Back to Cutlass

CUTLASS: wmma_ptx.h Source File

docs/wmma__ptx_8h_source.html

4.4.25.0 KB
Original Source

| | CUTLASS

CUDA Templates for Linear Algebra Subroutines and Solvers |

wmma_ptx.h

Go to the documentation of this file.

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 **************************************************************************************************/

28 #pragma once

29

30 #include "cutlass/arch/wmma.h"

31

32 namespace cutlass {

33 namespace arch {

34

40

44 template <

45typename Shape_,

46typename Element_,

47typename Layout_,

48 MemoryKind Memory = MemoryKind::kShared

49 >

50 struct PtxWmmaLoadA;

52

53 template <

54typename Shape_,

55typename Element_,

56typename Layout_,

57 MemoryKind Memory = MemoryKind::kShared

58 >

59 struct PtxWmmaLoadB;

61

62 template <

63typename Shape_,

64typename Element_,

65typename Layout_,

66 MemoryKind Memory = MemoryKind::kShared

67 >

68 struct PtxWmmaLoadC;

70

71

75 template <

76typename Shape_,

77typename ElementA_,

78typename LayoutA_,

79typename ElementB_,

80typename LayoutB_,

81typename ElementC_,

82typename LayoutC_,

83typename Operator = cutlass::arch::OpMultiplyAdd

84 >

85 struct PtxWmma;

87

91 template <

92typename Shape_,

93typename Element_,

94typename Layout_,

95 MemoryKind Memory = MemoryKind::kShared

96 >

97 struct PtxWmmaStoreD;

99

101

102 } // namespace arch

103 } // namespace cutlass

104

cutlass

Definition: aligned_buffer.h:35

cutlass::arch::PtxWmmaLoadB

Definition: wmma_ptx.h:59

cutlass::arch::PtxWmmaLoadC

Definition: wmma_ptx.h:68

cutlass::arch::PtxWmma

WMMA Matrix multiply-add operation.

Definition: wmma_ptx.h:85

cutlass::arch::PtxWmmaStoreD

WMMA store for matrix D.

Definition: wmma_ptx.h:97

cutlass::arch::PtxWmmaLoadA

WMMA PTX string load for A, B, and C matrices.

Definition: wmma_ptx.h:50

wmma.h

Templates exposing architecture support for warp matrix multiply-add (WMMA) operations.


Generated by 1.8.11