epilogue {
59 namespace threadblock {
60
62
64 template <
65typename Shape_,
66typename WarpMmaSimt_,
67typename OutputOp_,
68int ElementsPerAccess
69 >
70 struct DefaultEpilogueSimt {
71
73using WarpMmaSimt = WarpMmaSimt_;
75static int const kElementsPerAccess = ElementsPerAccess;
76static const int kPartitionsK = Shape::kK / WarpMmaSimt::Shape::kK;
77
78using ElementOutput = typename OutputOp::ElementOutput;
79using LayoutC = typename WarpMmaSimt::LayoutC;
80using ElementAccumulator = typename WarpMmaSimt::ElementC;
81
82//
83// Thread map
84//
85
86using OutputTileThreadMap = typename cutlass::epilogue::threadblock::DefaultThreadMapSimt<
87Shape,
88typename WarpMmaSimt::Shape,
89typename WarpMmaSimt::Policy,
90kPartitionsK,
92 kElementsPerAccess
93 >::Type;
94
95using OutputTileIterator = cutlass::epilogue::threadblock::PredicatedTileIterator<
97 ElementOutput
98 >;
99
100using AccumulatorFragmentIterator = cutlass::epilogue::warp::FragmentIteratorSimt<
101typename WarpMmaSimt::Shape,
102typename WarpMmaSimt::ThreadMma,
103layout::RowMajor,
104typename WarpMmaSimt::Policy
105 >;
106
107using WarpTileIterator = cutlass::epilogue::warp::TileIteratorSimt<
108typename WarpMmaSimt::Shape,
109typename WarpMmaSimt::ThreadMma,
111 layout::RowMajor,
112typename WarpMmaSimt::Policy
113 >;
114
115using SharedLoadIterator = cutlass::epilogue::threadblock::SharedLoadIterator<
116typename OutputTileThreadMap::CompactedThreadMap,
117 ElementAccumulator
118 >;
119
121using Padding = typename WarpTileIterator::Padding;
122
123//
124// Define the epilogue
125//
126using Epilogue = cutlass::epilogue::threadblock::Epilogue<
127Shape,
128WarpMmaSimt,
129kPartitionsK,
131AccumulatorFragmentIterator,
132WarpTileIterator,
134OutputOp,
135Padding
136 >;
137 };
138
140
141 } // namespace threadblock
142 } // namespace epilogue
143 } // namespace cutlass
144
[regular_tile_iterator_pitch_linear.h](regular tile iterator pitch linear_8h.html)
Templates implementing loading of tiles from pitch-linear rank=2 tensors.
Definition: aligned_buffer.h:35
[tile_iterator_simt.h](tile iterator simt_8h.html)
cutlass::epilogue::threadblock::DefaultEpilogueSimt
Defines sensible defaults for epilogues for SimtOps.
Definition: default_epilogue_simt.h:70
cutlass::epilogue::threadblock::DefaultEpilogueSimt::kElementsPerAccess
static int const kElementsPerAccess
Definition: default_epilogue_simt.h:75
[predicated_tile_iterator.h](epilogue_2threadblock_2predicated tile iterator_8h.html)
Epilogue for threadblock scoped GEMMs using Tensor Ops.
cutlass::epilogue::threadblock::DefaultEpilogueSimt::kPartitionsK
static const int kPartitionsK
Definition: default_epilogue_simt.h:76
Defines common types used for all GEMM-like operators.
Functor performing conversion operations used by epilogues.
cutlass::epilogue::threadblock::DefaultEpilogueSimt::SharedLoadIterator
cutlass::epilogue::threadblock::SharedLoadIterator< typename OutputTileThreadMap::CompactedThreadMap, ElementAccumulator > SharedLoadIterator
Definition: default_epilogue_simt.h:118
cutlass::epilogue::threadblock::DefaultEpilogueSimt::WarpMmaSimt
WarpMmaSimt_ WarpMmaSimt
Definition: default_epilogue_simt.h:73
cutlass::epilogue::threadblock::DefaultThreadMapSimt
Defines the optimal thread map for SIMT accumulator layouts.
Definition: default_thread_map_simt.h:52
Statically sized array of elements that accommodates all CUTLASS-supported numeric types and is safe ...
cutlass::epilogue::threadblock::DefaultEpilogueSimt::OutputOp
OutputOp_ OutputOp
Definition: default_epilogue_simt.h:74
Functor performing linear combination operations used by epilogues.
cutlass::epilogue::threadblock::DefaultEpilogueSimt::AccumulatorFragmentIterator
cutlass::epilogue::warp::FragmentIteratorSimt< typename WarpMmaSimt::Shape, typename WarpMmaSimt::ThreadMma, layout::RowMajor, typename WarpMmaSimt::Policy > AccumulatorFragmentIterator
Definition: default_epilogue_simt.h:105
cutlass::epilogue::threadblock::DefaultEpilogueSimt::LayoutC
typename WarpMmaSimt::LayoutC LayoutC
Definition: default_epilogue_simt.h:79
cutlass::epilogue::warp::FragmentIteratorSimt
Fragment iterator for SIMT accumulator arrangements.
Definition: fragment_iterator_simt.h:60
cutlass::epilogue::threadblock::DefaultEpilogueSimt::ElementAccumulator
typename WarpMmaSimt::ElementC ElementAccumulator
Definition: default_epilogue_simt.h:80
[default_thread_map_simt.h](default thread map__simt_8h.html)
Top-level include for all CUTLASS numeric types.
[fragment_iterator_simt.h](fragment iterator simt_8h.html)
This defines a "fragment" iterator for visiting the fragments of an accumulator tile that participate...
[shared_load_iterator.h](shared load iterator_8h.html)
Epilogue for threadblock scoped GEMMs using Tensor Ops.
cutlass::epilogue::threadblock::DefaultEpilogueSimt::Padding
typename WarpTileIterator::Padding Padding
Hard-coded padding elements added.
Definition: default_epilogue_simt.h:121
cutlass::epilogue::threadblock::DefaultEpilogueSimt::Shape
Shape_ Shape
Definition: default_epilogue_simt.h:72
Mapping function for row-major matrices.
Definition: layout/matrix.h:50
cutlass::epilogue::threadblock::Epilogue
Epilogue operator without splitk.
Definition: epilogue.h:74
Epilogue for threadblock scoped GEMMs using Tensor Ops.
cutlass::epilogue::threadblock::PredicatedTileIterator
Definition: epilogue/threadblock/predicated_tile_iterator.h:65
cutlass::epilogue::threadblock::DefaultEpilogueSimt::ElementOutput
typename OutputOp::ElementOutput ElementOutput
Definition: default_epilogue_simt.h:78
cutlass::epilogue::threadblock::DefaultEpilogueSimt::WarpTileIterator
cutlass::epilogue::warp::TileIteratorSimt< typename WarpMmaSimt::Shape, typename WarpMmaSimt::ThreadMma, ElementAccumulator, layout::RowMajor, typename WarpMmaSimt::Policy > WarpTileIterator
Definition: default_epilogue_simt.h:113
cutlass::epilogue::threadblock::SharedLoadIterator
Definition: shared_load_iterator.h:61
cutlass::epilogue::threadblock::DefaultEpilogueSimt::OutputTileIterator
cutlass::epilogue::threadblock::PredicatedTileIterator< OutputTileThreadMap, ElementOutput > OutputTileIterator
Definition: default_epilogue_simt.h:98
Functor performing reduction operations used by epilogues.
Basic include for CUTLASS.
cutlass::epilogue::warp::TileIteratorSimt
Template for reading and writing tiles of accumulators to shared memory.
Definition: tile_iterator_simt.h:55
cutlass::epilogue::threadblock::DefaultEpilogueSimt::OutputTileThreadMap
typename cutlass::epilogue::threadblock::DefaultThreadMapSimt< Shape, typename WarpMmaSimt::Shape, typename WarpMmaSimt::Policy, kPartitionsK, ElementOutput, kElementsPerAccess >::Type OutputTileThreadMap
Definition: default_epilogue_simt.h:93
Generated by 1.8.11