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

• develop
• include
• ck
• tensor_operation
• gpu
• device

device_grouped_conv_bwd_weight.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 #define DISABLE_SPLIT_K_AUTODEDUCE_FOR_ONE_STAGE_KERNELS 1

15 

16 template <ck::index_t NDimSpatial,

17  typename InLayout,

18  typename WeiLayout,

19  typename OutLayout,

20  typename InDataType,

21  typename WeiDataType,

22  typename OutDataType,

23  typename InElementwiseOperation,

24  typename WeiElementwiseOperation,

25  typename OutElementwiseOperation,

26  typename ComputeTypeA = InDataType,

27  typename ComputeTypeB = ComputeTypeA>

28 struct DeviceGroupedConvBwdWeight : public BaseOperator

29 {

30  virtual std::unique_ptr<BaseArgument>

31  MakeArgumentPointer(const void* p_in,

32  void* p_wei,

33  const void* p_out,

34  const std::array<index_t, NDimSpatial + 3>& a_g_n_c_wis_lengths, // input

35  const std::array<index_t, NDimSpatial + 3>& a_g_n_c_wis_strides,

36  const std::array<index_t, NDimSpatial + 3>& b_g_k_c_xs_lengths, // weight

37  const std::array<index_t, NDimSpatial + 3>& b_g_k_c_xs_strides,

38  const std::array<index_t, NDimSpatial + 3>& e_g_n_k_wos_lengths, // output

39  const std::array<index_t, NDimSpatial + 3>& e_g_n_k_wos_strides,

40  const std::array<ck::index_t, NDimSpatial>& conv_filter_strides,

41  const std::array<ck::index_t, NDimSpatial>& conv_filter_dilations,

42  const std::array<ck::index_t, NDimSpatial>& input_left_pads,

43  const std::array<ck::index_t, NDimSpatial>& input_right_pads,

44  InElementwiseOperation in_element_op,

45  WeiElementwiseOperation wei_element_op,

46  OutElementwiseOperation out_element_op,

47  ck::index_t split_k) = 0;

48 

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

50 };

51 

52 } // namespace device

53 } // namespace tensor_operation

54 } // namespace ck

device_base.hpp

ck

Definition: ck.hpp:268

ck::index_t

int32_t index_t

Definition: ck.hpp:299

ck::tensor_operation::device::BaseOperator

Definition: device_base.hpp:223

ck::tensor_operation::device::DeviceGroupedConvBwdWeight

Definition: device_grouped_conv_bwd_weight.hpp:29

ck::tensor_operation::device::DeviceGroupedConvBwdWeight::MakeInvokerPointer

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

ck::tensor_operation::device::DeviceGroupedConvBwdWeight::MakeArgumentPointer

virtual std::unique_ptr< BaseArgument > MakeArgumentPointer(const void *p_in, void *p_wei, const void *p_out, const std::array< index_t, NDimSpatial+3 > &a_g_n_c_wis_lengths, const std::array< index_t, NDimSpatial+3 > &a_g_n_c_wis_strides, const std::array< index_t, NDimSpatial+3 > &b_g_k_c_xs_lengths, const std::array< index_t, NDimSpatial+3 > &b_g_k_c_xs_strides, const std::array< index_t, NDimSpatial+3 > &e_g_n_k_wos_lengths, const std::array< index_t, NDimSpatial+3 > &e_g_n_k_wos_strides, const std::array< ck::index_t, NDimSpatial > &conv_filter_strides, const std::array< ck::index_t, NDimSpatial > &conv_filter_dilations, const std::array< ck::index_t, NDimSpatial > &input_left_pads, const std::array< ck::index_t, NDimSpatial > &input_right_pads, InElementwiseOperation in_element_op, WeiElementwiseOperation wei_element_op, OutElementwiseOperation out_element_op, ck::index_t split_k)=0