/home/docs/checkouts/readthedocs.org/user_builds/advanced-micro-devices-composable-kernel/checkouts/docs-6.4.3/include/ck/tensor_operation/gpu/device/device_gemm_multiple_d_multiple_r.hpp Source File

• docs-6.4.3
• include
• ck
• tensor_operation
• gpu
• device

device_gemm_multiple_d_multiple_r.hpp

Go to the documentation of this file.

1 // SPDX-License-Identifier: MIT

2 // Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.

3 

4 #pragma once

5 

6 #include <array>

7 

8 #include "ck/tensor_operation/gpu/device/device_base.hpp"

9 

10 namespace ck {

11 namespace tensor_operation {

12 namespace device {

13 

14 // FIXME: DeviceGemmReduce type need to well define the problem

15 // GEMM:

16 // input : A[AK0, M, AK1]

17 // input : B[AK0, N, AK1]

18 // input : D0[M, N], D1[M, N], ...

19 // output : E[M, N]

20 // output : R0[M], R1[M], ...

21 // C = a_op(A) * b_op(B)

22 // E = cde_op(C, D0, D1, ...)

23 // Q0 = reduce0(q_op0(E)), Q1 = reduce1(q_op0(E)), ...

24 // R0 = r_op0(Q0), R1 = r_op1(Q1), ...

25 // Assume:

26 // D0, D1, ... and E have the same layout

27 template <typename ALayout,

28  typename BLayout,

29  typename DELayout,

30  typename ADataType,

31  typename BDataType,

32  typename DsDataType,

33  typename EDataType,

34  typename RsDataType,

35  typename AElementwiseOperation,

36  typename BElementwiseOperation,

37  typename CDEElementwiseOperation,

38  typename QsElementwiseOperation,

39  typename RsElementwiseOperation>

40 struct DeviceGemmMultipleDMultipleR : public BaseOperator

41 {

42  static constexpr index_t NumDTensor = DsDataType::Size();

43  static constexpr index_t NumRTensor = RsDataType::Size();

44 

45  virtual std::unique_ptr<BaseArgument>

46  MakeArgumentPointer(const void* p_a,

47  const void* p_b,

48  std::array<const void*, NumDTensor> p_ds,

49  void* p_e,

50  std::array<void*, NumRTensor> p_rs,

51  ck::index_t M,

52  ck::index_t N,

53  ck::index_t K,

54  ck::index_t StrideA,

55  ck::index_t StrideB,

56  std::array<ck::index_t, NumDTensor> StrideDs,

57  ck::index_t StrideE,

58  AElementwiseOperation a_element_op,

59  BElementwiseOperation b_element_op,

60  CDEElementwiseOperation cde_element_op,

61  QsElementwiseOperation qs_element_op,

62  RsElementwiseOperation rs_element_op) = 0;

63 

64  virtual std::unique_ptr<BaseInvoker> MakeInvokerPointer() = 0;

65 };

66 

67 template <typename ALayout,

68  typename BLayout,

69  typename DELayout,

70  typename ADataType,

71  typename BDataType,

72  typename DsDataType,

73  typename EDataType,

74  typename RsDataType,

75  typename AElementwiseOperation,

76  typename BElementwiseOperation,

77  typename CDEElementwiseOperation,

78  typename QsElementwiseOperation,

79  typename RsElementwiseOperation>

80 using DeviceGemmMultipleDMultipleRPtr =

81  std::unique_ptr<DeviceGemmMultipleDMultipleR<ALayout,

82  BLayout,

83  DELayout,

84  ADataType,

85  BDataType,

86  DsDataType,

87  EDataType,

88  RsDataType,

89  AElementwiseOperation,

90  BElementwiseOperation,

91  CDEElementwiseOperation,

92  QsElementwiseOperation,

93  RsElementwiseOperation>>;

94 

95 } // namespace device

96 } // namespace tensor_operation

97 } // namespace ck

device_base.hpp

ck::tensor_operation::device::DeviceGemmMultipleDMultipleRPtr

std::unique_ptr< DeviceGemmMultipleDMultipleR< ALayout, BLayout, DELayout, ADataType, BDataType, DsDataType, EDataType, RsDataType, AElementwiseOperation, BElementwiseOperation, CDEElementwiseOperation, QsElementwiseOperation, RsElementwiseOperation > > DeviceGemmMultipleDMultipleRPtr

Definition: device_gemm_multiple_d_multiple_r.hpp:93

ck

Definition: ck.hpp:264

ck::index_t

int32_t index_t

Definition: ck.hpp:289

ck::tensor_operation::device::BaseOperator

Definition: device_base.hpp:76

ck::tensor_operation::device::DeviceGemmMultipleDMultipleR

Definition: device_gemm_multiple_d_multiple_r.hpp:41

ck::tensor_operation::device::DeviceGemmMultipleDMultipleR::NumDTensor

static constexpr index_t NumDTensor

Definition: device_gemm_multiple_d_multiple_r.hpp:42

ck::tensor_operation::device::DeviceGemmMultipleDMultipleR::MakeInvokerPointer

virtual std::unique_ptr< BaseInvoker > MakeInvokerPointer()=0

ck::tensor_operation::device::DeviceGemmMultipleDMultipleR::NumRTensor

static constexpr index_t NumRTensor

Definition: device_gemm_multiple_d_multiple_r.hpp:43

ck::tensor_operation::device::DeviceGemmMultipleDMultipleR::MakeArgumentPointer

virtual std::unique_ptr< BaseArgument > MakeArgumentPointer(const void *p_a, const void *p_b, std::array< const void *, NumDTensor > p_ds, void *p_e, std::array< void *, NumRTensor > p_rs, ck::index_t M, ck::index_t N, ck::index_t K, ck::index_t StrideA, ck::index_t StrideB, std::array< ck::index_t, NumDTensor > StrideDs, ck::index_t StrideE, AElementwiseOperation a_element_op, BElementwiseOperation b_element_op, CDEElementwiseOperation cde_element_op, QsElementwiseOperation qs_element_op, RsElementwiseOperation rs_element_op)=0