update profiler for conv bwd weight

library/include/ck/library/tensor_operation_instance/gpu/convolution_backward_weight.hpp

+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#pragma once
+
+#include <cstdlib>
+
+#include "ck/ck.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/device/device_conv_bwd_weight.hpp"
+#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+
+#include "ck/library/tensor_operation_instance/device_operation_instance_factory.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace instance {
+
+// conv1d backward weight
+void add_device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_bf16_f32_bf16_instances(
+    std::vector<std::unique_ptr<DeviceConvBwdWeight<1,
+                                                    NWC,
+                                                    KXC,
+                                                    NWK,
+                                                    BF16,
+                                                    F32,
+                                                    BF16,
+                                                    PassThrough,
+                                                    PassThrough,
+                                                    PassThrough>>>& instances);
+
+void add_device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_f16_instances(
+    std::vector<std::unique_ptr<DeviceConvBwdWeight<1,
+                                                    NWC,
+                                                    KXC,
+                                                    NWK,
+                                                    F16,
+                                                    F16,
+                                                    F16,
+                                                    PassThrough,
+                                                    PassThrough,
+                                                    PassThrough>>>& instances);
+
+void add_device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_f32_instances(
+    std::vector<std::unique_ptr<DeviceConvBwdWeight<1,
+                                                    NWC,
+                                                    KXC,
+                                                    NWK,
+                                                    F32,
+                                                    F32,
+                                                    F32,
+                                                    PassThrough,
+                                                    PassThrough,
+                                                    PassThrough>>>& instances);
+
+// conv2d backward weight
+void add_device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_bf16_f32_bf16_instances(
+    std::vector<std::unique_ptr<DeviceConvBwdWeight<2,
+                                                    NHWC,
+                                                    KYXC,
+                                                    NHWK,
+                                                    BF16,
+                                                    F32,
+                                                    BF16,
+                                                    PassThrough,
+                                                    PassThrough,
+                                                    PassThrough>>>& instances);
+
+void add_device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_f16_instances(
+    std::vector<std::unique_ptr<DeviceConvBwdWeight<2,
+                                                    NHWC,
+                                                    KYXC,
+                                                    NHWK,
+                                                    F16,
+                                                    F16,
+                                                    F16,
+                                                    PassThrough,
+                                                    PassThrough,
+                                                    PassThrough>>>& instances);
+
+void add_device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_f32_instances(
+    std::vector<std::unique_ptr<DeviceConvBwdWeight<2,
+                                                    NHWC,
+                                                    KYXC,
+                                                    NHWK,
+                                                    F32,
+                                                    F32,
+                                                    F32,
+                                                    PassThrough,
+                                                    PassThrough,
+                                                    PassThrough>>>& instances);
+
+// conv3d backward weight
+void add_device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_bf16_f32_bf16_instances(
+    std::vector<std::unique_ptr<DeviceConvBwdWeight<3,
+                                                    NDHWC,
+                                                    KZYXC,
+                                                    NDHWK,
+                                                    BF16,
+                                                    F32,
+                                                    BF16,
+                                                    PassThrough,
+                                                    PassThrough,
+                                                    PassThrough>>>& instances);
+
+void add_device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_f16_instances(
+    std::vector<std::unique_ptr<DeviceConvBwdWeight<3,
+                                                    NDHWC,
+                                                    KZYXC,
+                                                    NDHWK,
+                                                    F16,
+                                                    F16,
+                                                    F16,
+                                                    PassThrough,
+                                                    PassThrough,
+                                                    PassThrough>>>& instances);
+
+void add_device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_f32_instances(
+    std::vector<std::unique_ptr<DeviceConvBwdWeight<3,
+                                                    NDHWC,
+                                                    KZYXC,
+                                                    NDHWK,
+                                                    F32,
+                                                    F32,
+                                                    F32,
+                                                    PassThrough,
+                                                    PassThrough,
+                                                    PassThrough>>>& instances);
+
+template <ck::index_t NumDimSpatial,
+          typename InLayout,
+          typename WeiLayout,
+          typename OutLayout,
+          typename InDataType,
+          typename WeiDataType,
+          typename OutDataType>
+struct DeviceOperationInstanceFactory<ck::tensor_operation::device::DeviceConvBwdWeight<
+    NumDimSpatial,
+    InLayout,
+    WeiLayout,
+    OutLayout,
+    InDataType,
+    WeiDataType,
+    OutDataType,
+    ck::tensor_operation::element_wise::PassThrough,
+    ck::tensor_operation::element_wise::PassThrough,
+    ck::tensor_operation::element_wise::PassThrough>>
+{
+    using DeviceOp = DeviceConvBwdWeight<NumDimSpatial,
+                                         InLayout,
+                                         WeiLayout,
+                                         OutLayout,
+                                         InDataType,
+                                         WeiDataType,
+                                         OutDataType,
+                                         ck::tensor_operation::element_wise::PassThrough,
+                                         ck::tensor_operation::element_wise::PassThrough,
+                                         ck::tensor_operation::element_wise::PassThrough>;
+
+    static auto GetInstances()
+    {
+        std::vector<std::unique_ptr<DeviceOp>> op_ptrs;
+
+        if constexpr(NumDimSpatial == 1 && is_same_v<InLayout, NWC> && is_same_v<WeiLayout, KXC> &&
+                     is_same_v<OutLayout, NWK>)
+        {
+            if constexpr(is_same_v<InDataType, float> && is_same_v<WeiDataType, float> &&
+                         is_same_v<OutDataType, float>)
+            {
+                add_device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_f32_instances(op_ptrs);
+            }
+            else if constexpr(is_same_v<InDataType, half_t> && is_same_v<WeiDataType, half_t> &&
+                              is_same_v<OutDataType, half_t>)
+            {
+                add_device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_f16_instances(op_ptrs);
+            }
+            else if constexpr(is_same_v<InDataType, ck::bhalf_t> && is_same_v<WeiDataType, float> &&
+                              is_same_v<OutDataType, ck::bhalf_t>)
+            {
+                add_device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_bf16_f32_bf16_instances(op_ptrs);
+            }
+        }
+        else if constexpr(NumDimSpatial == 2 && is_same_v<InLayout, NHWC> &&
+                          is_same_v<WeiLayout, KYXC> && is_same_v<OutLayout, NHWK>)
+        {
+            if constexpr(is_same_v<InDataType, float> && is_same_v<WeiDataType, float> &&
+                         is_same_v<OutDataType, float>)
+            {
+                add_device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_f32_instances(op_ptrs);
+            }
+            else if constexpr(is_same_v<InDataType, half_t> && is_same_v<WeiDataType, half_t> &&
+                              is_same_v<OutDataType, half_t>)
+            {
+                add_device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_f16_instances(op_ptrs);
+            }
+            else if constexpr(is_same_v<InDataType, ck::bhalf_t> && is_same_v<WeiDataType, float> &&
+                              is_same_v<OutDataType, ck::bhalf_t>)
+            {
+                add_device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_bf16_f32_bf16_instances(op_ptrs);
+            }
+        }
+        else if constexpr(NumDimSpatial == 3 && is_same_v<InLayout, NDHWC> &&
+                          is_same_v<WeiLayout, KZYXC> && is_same_v<OutLayout, NDHWK>)
+        {
+            if constexpr(is_same_v<InDataType, float> && is_same_v<WeiDataType, float> &&
+                         is_same_v<OutDataType, float>)
+            {
+                add_device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_f32_instances(op_ptrs);
+            }
+            else if constexpr(is_same_v<InDataType, half_t> && is_same_v<WeiDataType, half_t> &&
+                              is_same_v<OutDataType, half_t>)
+            {
+                add_device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_f16_instances(op_ptrs);
+            }
+            else if constexpr(is_same_v<InDataType, ck::bhalf_t> && is_same_v<WeiDataType, float> &&
+                              is_same_v<OutDataType, ck::bhalf_t>)
+            {
+                add_device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_bf16_f32_bf16_instances(op_ptrs);
+            }
+        }
+
+        return op_ptrs;
+    }
+};
+
+} // namespace instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck

library/include/ck/library/tensor_operation_instance/gpu/convolution_forward.hpp

@@ -17,7 +17,7 @@ namespace tensor_operation {
 namespace device {
 namespace instance {
 
-// conv1d_fwd
+// conv1d forward
 void add_device_conv1d_fwd_xdl_nwc_kxc_nwk_bf16_instances(
     std::vector<std::unique_ptr<
         DeviceConvFwd<1, NWC, KXC, NWK, BF16, BF16, BF16, PassThrough, PassThrough, PassThrough>>>&
@@ -45,7 +45,7 @@ void add_device_conv1d_fwd_xdl_nwc_kxc_nwk_int8_instances(
                                               PassThrough,
                                               PassThrough>>>& instances);
 
-// conv2d_fwd
+// conv2d forward
 void add_device_conv2d_fwd_xdl_c_shuffle_nhwc_kyxc_nhwk_f16_instances(
     std::vector<std::unique_ptr<
         DeviceConvFwd<2, NHWC, KYXC, NHWK, F16, F16, F16, PassThrough, PassThrough, PassThrough>>>&
@@ -119,7 +119,7 @@ void add_device_convnd_2d_fwd_xdl_nhwc_kyxc_nhwk_int8_instances(
                                               PassThrough,
                                               PassThrough>>>& instances);
 
-// conv3d_fwd
+// conv3d forward
 void add_device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk_bf16_instances(
     std::vector<std::unique_ptr<DeviceConvFwd<3,
                                               NDHWC,

library/src/tensor_operation_instance/gpu/CMakeLists.txt

@@ -23,8 +23,9 @@ add_subdirectory(conv2d_fwd_bias_relu)
 add_subdirectory(conv2d_fwd_bias_relu_add)
 add_subdirectory(conv2d_bwd_data)
 add_subdirectory(convnd_bwd_data)
+add_subdirectory(conv1d_bwd_weight)
 add_subdirectory(conv2d_bwd_weight)
-add_subdirectory(convnd_bwd_weight)
+add_subdirectory(conv3d_bwd_weight)
 add_subdirectory(reduce)
 add_subdirectory(normalization)
 add_subdirectory(elementwise)
@@ -47,8 +48,9 @@ add_library(device_operations STATIC
     $<TARGET_OBJECTS:device_conv2d_fwd_bias_relu_add_instance>
     $<TARGET_OBJECTS:device_conv2d_bwd_data_instance>
     $<TARGET_OBJECTS:device_convnd_bwd_data_instance>
+    $<TARGET_OBJECTS:device_conv1d_bwd_weight_instance>
     $<TARGET_OBJECTS:device_conv2d_bwd_weight_instance>
-    $<TARGET_OBJECTS:device_convnd_bwd_weight_instance>
+    $<TARGET_OBJECTS:device_conv3d_bwd_weight_instance>
     $<TARGET_OBJECTS:device_reduce_instance>
     $<TARGET_OBJECTS:device_normalization_instance>
     $<TARGET_OBJECTS:device_elementwise_instance>

library/src/tensor_operation_instance/gpu/conv1d_bwd_weight/CMakeLists.txt

+#device_conv1d_bwd_weight_instance
+set(DEVICE_CONV1D_BWD_WEIGHT_INSTANCE_SOURCE 
+    device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_f16_instance.cpp;
+    device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_f32_instance.cpp;
+    device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_bf16_instance.cpp;
+)
+
+add_library(device_conv1d_bwd_weight_instance OBJECT ${DEVICE_CONV1D_BWD_WEIGHT_INSTANCE_SOURCE})
+target_compile_features(device_conv1d_bwd_weight_instance PUBLIC) 
+set_target_properties(device_conv1d_bwd_weight_instance PROPERTIES POSITION_INDEPENDENT_CODE ON)
+rocm_install(TARGETS device_conv1d_bwd_weight_instance)
+
+clang_tidy_check(device_conv1d_bwd_weight_instance)

library/src/tensor_operation_instance/gpu/conv1d_bwd_weight/device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_bf16_instance.cpp

+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include <cstdlib>
+
+#include "ck/ck.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/device/device_convnd_bwd_weight_nwc_kxc_nwk_xdl_cshuffle.hpp"
+#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+
+#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace instance {
+
+using BF16 = bhalf_t;
+using F32  = float;
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using NWC = ck::tensor_layout::convolution::NWC;
+using KXC = ck::tensor_layout::convolution::KXC;
+using NWK = ck::tensor_layout::convolution::NWK;
+
+using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+
+static constexpr auto ConvBwdWeightDefault =
+    ck::tensor_operation::device::ConvolutionBackwardWeightSpecialization::Default;
+
+static constexpr auto ConvBwdWeightFilter1x1Stride1Pad0 =
+    ck::tensor_operation::device::ConvolutionBackwardWeightSpecialization::Filter1x1Stride1Pad0;
+
+// Compilation parameters for in[n, wi, c] * wei[k, x, c] = out[n, wo, k]
+using device_conv1d_bwd_weight_xdl_c_shuffle_nwc_kxc_nwk_bf16_f32_bf16_instances = std::tuple<
+    // clang-format off
+        //#########################################|     Num| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                       ConvBackward| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|   BBlockTransfer|  BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|   CBlockTransfer|  CBlockTransfer|
+        //#########################################|     Dim|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                             Weight|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|    ThreadCluster|   ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|   ClusterLengths| ScalarPerVector|
+        //#########################################| Spatial|       |        |        |        |   Operation|   Operation|   Operation|                     Specialization|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          |  Lengths_K0_N_K1|    ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle| MBlock_MPerBlock|    NWaveNPerXdl|
+        //#########################################|        |       |        |        |        |            |            |            |                                   |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                 |                |               |              |               |               |          |            |            | NBlock_NPerBlock|                |
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,   256,   256,   128,     4,  8,   32,   32,    4,    2,  S<1, 4, 32, 2>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 16, 4>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 8>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,   256,   128,   256,     4,  8,   32,   32,    2,    4,  S<1, 4, 16, 4>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              2,      true,   S<1, 4, 32, 2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 8>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,   128,   128,   128,     4,  8,   32,   32,    4,    2,  S<1, 4, 16, 2>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 16, 2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,   256,   128,   128,     4,  8,   32,   32,    2,    2,  S<1, 4, 16, 4>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              2,      true,   S<1, 4, 16, 4>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,   128,   128,    64,     4,  8,   32,   32,    2,    2,  S<1, 4, 16, 2>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 8,  4>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,   128,    64,   128,     4,  8,   32,   32,    2,    2,  S<1, 4, 8,  4>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              2,      true,   S<1, 4, 16, 2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,    64,    64,    64,     4,  8,   32,   32,    2,    2,  S<1, 4, 8,  2>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 8,  2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 16, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,   256,   128,    64,     4,  8,   32,   32,    2,    1,  S<1, 4, 16, 4>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              2,      true,   S<1, 4, 8,  8>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              1,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,   256,    64,   128,     4,  8,   32,   32,    1,    2,  S<1, 4, 8,  8>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              1,      true,   S<1, 4, 16, 4>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,   128,   128,    32,     4,  8,   32,   32,    2,    1,  S<1, 4, 16, 2>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 4,  8>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              1,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,   128,    32,   128,     4,  8,   32,   32,    1,    2,  S<1, 4, 4,  8>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              1,      true,   S<1, 4, 16, 2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,    64,    64,    32,     4,  8,   32,   32,    2,    1,  S<1, 4, 8,  2>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 4,  4>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 16, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,    64,    32,    64,     4,  8,   32,   32,    1,    2,  S<1, 4, 4,  4>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              2,      true,   S<1, 4, 8,  2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 16, 1, 4>,               4>
+    // clang-format on
+    >;
+
+using device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_1x1_s1_p0_bf16_f32_bf16_instances = std::tuple<
+    // clang-format off
+        //#########################################|     Num| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                       ConvBackward| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|   BBlockTransfer|  BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|   CBlockTransfer|  CBlockTransfer|
+        //#########################################|     Dim|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                             Weight|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|    ThreadCluster|   ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|   ClusterLengths| ScalarPerVector|
+        //#########################################| Spatial|       |        |        |        |   Operation|   Operation|   Operation|                     Specialization|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          |  Lengths_K0_N_K1|    ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle| MBlock_MPerBlock|    NWaveNPerXdl|
+        //#########################################|        |       |        |        |        |            |            |            |                                   |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                 |                |               |              |               |               |          |            |            | NBlock_NPerBlock|                |
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,   256,   256,   128,     4,  8,   32,   32,    4,    2,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 16, 4>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 8>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,   256,   128,   256,     4,  8,   32,   32,    2,    4,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,   S<1, 4, 32, 2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 8>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,   128,   128,   128,     4,  8,   32,   32,    4,    2,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 16, 2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,   256,   128,   128,     4,  8,   32,   32,    2,    2,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,   S<1, 4, 16, 4>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,   128,   128,    64,     4,  8,   32,   32,    2,    2,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 8,  4>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,   128,    64,   128,     4,  8,   32,   32,    2,    2,  S<1, 4, 8,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,   S<1, 4, 16, 2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,    64,    64,    64,     4,  8,   32,   32,    2,    2,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 8,  2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 16, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,   256,   128,    64,     4,  8,   32,   32,    2,    1,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,   S<1, 4, 8,  8>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              1,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,   256,    64,   128,     4,  8,   32,   32,    1,    2,  S<1, 4, 8,  8>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              1,      true,   S<1, 4, 16, 4>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,   128,   128,    32,     4,  8,   32,   32,    2,    1,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 4,  8>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              1,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,   128,    32,   128,     4,  8,   32,   32,    1,    2,  S<1, 4, 4,  8>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              1,      true,   S<1, 4, 16, 2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,    64,    64,    32,     4,  8,   32,   32,    2,    1,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 4,  4>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 16, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,    64,    32,    64,     4,  8,   32,   32,    1,    2,  S<1, 4, 4,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,   S<1, 4, 8,  2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 16, 1, 4>,               4>
+    // clang-format on
+    >;
+
+void add_device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_bf16_f32_bf16_instances(
+    std::vector<std::unique_ptr<DeviceConvBwdWeight<1,
+                                                    NWC,
+                                                    KXC,
+                                                    NWK,
+                                                    BF16,
+                                                    F32,
+                                                    BF16,
+                                                    PassThrough,
+                                                    PassThrough,
+                                                    PassThrough>>>& instances)
+{
+    add_device_operation_instances(
+        instances, device_conv1d_bwd_weight_xdl_c_shuffle_nwc_kxc_nwk_bf16_f32_bf16_instances{});
+    add_device_operation_instances(
+        instances, device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_1x1_s1_p0_bf16_f32_bf16_instances{});
+}
+
+} // namespace instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck

library/src/tensor_operation_instance/gpu/conv1d_bwd_weight/device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_f16_instance.cpp

+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include <cstdlib>
+
+#include "ck/ck.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/device/device_convnd_bwd_weight_nwc_kxc_nwk_xdl_cshuffle.hpp"
+#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+
+#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace instance {
+
+using F16 = ck::half_t;
+using F32 = float;
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using NWC = ck::tensor_layout::convolution::NWC;
+using KXC = ck::tensor_layout::convolution::KXC;
+using NWK = ck::tensor_layout::convolution::NWK;
+
+using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+
+static constexpr auto ConvBwdWeightDefault =
+    ck::tensor_operation::device::ConvolutionBackwardWeightSpecialization::Default;
+
+static constexpr auto ConvBwdWeightFilter1x1Stride1Pad0 =
+    ck::tensor_operation::device::ConvolutionBackwardWeightSpecialization::Filter1x1Stride1Pad0;
+
+// Compilation parameters for in[n, wi, c] * wei[k, x, c] = out[n, wo, k]
+using device_conv1d_bwd_weight_xdl_c_shuffle_nwc_kxc_nwk_f16_default_instances = std::tuple<
+    // clang-format off
+        //#########################################|     Num| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                      ConvBackward| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|   CBlockTransfer|  CBlockTransfer|
+        //#########################################|     Dim|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                            Weight|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|   ClusterLengths| ScalarPerVector|
+        //#########################################| Spatial|       |        |        |        |   Operation|   Operation|   Operation|                    Specialization|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle| MBlock_MPerBlock|    NWaveNPerXdl|
+        //#########################################|        |       |        |        |        |            |            |            |                                  |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            | NBlock_NPerBlock|                |
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   256,   256,   128,     4,  8,   32,   32,    4,    2,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 8>,               8>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   256,   128,   256,     4,  8,   32,   32,    2,    4,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 8>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   128,   128,   128,     4,  8,   32,   32,    4,    2,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   256,   128,   128,     4,  8,   32,   32,    2,    2,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   128,   128,    64,     4,  8,   32,   32,    2,    2,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 8,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   128,    64,   128,     4,  8,   32,   32,    2,    2,  S<1, 4, 8,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,    64,    64,    64,     4,  8,   32,   32,    2,    2,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 16, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   256,   128,    64,     4,  8,   32,   32,    2,    1,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,  S<1, 4, 8,  8>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              1,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   256,    64,   128,     4,  8,   32,   32,    1,    2,  S<1, 4, 8,  8>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              1,      true,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   128,   128,    32,     4,  8,   32,   32,    2,    1,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 4,  8>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              1,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   128,    32,   128,     4,  8,   32,   32,    1,    2,  S<1, 4, 4,  8>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              1,      true,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,    64,    64,    32,     4,  8,   32,   32,    2,    1,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 4,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 16, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,    64,    32,    64,     4,  8,   32,   32,    1,    2,  S<1, 4, 4,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 16, 1, 4>,               8>
+    // clang-format on
+    >;
+
+using device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_1x1_s1_p0_f16_instances = std::tuple<
+    // clang-format off
+        //#########################################|     Num| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                      ConvBackward| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|   CBlockTransfer|  CBlockTransfer|
+        //#########################################|     Dim|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                            Weight|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|   ClusterLengths| ScalarPerVector|
+        //#########################################| Spatial|       |        |        |        |   Operation|   Operation|   Operation|                    Specialization|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle| MBlock_MPerBlock|    NWaveNPerXdl|
+        //#########################################|        |       |        |        |        |            |            |            |                                  |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            | NBlock_NPerBlock|                |
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   256,   256,   128,     4,  8,   32,   32,    4,    2,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 8>,               8>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   256,   128,   256,     4,  8,   32,   32,    2,    4,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 8>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   128,   128,   128,     4,  8,   32,   32,    4,    2,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   256,   128,   128,     4,  8,   32,   32,    2,    2,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   128,   128,    64,     4,  8,   32,   32,    2,    2,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 8,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   128,    64,   128,     4,  8,   32,   32,    2,    2,  S<1, 4, 8,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,    64,    64,    64,     4,  8,   32,   32,    2,    2,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 16, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   256,   128,    64,     4,  8,   32,   32,    2,    1,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,  S<1, 4, 8,  8>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              1,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   256,    64,   128,     4,  8,   32,   32,    1,    2,  S<1, 4, 8,  8>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              1,      true,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   128,   128,    32,     4,  8,   32,   32,    2,    1,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 4,  8>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              1,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   128,    32,   128,     4,  8,   32,   32,    1,    2,  S<1, 4, 4,  8>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              1,      true,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,    64,    64,    32,     4,  8,   32,   32,    2,    1,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 4,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 16, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,    64,    32,    64,     4,  8,   32,   32,    1,    2,  S<1, 4, 4,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 16, 1, 4>,               8>
+    // clang-format on
+    >;
+
+void add_device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_f16_instances(
+    std::vector<std::unique_ptr<DeviceConvBwdWeight<1,
+                                                    NWC,
+                                                    KXC,
+                                                    NWK,
+                                                    F16,
+                                                    F16,
+                                                    F16,
+                                                    PassThrough,
+                                                    PassThrough,
+                                                    PassThrough>>>& instances)
+{
+    add_device_operation_instances(
+        instances, device_conv1d_bwd_weight_xdl_c_shuffle_nwc_kxc_nwk_f16_default_instances{});
+    add_device_operation_instances(
+        instances, device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_1x1_s1_p0_f16_instances{});
+}
+
+} // namespace instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck

library/src/tensor_operation_instance/gpu/conv1d_bwd_weight/device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_f32_instance.cpp

+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include <cstdlib>
+
+#include "ck/ck.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/device/device_convnd_bwd_weight_nwc_kxc_nwk_xdl_cshuffle.hpp"
+#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+
+#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace instance {
+
+using F32 = float;
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using NWC = ck::tensor_layout::convolution::NWC;
+using KXC = ck::tensor_layout::convolution::KXC;
+using NWK = ck::tensor_layout::convolution::NWK;
+
+using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+
+static constexpr auto ConvBwdWeightDefault =
+    ck::tensor_operation::device::ConvolutionBackwardWeightSpecialization::Default;
+
+static constexpr auto ConvBwdWeightFilter1x1Stride1Pad0 =
+    ck::tensor_operation::device::ConvolutionBackwardWeightSpecialization::Filter1x1Stride1Pad0;
+
+// Compilation parameters for in[n, wi, c] * wei[k, x, c] = out[n, wo, k]
+using device_conv1d_bwd_weight_xdl_c_shuffle_nwc_kxc_nwk_f32_default_instances = std::tuple<
+    // clang-format off
+        //#########################################|     Num| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                      ConvBackward| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer|   ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|   CBlockTransfer|  CBlockTransfer|
+        //#########################################|     Dim|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                            Weight|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|    ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|   ClusterLengths| ScalarPerVector|
+        //#########################################| Spatial|       |        |        |        |   Operation|   Operation|   Operation|                    Specialization|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|     ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle| MBlock_MPerBlock|    NWaveNPerXdl|
+        //#########################################|        |       |        |        |        |            |            |            |                                  |      |      |      |      |   |     |     |     |     |                |                 |               |               |               |               |          |                |               |               |              |               |               |          |            |            | NBlock_NPerBlock|                |
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   256,   256,   128,     4,  4,   32,   32,    4,    2,  S<1, 4, 64, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              2,      true,           1,           1,   S<1, 32, 1, 8>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   256,   128,   256,     4,  4,   32,   32,    2,    4,  S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              2,      true,  S<1, 4, 64, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 32, 1, 8>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   128,   128,   128,     4,  4,   32,   32,    4,    2,  S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 32, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   256,   128,   128,     4,  4,   32,   32,    2,    2,  S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              2,      true,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   128,   128,    64,     4,  4,   32,   32,    2,    2,  S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   128,    64,   128,     4,  4,   32,   32,    2,    2,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              2,      true,  S<1, 4, 32, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,    64,    64,    64,     4,  4,   32,   32,    2,    2,  S<1, 4, 16, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 16, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 16, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   256,   128,    64,     4,  4,   32,   32,    2,    1,  S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              2,      true,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              1,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   256,    64,   128,     4,  4,   32,   32,    1,    2,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              1,      true,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   128,   128,    32,     4,  4,   32,   32,    2,    1,  S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 8,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              1,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   128,    32,   128,     4,  4,   32,   32,    1,    2,  S<1, 4, 8,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              1,      true,  S<1, 4, 32, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,    64,    64,    32,     4,  4,   32,   32,    2,    1,  S<1, 4, 16, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              2,      true,           1,           1,   S<1, 16, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,    64,    32,    64,     4,  4,   32,   32,    1,    2,  S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              2,      true,  S<1, 4, 16, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 16, 1, 4>,               4>
+    // clang-format on
+    >;
+
+using device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_1x1_s1_p0_f32_instances = std::tuple<
+    // clang-format off
+        //#########################################|     Num| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                      ConvBackward| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|   CBlockTransfer|  CBlockTransfer|
+        //#########################################|     Dim|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                            Weight|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|   ClusterLengths| ScalarPerVector|
+        //#########################################| Spatial|       |        |        |        |   Operation|   Operation|   Operation|                    Specialization|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle| MBlock_MPerBlock|    NWaveNPerXdl|
+        //#########################################|        |       |        |        |        |            |            |            |                                  |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            | NBlock_NPerBlock|                |
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   256,   256,   128,     4,  4,   32,   32,    4,    2,  S<1, 4, 64, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              2,      true,           1,           1,   S<1, 32, 1, 8>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   256,   128,   256,     4,  4,   32,   32,    2,    4,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              2,      true,  S<1, 4, 64, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 32, 1, 8>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   128,   128,   128,     4,  4,   32,   32,    4,    2,  S<1, 4, 32, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 32, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   256,   128,   128,     4,  4,   32,   32,    2,    2,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              2,      true,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   128,   128,    64,     4,  4,   32,   32,    2,    2,  S<1, 4, 32, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   128,    64,   128,     4,  4,   32,   32,    2,    2,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              2,      true,  S<1, 4, 32, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,    64,    64,    64,     4,  4,   32,   32,    2,    2,  S<1, 4, 16, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 16, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 16, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   256,   128,    64,     4,  4,   32,   32,    2,    1,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              2,      true,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              1,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   256,    64,   128,     4,  4,   32,   32,    1,    2,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              1,      true,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   128,   128,    32,     4,  4,   32,   32,    2,    1,  S<1, 4, 32, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 8,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              1,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   128,    32,   128,     4,  4,   32,   32,    1,    2,  S<1, 4, 8,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              1,      true,  S<1, 4, 32, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,    64,    64,    32,     4,  4,   32,   32,    2,    1,  S<1, 4, 16, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              2,      true,           1,           1,   S<1, 16, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       1,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,    64,    32,    64,     4,  4,   32,   32,    1,    2,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              2,      true,  S<1, 4, 16, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 16, 1, 4>,               4>
+    // clang-format on
+    >;
+
+void add_device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_f32_instances(
+    std::vector<std::unique_ptr<DeviceConvBwdWeight<1,
+                                                    NWC,
+                                                    KXC,
+                                                    NWK,
+                                                    F32,
+                                                    F32,
+                                                    F32,
+                                                    PassThrough,
+                                                    PassThrough,
+                                                    PassThrough>>>& instances)
+{
+    add_device_operation_instances(
+        instances, device_conv1d_bwd_weight_xdl_c_shuffle_nwc_kxc_nwk_f32_default_instances{});
+    add_device_operation_instances(
+        instances, device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_1x1_s1_p0_f32_instances{});
+}
+
+} // namespace instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck

library/src/tensor_operation_instance/gpu/conv2d_bwd_weight/CMakeLists.txt

-# device_conv2d_bwd_weight_instance
+#device_conv2d_bwd_weight_instance
 set(DEVICE_CONV2D_BWD_WEIGHT_INSTANCE_SOURCE 
     device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_f16_instance.cpp;
     device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_f32_instance.cpp;
+    device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_bf16_instance.cpp;
 )
+
 add_library(device_conv2d_bwd_weight_instance OBJECT ${DEVICE_CONV2D_BWD_WEIGHT_INSTANCE_SOURCE})
 target_compile_features(device_conv2d_bwd_weight_instance PUBLIC) 
 set_target_properties(device_conv2d_bwd_weight_instance PROPERTIES POSITION_INDEPENDENT_CODE ON)

library/src/tensor_operation_instance/gpu/conv2d_bwd_weight/device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_bf16_instance.cpp

+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include <cstdlib>
+
+#include "ck/ck.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/device/device_convnd_bwd_weight_nwc_kxc_nwk_xdl_cshuffle.hpp"
+#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+
+#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace instance {
+
+using BF16 = bhalf_t;
+using F32  = float;
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using NHWC = ck::tensor_layout::convolution::NHWC;
+using KYXC = ck::tensor_layout::convolution::KYXC;
+using NHWK = ck::tensor_layout::convolution::NHWK;
+
+using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+
+static constexpr auto ConvBwdWeightDefault =
+    ck::tensor_operation::device::ConvolutionBackwardWeightSpecialization::Default;
+
+static constexpr auto ConvBwdWeightFilter1x1Stride1Pad0 =
+    ck::tensor_operation::device::ConvolutionBackwardWeightSpecialization::Filter1x1Stride1Pad0;
+
+// Compilation parameters for in[n, hi, wi, c] * wei[k, y, x, c] = out[n, ho, wo, k]
+using device_conv2d_bwd_weight_xdl_c_shuffle_nhwc_kyxc_nhwk_bf16_f32_bf16_instances = std::tuple<
+    // clang-format off
+        //#########################################|     Num| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                       ConvBackward| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|   BBlockTransfer|  BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|   CBlockTransfer|  CBlockTransfer|
+        //#########################################|     Dim|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                             Weight|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|    ThreadCluster|   ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|   ClusterLengths| ScalarPerVector|
+        //#########################################| Spatial|       |        |        |        |   Operation|   Operation|   Operation|                     Specialization|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          |  Lengths_K0_N_K1|    ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle| MBlock_MPerBlock|    NWaveNPerXdl|
+        //#########################################|        |       |        |        |        |            |            |            |                                   |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                 |                |               |              |               |               |          |            |            | NBlock_NPerBlock|                |
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,   256,   256,   128,     4,  8,   32,   32,    4,    2,  S<1, 4, 32, 2>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 16, 4>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 8>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,   256,   128,   256,     4,  8,   32,   32,    2,    4,  S<1, 4, 16, 4>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              2,      true,   S<1, 4, 32, 2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 8>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,   128,   128,   128,     4,  8,   32,   32,    4,    2,  S<1, 4, 16, 2>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 16, 2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,   256,   128,   128,     4,  8,   32,   32,    2,    2,  S<1, 4, 16, 4>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              2,      true,   S<1, 4, 16, 4>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,   128,   128,    64,     4,  8,   32,   32,    2,    2,  S<1, 4, 16, 2>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 8,  4>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,   128,    64,   128,     4,  8,   32,   32,    2,    2,  S<1, 4, 8,  4>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              2,      true,   S<1, 4, 16, 2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,    64,    64,    64,     4,  8,   32,   32,    2,    2,  S<1, 4, 8,  2>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 8,  2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 16, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,   256,   128,    64,     4,  8,   32,   32,    2,    1,  S<1, 4, 16, 4>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              2,      true,   S<1, 4, 8,  8>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              1,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,   256,    64,   128,     4,  8,   32,   32,    1,    2,  S<1, 4, 8,  8>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              1,      true,   S<1, 4, 16, 4>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,   128,   128,    32,     4,  8,   32,   32,    2,    1,  S<1, 4, 16, 2>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 4,  8>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              1,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,   128,    32,   128,     4,  8,   32,   32,    1,    2,  S<1, 4, 4,  8>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              1,      true,   S<1, 4, 16, 2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,    64,    64,    32,     4,  8,   32,   32,    2,    1,  S<1, 4, 8,  2>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 4,  4>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 16, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,    64,    32,    64,     4,  8,   32,   32,    1,    2,  S<1, 4, 4,  4>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              2,      true,   S<1, 4, 8,  2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 16, 1, 4>,               4>
+    // clang-format on
+    >;
+
+using device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_1x1_s1_p0_bf16_f32_bf16_instances = std::tuple<
+    // clang-format off
+        //#########################################|     Num| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                       ConvBackward| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|   BBlockTransfer|  BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|   CBlockTransfer|  CBlockTransfer|
+        //#########################################|     Dim|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                             Weight|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|    ThreadCluster|   ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|   ClusterLengths| ScalarPerVector|
+        //#########################################| Spatial|       |        |        |        |   Operation|   Operation|   Operation|                     Specialization|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          |  Lengths_K0_N_K1|    ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle| MBlock_MPerBlock|    NWaveNPerXdl|
+        //#########################################|        |       |        |        |        |            |            |            |                                   |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                 |                |               |              |               |               |          |            |            | NBlock_NPerBlock|                |
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,   256,   256,   128,     4,  8,   32,   32,    4,    2,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 16, 4>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 8>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,   256,   128,   256,     4,  8,   32,   32,    2,    4,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,   S<1, 4, 32, 2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 8>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,   128,   128,   128,     4,  8,   32,   32,    4,    2,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 16, 2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,   256,   128,   128,     4,  8,   32,   32,    2,    2,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,   S<1, 4, 16, 4>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,   128,   128,    64,     4,  8,   32,   32,    2,    2,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 8,  4>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,   128,    64,   128,     4,  8,   32,   32,    2,    2,  S<1, 4, 8,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,   S<1, 4, 16, 2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,    64,    64,    64,     4,  8,   32,   32,    2,    2,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 8,  2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 16, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,   256,   128,    64,     4,  8,   32,   32,    2,    1,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,   S<1, 4, 8,  8>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              1,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,   256,    64,   128,     4,  8,   32,   32,    1,    2,  S<1, 4, 8,  8>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              1,      true,   S<1, 4, 16, 4>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,   128,   128,    32,     4,  8,   32,   32,    2,    1,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 4,  8>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              1,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,   128,    32,   128,     4,  8,   32,   32,    1,    2,  S<1, 4, 4,  8>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              1,      true,   S<1, 4, 16, 2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,    64,    64,    32,     4,  8,   32,   32,    2,    1,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 4,  4>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 16, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       2,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,    64,    32,    64,     4,  8,   32,   32,    1,    2,  S<1, 4, 4,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,   S<1, 4, 8,  2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 16, 1, 4>,               4>
+    // clang-format on
+    >;
+
+void add_device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_bf16_f32_bf16_instances(
+    std::vector<std::unique_ptr<DeviceConvBwdWeight<2,
+                                                    NHWC,
+                                                    KYXC,
+                                                    NHWK,
+                                                    BF16,
+                                                    F32,
+                                                    BF16,
+                                                    PassThrough,
+                                                    PassThrough,
+                                                    PassThrough>>>& instances)
+{
+    add_device_operation_instances(
+        instances, device_conv2d_bwd_weight_xdl_c_shuffle_nhwc_kyxc_nhwk_bf16_f32_bf16_instances{});
+    add_device_operation_instances(
+        instances, device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_1x1_s1_p0_bf16_f32_bf16_instances{});
+}
+
+} // namespace instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck

library/src/tensor_operation_instance/gpu/conv3d_bwd_weight/CMakeLists.txt

+#device_conv3d_bwd_weight_instance
+set(DEVICE_CONV3D_BWD_WEIGHT_INSTANCE_SOURCE 
+    device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_f16_instance.cpp;
+    device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_f32_instance.cpp;
+    device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_bf16_instance.cpp;
+)
+
+add_library(device_conv3d_bwd_weight_instance OBJECT ${DEVICE_CONV3D_BWD_WEIGHT_INSTANCE_SOURCE})
+target_compile_features(device_conv3d_bwd_weight_instance PUBLIC) 
+set_target_properties(device_conv3d_bwd_weight_instance PROPERTIES POSITION_INDEPENDENT_CODE ON)
+rocm_install(TARGETS device_conv3d_bwd_weight_instance)
+
+clang_tidy_check(device_conv3d_bwd_weight_instance)

library/src/tensor_operation_instance/gpu/conv3d_bwd_weight/device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_bf16_instance.cpp

+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include <cstdlib>
+
+#include "ck/ck.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/device/device_convnd_bwd_weight_nwc_kxc_nwk_xdl_cshuffle.hpp"
+#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+
+#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace instance {
+
+using BF16 = bhalf_t;
+using F32  = float;
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using NDHWC = ck::tensor_layout::convolution::NDHWC;
+using KZYXC = ck::tensor_layout::convolution::KZYXC;
+using NDHWK = ck::tensor_layout::convolution::NDHWK;
+
+using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+
+static constexpr auto ConvBwdWeightDefault =
+    ck::tensor_operation::device::ConvolutionBackwardWeightSpecialization::Default;
+
+static constexpr auto ConvBwdWeightFilter1x1Stride1Pad0 =
+    ck::tensor_operation::device::ConvolutionBackwardWeightSpecialization::Filter1x1Stride1Pad0;
+
+// Compilation parameters for in[n, di, hi, wi, c] * wei[k, z, y, x, c] = out[n, do, ho, wo, k]
+using device_conv3d_bwd_weight_xdl_c_shuffle_ndhwc_kzyxc_ndhwk_bf16_f32_bf16_instances = std::tuple<
+    // clang-format off
+        //#########################################|     Num| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                       ConvBackward| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|   BBlockTransfer|  BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|   CBlockTransfer|  CBlockTransfer|
+        //#########################################|     Dim|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                             Weight|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|    ThreadCluster|   ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|   ClusterLengths| ScalarPerVector|
+        //#########################################| Spatial|       |        |        |        |   Operation|   Operation|   Operation|                     Specialization|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          |  Lengths_K0_N_K1|    ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle| MBlock_MPerBlock|    NWaveNPerXdl|
+        //#########################################|        |       |        |        |        |            |            |            |                                   |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                 |                |               |              |               |               |          |            |            | NBlock_NPerBlock|                |
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,   256,   256,   128,     4,  8,   32,   32,    4,    2,  S<1, 4, 32, 2>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 16, 4>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 8>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,   256,   128,   256,     4,  8,   32,   32,    2,    4,  S<1, 4, 16, 4>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              2,      true,   S<1, 4, 32, 2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 8>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,   128,   128,   128,     4,  8,   32,   32,    4,    2,  S<1, 4, 16, 2>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 16, 2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,   256,   128,   128,     4,  8,   32,   32,    2,    2,  S<1, 4, 16, 4>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              2,      true,   S<1, 4, 16, 4>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,   128,   128,    64,     4,  8,   32,   32,    2,    2,  S<1, 4, 16, 2>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 8,  4>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,   128,    64,   128,     4,  8,   32,   32,    2,    2,  S<1, 4, 8,  4>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              2,      true,   S<1, 4, 16, 2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,    64,    64,    64,     4,  8,   32,   32,    2,    2,  S<1, 4, 8,  2>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 8,  2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 16, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,   256,   128,    64,     4,  8,   32,   32,    2,    1,  S<1, 4, 16, 4>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              2,      true,   S<1, 4, 8,  8>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              1,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,   256,    64,   128,     4,  8,   32,   32,    1,    2,  S<1, 4, 8,  8>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              1,      true,   S<1, 4, 16, 4>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,   128,   128,    32,     4,  8,   32,   32,    2,    1,  S<1, 4, 16, 2>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 4,  8>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              1,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,   128,    32,   128,     4,  8,   32,   32,    1,    2,  S<1, 4, 4,  8>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              1,      true,   S<1, 4, 16, 2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,    64,    64,    32,     4,  8,   32,   32,    2,    1,  S<1, 4, 8,  2>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 4,  4>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 16, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,               ConvBwdWeightDefault,    64,    32,    64,     4,  8,   32,   32,    1,    2,  S<1, 4, 4,  4>, S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              8,              2,      true,   S<1, 4, 8,  2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 16, 1, 4>,               4>
+    // clang-format on
+    >;
+
+using device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_1x1_s1_p0_bf16_f32_bf16_instances = std::tuple<
+    // clang-format off
+        //#########################################|     Num| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                       ConvBackward| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|   BBlockTransfer|  BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|   CBlockTransfer|  CBlockTransfer|
+        //#########################################|     Dim|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                             Weight|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|    ThreadCluster|   ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|   ClusterLengths| ScalarPerVector|
+        //#########################################| Spatial|       |        |        |        |   Operation|   Operation|   Operation|                     Specialization|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          |  Lengths_K0_N_K1|    ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle| MBlock_MPerBlock|    NWaveNPerXdl|
+        //#########################################|        |       |        |        |        |            |            |            |                                   |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                 |                |               |              |               |               |          |            |            | NBlock_NPerBlock|                |
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,   256,   256,   128,     4,  8,   32,   32,    4,    2,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 16, 4>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 8>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,   256,   128,   256,     4,  8,   32,   32,    2,    4,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,   S<1, 4, 32, 2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 8>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,   128,   128,   128,     4,  8,   32,   32,    4,    2,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 16, 2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,   256,   128,   128,     4,  8,   32,   32,    2,    2,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,   S<1, 4, 16, 4>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,   128,   128,    64,     4,  8,   32,   32,    2,    2,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 8,  4>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,   128,    64,   128,     4,  8,   32,   32,    2,    2,  S<1, 4, 8,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,   S<1, 4, 16, 2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,    64,    64,    64,     4,  8,   32,   32,    2,    2,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 8,  2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 16, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,   256,   128,    64,     4,  8,   32,   32,    2,    1,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,   S<1, 4, 8,  8>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              1,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,   256,    64,   128,     4,  8,   32,   32,    1,    2,  S<1, 4, 8,  8>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              1,      true,   S<1, 4, 16, 4>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,   128,   128,    32,     4,  8,   32,   32,    2,    1,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 4,  8>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              1,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,   128,    32,   128,     4,  8,   32,   32,    1,    2,  S<1, 4, 4,  8>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              1,      true,   S<1, 4, 16, 2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,    64,    64,    32,     4,  8,   32,   32,    2,    1,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,   S<1, 4, 4,  4>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 16, 1, 4>,               4>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,   BF16,     F32,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,    64,    32,    64,     4,  8,   32,   32,    1,    2,  S<1, 4, 4,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,   S<1, 4, 8,  2>,   S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 16, 1, 4>,               4>
+    // clang-format on
+    >;
+
+void add_device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_bf16_f32_bf16_instances(
+    std::vector<std::unique_ptr<DeviceConvBwdWeight<3,
+                                                    NDHWC,
+                                                    KZYXC,
+                                                    NDHWK,
+                                                    BF16,
+                                                    F32,
+                                                    BF16,
+                                                    PassThrough,
+                                                    PassThrough,
+                                                    PassThrough>>>& instances)
+{
+    add_device_operation_instances(
+        instances,
+        device_conv3d_bwd_weight_xdl_c_shuffle_ndhwc_kzyxc_ndhwk_bf16_f32_bf16_instances{});
+    add_device_operation_instances(
+        instances,
+        device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_1x1_s1_p0_bf16_f32_bf16_instances{});
+}
+
+} // namespace instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck

library/src/tensor_operation_instance/gpu/conv3d_bwd_weight/device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_f16_instance.cpp

+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include <cstdlib>
+
+#include "ck/ck.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/device/device_convnd_bwd_weight_nwc_kxc_nwk_xdl_cshuffle.hpp"
+#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+
+#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace instance {
+
+using F16 = ck::half_t;
+using F32 = float;
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using NDHWC = ck::tensor_layout::convolution::NDHWC;
+using KZYXC = ck::tensor_layout::convolution::KZYXC;
+using NDHWK = ck::tensor_layout::convolution::NDHWK;
+
+using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+
+static constexpr auto ConvBwdWeightDefault =
+    ck::tensor_operation::device::ConvolutionBackwardWeightSpecialization::Default;
+
+static constexpr auto ConvBwdWeightFilter1x1Stride1Pad0 =
+    ck::tensor_operation::device::ConvolutionBackwardWeightSpecialization::Filter1x1Stride1Pad0;
+
+// Compilation parameters for in[n, di, hi, wi, c] * wei[k, z, y, x, c] = out[n, do, ho, wo, k]
+using device_conv3d_bwd_weight_xdl_c_shuffle_ndhwc_kzyxc_ndhwk_f16_default_instances = std::tuple<
+    // clang-format off
+        //#########################################|     Num| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                      ConvBackward| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|   CBlockTransfer|  CBlockTransfer|
+        //#########################################|     Dim|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                            Weight|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|   ClusterLengths| ScalarPerVector|
+        //#########################################| Spatial|       |        |        |        |   Operation|   Operation|   Operation|                    Specialization|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle| MBlock_MPerBlock|    NWaveNPerXdl|
+        //#########################################|        |       |        |        |        |            |            |            |                                  |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            | NBlock_NPerBlock|                |
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   256,   256,   128,     4,  8,   32,   32,    4,    2,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 8>,               8>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   256,   128,   256,     4,  8,   32,   32,    2,    4,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 8>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   128,   128,   128,     4,  8,   32,   32,    4,    2,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   256,   128,   128,     4,  8,   32,   32,    2,    2,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   128,   128,    64,     4,  8,   32,   32,    2,    2,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 8,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   128,    64,   128,     4,  8,   32,   32,    2,    2,  S<1, 4, 8,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,    64,    64,    64,     4,  8,   32,   32,    2,    2,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 16, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   256,   128,    64,     4,  8,   32,   32,    2,    1,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,  S<1, 4, 8,  8>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              1,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   256,    64,   128,     4,  8,   32,   32,    1,    2,  S<1, 4, 8,  8>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              1,      true,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   128,   128,    32,     4,  8,   32,   32,    2,    1,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 4,  8>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              1,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   128,    32,   128,     4,  8,   32,   32,    1,    2,  S<1, 4, 4,  8>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              1,      true,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,    64,    64,    32,     4,  8,   32,   32,    2,    1,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 4,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 16, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,    64,    32,    64,     4,  8,   32,   32,    1,    2,  S<1, 4, 4,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 16, 1, 4>,               8>
+    // clang-format on
+    >;
+
+using device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_1x1_s1_p0_f16_instances = std::tuple<
+    // clang-format off
+        //#########################################|     Num| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                      ConvBackward| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|   CBlockTransfer|  CBlockTransfer|
+        //#########################################|     Dim|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                            Weight|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|   ClusterLengths| ScalarPerVector|
+        //#########################################| Spatial|       |        |        |        |   Operation|   Operation|   Operation|                    Specialization|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle| MBlock_MPerBlock|    NWaveNPerXdl|
+        //#########################################|        |       |        |        |        |            |            |            |                                  |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            | NBlock_NPerBlock|                |
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   256,   256,   128,     4,  8,   32,   32,    4,    2,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 8>,               8>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   256,   128,   256,     4,  8,   32,   32,    2,    4,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 8>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   128,   128,   128,     4,  8,   32,   32,    4,    2,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   256,   128,   128,     4,  8,   32,   32,    2,    2,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   128,   128,    64,     4,  8,   32,   32,    2,    2,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 8,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   128,    64,   128,     4,  8,   32,   32,    2,    2,  S<1, 4, 8,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,    64,    64,    64,     4,  8,   32,   32,    2,    2,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 16, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   256,   128,    64,     4,  8,   32,   32,    2,    1,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,  S<1, 4, 8,  8>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              1,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   256,    64,   128,     4,  8,   32,   32,    1,    2,  S<1, 4, 8,  8>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              1,      true,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   128,   128,    32,     4,  8,   32,   32,    2,    1,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 4,  8>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              1,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   128,    32,   128,     4,  8,   32,   32,    1,    2,  S<1, 4, 4,  8>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              1,      true,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 32, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,    64,    64,    32,     4,  8,   32,   32,    2,    1,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              4,      true,  S<1, 4, 4,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              2,      true,           1,           1,   S<1, 16, 1, 4>,               8>,   
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,    64,    32,    64,     4,  8,   32,   32,    1,    2,  S<1, 4, 4,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              8,              2,      true,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              8,              4,      true,           1,           1,   S<1, 16, 1, 4>,               8>
+    // clang-format on
+    >;
+
+void add_device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_f16_instances(
+    std::vector<std::unique_ptr<DeviceConvBwdWeight<3,
+                                                    NDHWC,
+                                                    KZYXC,
+                                                    NDHWK,
+                                                    F16,
+                                                    F16,
+                                                    F16,
+                                                    PassThrough,
+                                                    PassThrough,
+                                                    PassThrough>>>& instances)
+{
+    add_device_operation_instances(
+        instances,
+        device_conv3d_bwd_weight_xdl_c_shuffle_ndhwc_kzyxc_ndhwk_f16_default_instances{});
+    add_device_operation_instances(
+        instances, device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_1x1_s1_p0_f16_instances{});
+}
+
+} // namespace instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck

library/src/tensor_operation_instance/gpu/conv3d_bwd_weight/device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_f32_instance.cpp

+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+
+#include <cstdlib>
+
+#include "ck/ck.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/tensor_operation/gpu/device/device_convnd_bwd_weight_nwc_kxc_nwk_xdl_cshuffle.hpp"
+#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
+
+#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace instance {
+
+using F32 = float;
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using NDHWC = ck::tensor_layout::convolution::NDHWC;
+using KZYXC = ck::tensor_layout::convolution::KZYXC;
+using NDHWK = ck::tensor_layout::convolution::NDHWK;
+
+using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+
+static constexpr auto ConvBwdWeightDefault =
+    ck::tensor_operation::device::ConvolutionBackwardWeightSpecialization::Default;
+
+static constexpr auto ConvBwdWeightFilter1x1Stride1Pad0 =
+    ck::tensor_operation::device::ConvolutionBackwardWeightSpecialization::Filter1x1Stride1Pad0;
+
+// Compilation parameters for in[n, di, hi, wi, c] * wei[k, z, y, x, c] = out[n, do, ho, wo, k]
+using device_conv3d_bwd_weight_xdl_c_shuffle_ndhwc_kzyxc_ndhwk_f32_default_instances = std::tuple<
+    // clang-format off
+        //#########################################|     Num| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                      ConvBackward| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer|   ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|   CBlockTransfer|  CBlockTransfer|
+        //#########################################|     Dim|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                            Weight|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|    ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|   ClusterLengths| ScalarPerVector|
+        //#########################################| Spatial|       |        |        |        |   Operation|   Operation|   Operation|                    Specialization|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|     ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle| MBlock_MPerBlock|    NWaveNPerXdl|
+        //#########################################|        |       |        |        |        |            |            |            |                                  |      |      |      |      |   |     |     |     |     |                |                 |               |               |               |               |          |                |               |               |              |               |               |          |            |            | NBlock_NPerBlock|                |
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   256,   256,   128,     4,  4,   32,   32,    4,    2,  S<1, 4, 64, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              2,      true,           1,           1,   S<1, 32, 1, 8>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   256,   128,   256,     4,  4,   32,   32,    2,    4,  S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              2,      true,  S<1, 4, 64, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 32, 1, 8>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   128,   128,   128,     4,  4,   32,   32,    4,    2,  S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 32, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   256,   128,   128,     4,  4,   32,   32,    2,    2,  S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              2,      true,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   128,   128,    64,     4,  4,   32,   32,    2,    2,  S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   128,    64,   128,     4,  4,   32,   32,    2,    2,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              2,      true,  S<1, 4, 32, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,    64,    64,    64,     4,  4,   32,   32,    2,    2,  S<1, 4, 16, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 16, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 16, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   256,   128,    64,     4,  4,   32,   32,    2,    1,  S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              2,      true,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              1,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   256,    64,   128,     4,  4,   32,   32,    1,    2,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              1,      true,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   128,   128,    32,     4,  4,   32,   32,    2,    1,  S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 8,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              1,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,   128,    32,   128,     4,  4,   32,   32,    1,    2,  S<1, 4, 8,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              1,      true,  S<1, 4, 32, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,    64,    64,    32,     4,  4,   32,   32,    2,    1,  S<1, 4, 16, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              2,      true,           1,           1,   S<1, 16, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,              ConvBwdWeightDefault,    64,    32,    64,     4,  4,   32,   32,    1,    2,  S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,              2,              4,              2,      true,  S<1, 4, 16, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 16, 1, 4>,               4>
+    // clang-format on
+    >;
+
+using device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_1x1_s1_p0_f32_instances = std::tuple<
+    // clang-format off
+        //#########################################|     Num| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                      ConvBackward| Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|  BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|   CBlockTransfer|  CBlockTransfer|
+        //#########################################|     Dim|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                            Weight|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|   ClusterLengths| ScalarPerVector|
+        //#########################################| Spatial|       |        |        |        |   Operation|   Operation|   Operation|                    Specialization|      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle| MBlock_MPerBlock|    NWaveNPerXdl|
+        //#########################################|        |       |        |        |        |            |            |            |                                  |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |            |            | NBlock_NPerBlock|                |
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   256,   256,   128,     4,  4,   32,   32,    4,    2,  S<1, 4, 64, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              2,      true,           1,           1,   S<1, 32, 1, 8>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   256,   128,   256,     4,  4,   32,   32,    2,    4,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              2,      true,  S<1, 4, 64, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 32, 1, 8>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   128,   128,   128,     4,  4,   32,   32,    4,    2,  S<1, 4, 32, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 32, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   256,   128,   128,     4,  4,   32,   32,    2,    2,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              2,      true,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   128,   128,    64,     4,  4,   32,   32,    2,    2,  S<1, 4, 32, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   128,    64,   128,     4,  4,   32,   32,    2,    2,  S<1, 4, 16, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              2,      true,  S<1, 4, 32, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,    64,    64,    64,     4,  4,   32,   32,    2,    2,  S<1, 4, 16, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 16, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 16, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   256,   128,    64,     4,  4,   32,   32,    2,    1,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              2,      true,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              1,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   256,    64,   128,     4,  4,   32,   32,    1,    2,  S<1, 4, 16, 4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              1,      true,  S<1, 4, 32, 2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              2,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   128,   128,    32,     4,  4,   32,   32,    2,    1,  S<1, 4, 32, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 8,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              1,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,   128,    32,   128,     4,  4,   32,   32,    1,    2,  S<1, 4, 8,  4>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              1,      true,  S<1, 4, 32, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 32, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,    64,    64,    32,     4,  4,   32,   32,    2,    1,  S<1, 4, 16, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              4,      true,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              2,      true,           1,           1,   S<1, 16, 1, 4>,               4>,
+        DeviceConvNdBwdWeightNwcKxcNwk_Xdl_CShuffle<       3,    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough, ConvBwdWeightFilter1x1Stride1Pad0,    64,    32,    64,     4,  4,   32,   32,    1,    2,  S<1, 4, 8,  2>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,              2,              4,              2,      true,  S<1, 4, 16, 1>,  S<0, 3, 1, 2>,  S<0, 2, 1, 3>,             2,              4,              4,      true,           1,           1,   S<1, 16, 1, 4>,               4>
+    // clang-format on
+    >;
+
+void add_device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_f32_instances(
+    std::vector<std::unique_ptr<DeviceConvBwdWeight<3,
+                                                    NDHWC,
+                                                    KZYXC,
+                                                    NDHWK,
+                                                    F32,
+                                                    F32,
+                                                    F32,
+                                                    PassThrough,
+                                                    PassThrough,
+                                                    PassThrough>>>& instances)
+{
+    add_device_operation_instances(
+        instances,
+        device_conv3d_bwd_weight_xdl_c_shuffle_ndhwc_kzyxc_ndhwk_f32_default_instances{});
+    add_device_operation_instances(
+        instances, device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_1x1_s1_p0_f32_instances{});
+}
+
+} // namespace instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck

library/src/tensor_operation_instance/gpu/convnd_bwd_weight/CMakeLists.txt

-#device_convnd_bwd_weight_instance
-set(DEVICE_CONVND_BWD_WEIGHT_INSTANCE_SOURCE 
-    device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_f16_instance.cpp;
-    device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_f32_instance.cpp;
-    device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_bf16_instance.cpp;
-    device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_f16_instance.cpp;
-    device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_f32_instance.cpp;
-    device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_bf16_instance.cpp;
-    device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_f16_instance.cpp;
-    device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_f32_instance.cpp;
-    device_conv3d_bwd_weight_xdl_ndhwc_kzyxc_ndhwk_bf16_instance.cpp;
-)
-
-add_library(device_convnd_bwd_weight_instance OBJECT ${DEVICE_CONVND_BWD_WEIGHT_INSTANCE_SOURCE})
-target_compile_features(device_convnd_bwd_weight_instance PUBLIC) 
-set_target_properties(device_convnd_bwd_weight_instance PROPERTIES POSITION_INDEPENDENT_CODE ON)
-rocm_install(TARGETS device_convnd_bwd_weight_instance)
-
-clang_tidy_check(device_convnd_bwd_weight_instance)

library/src/tensor_operation_instance/gpu/convnd_bwd_weight/device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_bf16_instance.cpp

-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include <cstdlib>
-#include <numeric>
-#include "ck/ck.hpp"
-#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/device_convnd_backward_weight_xdl_c_shuffle_nhwc_kyxc_nhwk.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
-#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
-
-namespace ck {
-namespace tensor_operation {
-namespace device {
-namespace instance {
-
-using BF16 = bhalf_t;
-using F32  = float;
-
-template <ck::index_t... Is>
-using S = ck::Sequence<Is...>;
-
-using PassThrough = ck::tensor_operation::element_wise::PassThrough;
-static constexpr auto ConvBwdWeightDefault =
-    ck::tensor_operation::device::ConvolutionBackwardWeightSpecialization::Default;
-
-static constexpr auto ConvBwdWeightFilter1x1Stride1Pad0 =
-    ck::tensor_operation::device::ConvolutionBackwardWeightSpecialization::Filter1x1Stride1Pad0;
-
-// Compilation parameters for in[n, hi, wi, c] * wei[k, y, x, c] = out[n, ho, wo, k]
-using device_conv1d_bwd_weight_xdl_c_shuffle_nwc_kxc_nwk_bf16_instances = std::tuple<
-    // clang-format off
-        //#################################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|          ConvBackward|    Num|  Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer|   ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|   BBlockTransfer|  BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|    CBlockTransfer| CBlockTransfer|
-        //#################################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                Weight|    Dim|   Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|    ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|    ThreadCluster|   ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|    ClusterLengths|ScalarPerVector|
-        //#################################################################################|       |        |        |        |   Operation|   Operation|   Operation|        Specialization|Spatial|       |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|     ArrangeOrder|               |               |      PerVector|   PerVector_K1|          |  Lengths_K0_N_K1|    ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|  MBlock_MPerBlock|   NWaveNPerXdl|
-        //#################################################################################|       |        |        |        |            |            |            |                      |       |       |      |      |      |   |     |     |     |     |                |                 |               |               |               |               |          |                 |                |               |              |               |               |          |            |            |  NBlock_NPerBlock|               |
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,      256,   256,   128,    4,  8,   32,   32,    4,    2,  S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 8>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,      256,   128,   256,    4,  8,   32,   32,    2,    4,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 8>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,      128,   128,   128,    4,  8,   32,   32,    4,    2,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,      256,   128,   128,    4,  8,   32,   32,    2,    2,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,      128,   128,    64,    4,  8,   32,   32,    2,    2,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 8,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,      128,    64,   128,    4,  8,   32,   32,    2,    2,  S<1, 4, 8,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,       64,    64,    64,    4,  8,   32,   32,    2,    2,  S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 16, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,      256,   128,    64,    4,  8,   32,   32,    2,    1,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 8,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,      256,    64,   128,    4,  8,   32,   32,    1,    2,  S<1, 4, 8,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,      128,   128,    32,    4,  8,   32,   32,    2,    1,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 4,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,      128,    32,   128,    4,  8,   32,   32,    1,    2,  S<1, 4, 4,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,       64,    64,    32,    4,  8,   32,   32,    2,    1,  S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 4,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 16, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,       64,    32,    64,    4,  8,   32,   32,    1,    2,  S<1, 4, 4,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 16, 1, 4>,        4>
-    // clang-format on
-    >;
-
-using device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_1x1_s1_p0_bf16_instances = std::tuple<
-    // clang-format off
-        //#################################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                       ConvBackward|    Num|  Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer|   ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|   BBlockTransfer|  BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|    CBlockTransfer| CBlockTransfer|
-        //#################################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                             Weight|    Dim|   Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|    ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|    ThreadCluster|   ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|    ClusterLengths|ScalarPerVector|
-        //#################################################################################|       |        |        |        |   Operation|   Operation|   Operation|                     Specialization|Spatial|       |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|     ArrangeOrder|               |               |      PerVector|   PerVector_K1|          |  Lengths_K0_N_K1|    ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|  MBlock_MPerBlock|   NWaveNPerXdl|
-        //#################################################################################|       |        |        |        |            |            |            |                                   |       |       |      |      |      |   |     |     |     |     |                |                 |               |               |               |               |          |                 |                |               |              |               |               |          |            |            |  NBlock_NPerBlock|               |
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,      256,   256,   128,    4,  8,   32,   32,    4,    2,  S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 8>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,      256,   128,   256,    4,  8,   32,   32,    2,    4,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 8>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,      128,   128,   128,    4,  8,   32,   32,    4,    2,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,      256,   128,   128,    4,  8,   32,   32,    2,    2,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,      128,   128,    64,    4,  8,   32,   32,    2,    2,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 8,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,      128,    64,   128,    4,  8,   32,   32,    2,    2,  S<1, 4, 8,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,       64,    64,    64,    4,  8,   32,   32,    2,    2,  S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 16, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,      256,   128,    64,    4,  8,   32,   32,    2,    1,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 8,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,      256,    64,   128,    4,  8,   32,   32,    1,    2,  S<1, 4, 8,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,      128,   128,    32,    4,  8,   32,   32,    2,    1,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 4,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,      128,    32,   128,    4,  8,   32,   32,    1,    2,  S<1, 4, 4,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,       64,    64,    32,    4,  8,   32,   32,    2,    1,  S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 4,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 16, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,       64,    32,    64,    4,  8,   32,   32,    1,    2,  S<1, 4, 4,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 16, 1, 4>,        4>
-    // clang-format on
-    >;
-
-void add_device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_bf16_instances(
-    std::vector<DeviceConvBwdWeightPtr<PassThrough, PassThrough, PassThrough>>& instances)
-{
-    add_device_operation_instances(
-        instances, device_conv1d_bwd_weight_xdl_c_shuffle_nwc_kxc_nwk_bf16_instances{});
-    add_device_operation_instances(
-        instances, device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_1x1_s1_p0_bf16_instances{});
-}
-
-} // namespace instance
-} // namespace device
-} // namespace tensor_operation
-} // namespace ck

library/src/tensor_operation_instance/gpu/convnd_bwd_weight/device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_f16_instance.cpp

-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include <cstdlib>
-#include <numeric>
-#include "ck/ck.hpp"
-#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/device_convnd_backward_weight_xdl_c_shuffle_nhwc_kyxc_nhwk.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
-#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
-
-namespace ck {
-namespace tensor_operation {
-namespace device {
-namespace instance {
-
-using F16 = ck::half_t;
-using F32 = float;
-
-template <ck::index_t... Is>
-using S = ck::Sequence<Is...>;
-
-using PassThrough = ck::tensor_operation::element_wise::PassThrough;
-static constexpr auto ConvBwdWeightDefault =
-    ck::tensor_operation::device::ConvolutionBackwardWeightSpecialization::Default;
-
-static constexpr auto ConvBwdWeightFilter1x1Stride1Pad0 =
-    ck::tensor_operation::device::ConvolutionBackwardWeightSpecialization::Filter1x1Stride1Pad0;
-
-// Compilation parameters for in[n, hi, wi, c] * wei[k, y, x, c] = out[n, ho, wo, k]
-using device_conv1d_bwd_weight_xdl_c_shuffle_nwc_kxc_nwk_f16_instances = std::tuple<
-    // clang-format off
-        //#################################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|          ConvBackward|    Num|  Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer|   ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|   BBlockTransfer|  BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|    CBlockTransfer| CBlockTransfer|
-        //#################################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                Weight|    Dim|   Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|    ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|    ThreadCluster|   ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|    ClusterLengths|ScalarPerVector|
-        //#################################################################################|       |        |        |        |   Operation|   Operation|   Operation|        Specialization|Spatial|       |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|     ArrangeOrder|               |               |      PerVector|   PerVector_K1|          |  Lengths_K0_N_K1|    ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|  MBlock_MPerBlock|   NWaveNPerXdl|
-        //#################################################################################|       |        |        |        |            |            |            |                      |       |       |      |      |      |   |     |     |     |     |                |                 |               |               |               |               |          |                 |                |               |              |               |               |          |            |            |  NBlock_NPerBlock|               |
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,      256,   256,   128,    4,  8,   32,   32,    4,    2,  S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 8>,        8>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,      256,   128,   256,    4,  8,   32,   32,    2,    4,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 8>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,      128,   128,   128,    4,  8,   32,   32,    4,    2,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,      256,   128,   128,    4,  8,   32,   32,    2,    2,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,      128,   128,    64,    4,  8,   32,   32,    2,    2,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 8,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,      128,    64,   128,    4,  8,   32,   32,    2,    2,  S<1, 4, 8,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,       64,    64,    64,    4,  8,   32,   32,    2,    2,  S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 16, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,      256,   128,    64,    4,  8,   32,   32,    2,    1,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 8,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,      256,    64,   128,    4,  8,   32,   32,    1,    2,  S<1, 4, 8,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,      128,   128,    32,    4,  8,   32,   32,    2,    1,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 4,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,      128,    32,   128,    4,  8,   32,   32,    1,    2,  S<1, 4, 4,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,       64,    64,    32,    4,  8,   32,   32,    2,    1,  S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 4,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 16, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,       64,    32,    64,    4,  8,   32,   32,    1,    2,  S<1, 4, 4,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 16, 1, 4>,        8>
-    // clang-format on
-    >;
-
-using device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_1x1_s1_p0_f16_instances = std::tuple<
-    // clang-format off
-        //#################################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                       ConvBackward|    Num|  Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer|   ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|   BBlockTransfer|  BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|    CBlockTransfer| CBlockTransfer|
-        //#################################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                             Weight|    Dim|   Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|    ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|    ThreadCluster|   ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|    ClusterLengths|ScalarPerVector|
-        //#################################################################################|       |        |        |        |   Operation|   Operation|   Operation|                     Specialization|Spatial|       |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|     ArrangeOrder|               |               |      PerVector|   PerVector_K1|          |  Lengths_K0_N_K1|    ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|  MBlock_MPerBlock|   NWaveNPerXdl|
-        //#################################################################################|       |        |        |        |            |            |            |                                   |       |       |      |      |      |   |     |     |     |     |                |                 |               |               |               |               |          |                 |                |               |              |               |               |          |            |            |  NBlock_NPerBlock|               |
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,      256,   256,   128,    4,  8,   32,   32,    4,    2,  S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 8>,        8>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,      256,   128,   256,    4,  8,   32,   32,    2,    4,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 8>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,      128,   128,   128,    4,  8,   32,   32,    4,    2,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,      256,   128,   128,    4,  8,   32,   32,    2,    2,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,      128,   128,    64,    4,  8,   32,   32,    2,    2,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 8,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,      128,    64,   128,    4,  8,   32,   32,    2,    2,  S<1, 4, 8,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,       64,    64,    64,    4,  8,   32,   32,    2,    2,  S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 16, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,      256,   128,    64,    4,  8,   32,   32,    2,    1,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 8,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,      256,    64,   128,    4,  8,   32,   32,    1,    2,  S<1, 4, 8,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,      128,   128,    32,    4,  8,   32,   32,    2,    1,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 4,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,      128,    32,   128,    4,  8,   32,   32,    1,    2,  S<1, 4, 4,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,       64,    64,    32,    4,  8,   32,   32,    2,    1,  S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 4,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 16, 1, 4>,        8>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F16,     F16,     F16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,       64,    32,    64,    4,  8,   32,   32,    1,    2,  S<1, 4, 4,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 16, 1, 4>,        8>
-    // clang-format on
-    >;
-
-void add_device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_f16_instances(
-    std::vector<DeviceConvBwdWeightPtr<PassThrough, PassThrough, PassThrough>>& instances)
-{
-    add_device_operation_instances(
-        instances, device_conv1d_bwd_weight_xdl_c_shuffle_nwc_kxc_nwk_f16_instances{});
-    add_device_operation_instances(
-        instances, device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_1x1_s1_p0_f16_instances{});
-}
-
-} // namespace instance
-} // namespace device
-} // namespace tensor_operation
-} // namespace ck

library/src/tensor_operation_instance/gpu/convnd_bwd_weight/device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_f32_instance.cpp

-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include <cstdlib>
-#include <numeric>
-#include "ck/ck.hpp"
-#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/device_convnd_backward_weight_xdl_c_shuffle_nhwc_kyxc_nhwk.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
-#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
-
-namespace ck {
-namespace tensor_operation {
-namespace device {
-namespace instance {
-
-using F32 = float;
-
-template <ck::index_t... Is>
-using S = ck::Sequence<Is...>;
-
-using PassThrough = ck::tensor_operation::element_wise::PassThrough;
-static constexpr auto ConvBwdWeightDefault =
-    ck::tensor_operation::device::ConvolutionBackwardWeightSpecialization::Default;
-
-static constexpr auto ConvBwdWeightFilter1x1Stride1Pad0 =
-    ck::tensor_operation::device::ConvolutionBackwardWeightSpecialization::Filter1x1Stride1Pad0;
-
-// Compilation parameters for in[n, hi, wi, c] * wei[k, y, x, c] = out[n, ho, wo, k]
-using device_conv1d_bwd_weight_xdl_c_shuffle_nwc_kxc_nwk_f32_instances = std::tuple<
-    // clang-format off
-        //#################################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|          ConvBackward|    Num|  Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer|   ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|   BBlockTransfer|  BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|    CBlockTransfer| CBlockTransfer|
-        //#################################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                Weight|    Dim|   Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|    ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|    ThreadCluster|   ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|    ClusterLengths|ScalarPerVector|
-        //#################################################################################|       |        |        |        |   Operation|   Operation|   Operation|        Specialization|Spatial|       |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|     ArrangeOrder|               |               |      PerVector|   PerVector_K1|          |  Lengths_K0_N_K1|    ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|  MBlock_MPerBlock|   NWaveNPerXdl|
-        //#################################################################################|       |        |        |        |            |            |            |                      |       |       |      |      |      |   |     |     |     |     |                |                 |               |               |               |               |          |                 |                |               |              |               |               |          |            |            |  NBlock_NPerBlock|               |
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,      256,   256,   128,    4,  4,   32,   32,    4,    2,  S<1, 4, 64, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,    true,     S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,      true,        1,            1,     S<1, 32, 1, 8>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,      256,   128,   256,    4,  4,   32,   32,    2,    4,  S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,    true,     S<1, 4, 64, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,      true,        1,            1,     S<1, 32, 1, 8>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,      128,   128,   128,    4,  4,   32,   32,    4,    2,  S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,    true,     S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,      256,   128,   128,    4,  4,   32,   32,    2,    2,  S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,    true,     S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,      128,   128,    64,    4,  4,   32,   32,    2,    2,  S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,      128,    64,   128,    4,  4,   32,   32,    2,    2,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,    true,     S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,       64,    64,    64,    4,  4,   32,   32,    2,    2,  S<1, 4, 16, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,    true,     S<1, 4, 16, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,      true,        1,            1,     S<1, 16, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,      256,   128,    64,    4,  4,   32,   32,    2,    1,  S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              1,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,      256,    64,   128,    4,  4,   32,   32,    1,    2,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              1,    true,     S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,      128,   128,    32,    4,  4,   32,   32,    2,    1,  S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,    true,     S<1, 4, 8,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              1,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,      128,    32,   128,    4,  4,   32,   32,    1,    2,  S<1, 4, 8,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              1,    true,     S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,       64,    64,    32,    4,  4,   32,   32,    2,    1,  S<1, 4, 16, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,    true,     S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,      true,        1,            1,     S<1, 16, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     1,       64,    32,    64,    4,  4,   32,   32,    1,    2,  S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,    true,     S<1, 4, 16, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,      true,        1,            1,     S<1, 16, 1, 4>,        4>
-    // clang-format on
-    >;
-
-using device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_1x1_s1_p0_f32_instances = std::tuple<
-    // clang-format off
-        //#################################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                       ConvBackward|    Num|  Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer|   ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|   BBlockTransfer|  BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|    CBlockTransfer| CBlockTransfer|
-        //#################################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                             Weight|    Dim|   Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|    ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|    ThreadCluster|   ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|    ClusterLengths|ScalarPerVector|
-        //#################################################################################|       |        |        |        |   Operation|   Operation|   Operation|                     Specialization|Spatial|       |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|     ArrangeOrder|               |               |      PerVector|   PerVector_K1|          |  Lengths_K0_N_K1|    ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|  MBlock_MPerBlock|   NWaveNPerXdl|
-        //#################################################################################|       |        |        |        |            |            |            |                                   |       |       |      |      |      |   |     |     |     |     |                |                 |               |               |               |               |          |                 |                |               |              |               |               |          |            |            |  NBlock_NPerBlock|               |
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,      256,   256,   128,    4,  4,   32,   32,    4,    2,  S<1, 4, 64, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,    true,     S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,      true,        1,            1,     S<1, 32, 1, 8>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,      256,   128,   256,    4,  4,   32,   32,    2,    4,  S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,    true,     S<1, 4, 64, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,      true,        1,            1,     S<1, 32, 1, 8>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,      128,   128,   128,    4,  4,   32,   32,    4,    2,  S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,    true,     S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,      256,   128,   128,    4,  4,   32,   32,    2,    2,  S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,    true,     S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,      128,   128,    64,    4,  4,   32,   32,    2,    2,  S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,      128,    64,   128,    4,  4,   32,   32,    2,    2,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,    true,     S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,       64,    64,    64,    4,  4,   32,   32,    2,    2,  S<1, 4, 16, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,    true,     S<1, 4, 16, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,      true,        1,            1,     S<1, 16, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,      256,   128,    64,    4,  4,   32,   32,    2,    1,  S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              1,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,      256,    64,   128,    4,  4,   32,   32,    1,    2,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              1,    true,     S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,      128,   128,    32,    4,  4,   32,   32,    2,    1,  S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,    true,     S<1, 4, 8,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              1,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,      128,    32,   128,    4,  4,   32,   32,    1,    2,  S<1, 4, 8,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              1,    true,     S<1, 4, 32, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,      true,        1,            1,     S<1, 32, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,       64,    64,    32,    4,  4,   32,   32,    2,    1,  S<1, 4, 16, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,    true,     S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,      true,        1,            1,     S<1, 16, 1, 4>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<    F32,     F32,     F32,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     1,       64,    32,    64,    4,  4,   32,   32,    1,    2,  S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              2,    true,     S<1, 4, 16, 1>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              4,              4,      true,        1,            1,     S<1, 16, 1, 4>,        4>
-    // clang-format on
-    >;
-
-void add_device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_f32_instances(
-    std::vector<DeviceConvBwdWeightPtr<PassThrough, PassThrough, PassThrough>>& instances)
-{
-    add_device_operation_instances(
-        instances, device_conv1d_bwd_weight_xdl_c_shuffle_nwc_kxc_nwk_f32_instances{});
-    add_device_operation_instances(
-        instances, device_conv1d_bwd_weight_xdl_nwc_kxc_nwk_1x1_s1_p0_f32_instances{});
-}
-
-} // namespace instance
-} // namespace device
-} // namespace tensor_operation
-} // namespace ck

library/src/tensor_operation_instance/gpu/convnd_bwd_weight/device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_bf16_instance.cpp

-// SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
-
-#include <cstdlib>
-#include <numeric>
-#include "ck/ck.hpp"
-#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
-#include "ck/tensor_operation/gpu/device/device_convnd_backward_weight_xdl_c_shuffle_nhwc_kyxc_nhwk.hpp"
-#include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
-#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
-
-namespace ck {
-namespace tensor_operation {
-namespace device {
-namespace instance {
-
-using BF16 = bhalf_t;
-using F32  = float;
-
-template <ck::index_t... Is>
-using S = ck::Sequence<Is...>;
-
-using PassThrough = ck::tensor_operation::element_wise::PassThrough;
-static constexpr auto ConvBwdWeightDefault =
-    ck::tensor_operation::device::ConvolutionBackwardWeightSpecialization::Default;
-
-static constexpr auto ConvBwdWeightFilter1x1Stride1Pad0 =
-    ck::tensor_operation::device::ConvolutionBackwardWeightSpecialization::Filter1x1Stride1Pad0;
-
-// Compilation parameters for in[n, hi, wi, c] * wei[k, y, x, c] = out[n, ho, wo, k]
-using device_conv2d_bwd_weight_xdl_c_shuffle_nhwc_kyxc_nhwk_bf16_instances = std::tuple<
-    // clang-format off
-        //#################################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|          ConvBackward|    Num|  Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer|   ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|   BBlockTransfer|  BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|    CBlockTransfer| CBlockTransfer|
-        //#################################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                Weight|    Dim|   Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|    ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|    ThreadCluster|   ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|    ClusterLengths|ScalarPerVector|
-        //#################################################################################|       |        |        |        |   Operation|   Operation|   Operation|        Specialization|Spatial|       |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|     ArrangeOrder|               |               |      PerVector|   PerVector_K1|          |  Lengths_K0_N_K1|    ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|  MBlock_MPerBlock|   NWaveNPerXdl|
-        //#################################################################################|       |        |        |        |            |            |            |                      |       |       |      |      |      |   |     |     |     |     |                |                 |               |               |               |               |          |                 |                |               |              |               |               |          |            |            |  NBlock_NPerBlock|               |
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     2,      256,   256,   128,    4,  8,   32,   32,    4,    2,  S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 8>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     2,      256,   128,   256,    4,  8,   32,   32,    2,    4,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 8>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     2,      128,   128,   128,    4,  8,   32,   32,    4,    2,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     2,      256,   128,   128,    4,  8,   32,   32,    2,    2,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     2,      128,   128,    64,    4,  8,   32,   32,    2,    2,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 8,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     2,      128,    64,   128,    4,  8,   32,   32,    2,    2,  S<1, 4, 8,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     2,       64,    64,    64,    4,  8,   32,   32,    2,    2,  S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 16, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     2,      256,   128,    64,    4,  8,   32,   32,    2,    1,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 8,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     2,      256,    64,   128,    4,  8,   32,   32,    1,    2,  S<1, 4, 8,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     2,      128,   128,    32,    4,  8,   32,   32,    2,    1,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 4,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     2,      128,    32,   128,    4,  8,   32,   32,    1,    2,  S<1, 4, 4,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     2,       64,    64,    32,    4,  8,   32,   32,    2,    1,  S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 4,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 16, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightDefault,     2,       64,    32,    64,    4,  8,   32,   32,    1,    2,  S<1, 4, 4,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 16, 1, 4>,        4>
-    // clang-format on
-    >;
-
-using device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_1x1_s1_p0_bf16_instances = std::tuple<
-    // clang-format off
-        //#################################################################################| InData| WeiData| OutData| AccData|          In|         Wei|         Out|                       ConvBackward|    Num|  Block|  MPer|  NPer| K0Per| K1| MPer| NPer| MXdl| NXdl|  ABlockTransfer|   ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds|   BBlockTransfer|  BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds|    CShuffle|    CShuffle|    CBlockTransfer| CBlockTransfer|
-        //#################################################################################|   Type|    Type|    Type|    Type| Elementwise| Elementwise| Elementwise|                             Weight|    Dim|   Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|    ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|    ThreadCluster|   ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave|    ClusterLengths|ScalarPerVector|
-        //#################################################################################|       |        |        |        |   Operation|   Operation|   Operation|                     Specialization|Spatial|       |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|     ArrangeOrder|               |               |      PerVector|   PerVector_K1|          |  Lengths_K0_N_K1|    ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |  PerShuffle|  PerShuffle|  MBlock_MPerBlock|   NWaveNPerXdl|
-        //#################################################################################|       |        |        |        |            |            |            |                                   |       |       |      |      |      |   |     |     |     |     |                |                 |               |               |               |               |          |                 |                |               |              |               |               |          |            |            |  NBlock_NPerBlock|               |
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     2,      256,   256,   128,    4,  8,   32,   32,    4,    2,  S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 8>,        4>,
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     2,      256,   128,   256,    4,  8,   32,   32,    2,    4,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 32, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 8>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     2,      128,   128,   128,    4,  8,   32,   32,    4,    2,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     2,      256,   128,   128,    4,  8,   32,   32,    2,    2,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     2,      128,   128,    64,    4,  8,   32,   32,    2,    2,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 8,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     2,      128,    64,   128,    4,  8,   32,   32,    2,    2,  S<1, 4, 8,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     2,       64,    64,    64,    4,  8,   32,   32,    2,    2,  S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 16, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     2,      256,   128,    64,    4,  8,   32,   32,    2,    1,  S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 8,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     2,      256,    64,   128,    4,  8,   32,   32,    1,    2,  S<1, 4, 8,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,    true,     S<1, 4, 16, 4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     2,      128,   128,    32,    4,  8,   32,   32,    2,    1,  S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 4,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     2,      128,    32,   128,    4,  8,   32,   32,    1,    2,  S<1, 4, 4,  8>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              1,    true,     S<1, 4, 16, 2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 32, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     2,       64,    64,    32,    4,  8,   32,   32,    2,    1,  S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,    true,     S<1, 4, 4,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,      true,        1,            1,     S<1, 16, 1, 4>,        4>,   
-        DeviceConvndBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K<   BF16,    BF16,    BF16,     F32, PassThrough, PassThrough, PassThrough,  ConvBwdWeightFilter1x1Stride1Pad0,     2,       64,    32,    64,    4,  8,   32,   32,    1,    2,  S<1, 4, 4,  4>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              2,    true,     S<1, 4, 8,  2>,   S<0, 3, 1, 2>,   S<0, 2, 1, 3>,             2,              8,              4,      true,        1,            1,     S<1, 16, 1, 4>,        4>
-    // clang-format on
-    >;
-
-void add_device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_bf16_instances(
-    std::vector<DeviceConvBwdWeightPtr<PassThrough, PassThrough, PassThrough>>& instances)
-{
-    add_device_operation_instances(
-        instances, device_conv2d_bwd_weight_xdl_c_shuffle_nhwc_kyxc_nhwk_bf16_instances{});
-    add_device_operation_instances(
-        instances, device_conv2d_bwd_weight_xdl_nhwc_kyxc_nhwk_1x1_s1_p0_bf16_instances{});
-}
-
-} // namespace instance
-} // namespace device
-} // namespace tensor_operation
-} // namespace ck