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

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

Composable Kernel: /home/docs/checkouts/readthedocs.org/user_builds/advanced-micro-devices-composable-kernel/checkouts/docs-6.4.3/include/ck/tensor_operation/gpu/device/device_softmax.hpp Source File
device_softmax.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 <memory>
7 #include <vector>
8 
9 #include "ck/ck.hpp"
10 #include "ck/tensor_operation/gpu/device/device_base.hpp"
11 
12 namespace ck {
13 namespace tensor_operation {
14 namespace device {
15 
16 template <typename InDataType,
17  typename AccDataType,
18  typename OutDataType,
19  typename InElementwiseOp,
20  typename AccElementwiseOp,
21  index_t Rank,
22  index_t NumReduceDim>
23 struct DeviceSoftmax : public BaseOperator
24 {
25  //
26  // @brief Makes a pointer to Argument class.
27  //
28  // @param[in] inLengths Input tensor extent(s) from high to low dimension
29  // @param[in] inStrides Input tensor stride(s) from high to low dimension
30  // @param[in] reduceDims The dimension(s) the normalization operation is applied
31  // @param[in] alpha double type value
32  // @param[in] beta double type value
33  // @param[in] in_dev Typeless const pointer in device memory storing the input
34  // tensor
35  // @param out_dev Typeless pointer in device memory storing the output tensor
36  // @param[in] in_elementwise_op The input elementwise operation.
37  // @param[in] acc_elementwise_op The accumulation elementwise operation.
38  //
39  // @return Unique pointer to the Argument class.
40  //
41  virtual std::unique_ptr<BaseArgument>
42  MakeArgumentPointer(const std::vector<index_t> inLengths,
43  const std::vector<index_t> inStrides,
44  const std::vector<int> reduceDims,
45  double alpha,
46  double beta,
47  const void* in_dev,
48  void* out_dev,
49  InElementwiseOp in_elementwise_op,
50  AccElementwiseOp acc_elementwise_op) = 0;
51 
52  virtual std::unique_ptr<BaseInvoker> MakeInvokerPointer() = 0;
53 };
54 
55 template <typename InDataType,
56  typename AccDataType,
57  typename OutDataType,
58  typename InElementwiseOp,
59  typename AccElementwiseOp,
60  index_t Rank,
61  index_t NumReduceDim>
62 using DeviceSoftmaxPtr = std::unique_ptr<DeviceSoftmax<InDataType,
63  AccDataType,
64  OutDataType,
65  InElementwiseOp,
66  AccElementwiseOp,
67  Rank,
68  NumReduceDim>>;
69 
70 } // namespace device
71 } // namespace tensor_operation
72 } // namespace ck
device_base.hpp
ck::tensor_operation::device::DeviceSoftmaxPtr
std::unique_ptr< DeviceSoftmax< InDataType, AccDataType, OutDataType, InElementwiseOp, AccElementwiseOp, Rank, NumReduceDim > > DeviceSoftmaxPtr
Definition: device_softmax.hpp:68
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::DeviceSoftmax
Definition: device_softmax.hpp:24
ck::tensor_operation::device::DeviceSoftmax::MakeArgumentPointer
virtual std::unique_ptr< BaseArgument > MakeArgumentPointer(const std::vector< index_t > inLengths, const std::vector< index_t > inStrides, const std::vector< int > reduceDims, double alpha, double beta, const void *in_dev, void *out_dev, InElementwiseOp in_elementwise_op, AccElementwiseOp acc_elementwise_op)=0
ck::tensor_operation::device::DeviceSoftmax::MakeInvokerPointer
virtual std::unique_ptr< BaseInvoker > MakeInvokerPointer()=0