Merge branch 'develop' into amd-develop

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

@@ -30,6 +30,76 @@ void add_device_grouped_conv2d_bwd_data_xdl_gnhwc_gkyxc_gnhwk_f16_instances(
                                                                   PassThrough,
                                                                   PassThrough>>>& instances);
 
+void add_device_grouped_conv2d_bwd_data_xdl_gnhwc_gkyxc_gnhwk_f32_instances(
+    std::vector<std::unique_ptr<DeviceGroupedConvBwdDataMultipleD<2,
+                                                                  GNHWK,
+                                                                  GKYXC,
+                                                                  Empty_Tuple,
+                                                                  GNHWC,
+                                                                  F32,
+                                                                  F32,
+                                                                  Empty_Tuple,
+                                                                  F32,
+                                                                  PassThrough,
+                                                                  PassThrough,
+                                                                  PassThrough>>>& instances);
+
+void add_device_grouped_conv2d_bwd_data_xdl_gnhwc_gkyxc_gnhwk_bf16_instances(
+    std::vector<std::unique_ptr<DeviceGroupedConvBwdDataMultipleD<2,
+                                                                  GNHWK,
+                                                                  GKYXC,
+                                                                  Empty_Tuple,
+                                                                  GNHWC,
+                                                                  BF16,
+                                                                  BF16,
+                                                                  Empty_Tuple,
+                                                                  BF16,
+                                                                  PassThrough,
+                                                                  PassThrough,
+                                                                  PassThrough>>>& instances);
+
+void add_device_grouped_conv2d_bwd_data_xdl_nhwgc_gkyxc_nhwgk_f16_instances(
+    std::vector<std::unique_ptr<DeviceGroupedConvBwdDataMultipleD<2,
+                                                                  NHWGK,
+                                                                  GKYXC,
+                                                                  Empty_Tuple,
+                                                                  NHWGC,
+                                                                  F16,
+                                                                  F16,
+                                                                  Empty_Tuple,
+                                                                  F16,
+                                                                  PassThrough,
+                                                                  PassThrough,
+                                                                  PassThrough>>>& instances);
+
+void add_device_grouped_conv2d_bwd_data_xdl_nhwgc_gkyxc_nhwgk_f32_instances(
+    std::vector<std::unique_ptr<DeviceGroupedConvBwdDataMultipleD<2,
+                                                                  NHWGK,
+                                                                  GKYXC,
+                                                                  Empty_Tuple,
+                                                                  NHWGC,
+                                                                  F32,
+                                                                  F32,
+                                                                  Empty_Tuple,
+                                                                  F32,
+                                                                  PassThrough,
+                                                                  PassThrough,
+                                                                  PassThrough>>>& instances);
+
+void add_device_grouped_conv2d_bwd_data_xdl_nhwgc_gkyxc_nhwgk_bf16_instances(
+    std::vector<std::unique_ptr<DeviceGroupedConvBwdDataMultipleD<2,
+                                                                  NHWGK,
+                                                                  GKYXC,
+                                                                  Empty_Tuple,
+                                                                  NHWGC,
+                                                                  BF16,
+                                                                  BF16,
+                                                                  Empty_Tuple,
+                                                                  BF16,
+                                                                  PassThrough,
+                                                                  PassThrough,
+                                                                  PassThrough>>>& instances);
+
 template <ck::index_t NumDimSpatial,
           typename OutLayout,
           typename WeiLayout,
@@ -78,6 +148,35 @@ struct DeviceOperationInstanceFactory<
             {
                 add_device_grouped_conv2d_bwd_data_xdl_gnhwc_gkyxc_gnhwk_f16_instances(op_ptrs);
             }
+            else if constexpr(is_same_v<InDataType, F32> && is_same_v<WeiDataType, F32> &&
+                              is_same_v<OutDataType, F32>)
+            {
+                add_device_grouped_conv2d_bwd_data_xdl_gnhwc_gkyxc_gnhwk_f32_instances(op_ptrs);
+            }
+            else if constexpr(is_same_v<InDataType, BF16> && is_same_v<WeiDataType, BF16> &&
+                              is_same_v<OutDataType, BF16>)
+            {
+                add_device_grouped_conv2d_bwd_data_xdl_gnhwc_gkyxc_gnhwk_bf16_instances(op_ptrs);
+            }
+        }
+        else if constexpr(NumDimSpatial == 2 && is_same_v<InLayout, NHWGC> &&
+                          is_same_v<WeiLayout, GKYXC> && is_same_v<OutLayout, NHWGK>)
+        {
+            if constexpr(is_same_v<InDataType, F16> && is_same_v<WeiDataType, F16> &&
+                         is_same_v<OutDataType, F16>)
+            {
+                add_device_grouped_conv2d_bwd_data_xdl_nhwgc_gkyxc_nhwgk_f16_instances(op_ptrs);
+            }
+            else if constexpr(is_same_v<InDataType, F32> && is_same_v<WeiDataType, F32> &&
+                              is_same_v<OutDataType, F32>)
+            {
+                add_device_grouped_conv2d_bwd_data_xdl_nhwgc_gkyxc_nhwgk_f32_instances(op_ptrs);
+            }
+            else if constexpr(is_same_v<InDataType, BF16> && is_same_v<WeiDataType, BF16> &&
+                              is_same_v<OutDataType, BF16>)
+            {
+                add_device_grouped_conv2d_bwd_data_xdl_nhwgc_gkyxc_nhwgk_bf16_instances(op_ptrs);
+            }
         }
 
         return op_ptrs;

library/include/ck/library/utility/host_tensor.hpp

@@ -13,6 +13,7 @@
 
 #include "ck/utility/data_type.hpp"
 #include "ck/utility/span.hpp"
+#include "ck/utility/type_convert.hpp"
 
 #include "ck/library/utility/algorithm.hpp"
 #include "ck/library/utility/ranges.hpp"

library/src/tensor_operation_instance/gpu/batched_gemm_multi_d/device_batched_gemm_multi_d_dl_f16_f16_f16_gkm_gkn_gmn_instance.cpp

 // SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+// Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
 
 #include <cstdlib>
 

library/src/tensor_operation_instance/gpu/batched_gemm_multi_d/device_batched_gemm_multi_d_dl_f16_f16_f16_gkm_gkn_gmn_irregular_instance.cpp

 // SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+// Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
 
 #include <cstdlib>
 

library/src/tensor_operation_instance/gpu/batched_gemm_multi_d/device_batched_gemm_multi_d_dl_f16_f16_f16_gkm_gnk_gmn_instance.cpp

 // SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+// Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
 
 #include <cstdlib>
 

library/src/tensor_operation_instance/gpu/batched_gemm_multi_d/device_batched_gemm_multi_d_dl_f16_f16_f16_gkm_gnk_gmn_irregular_instance.cpp

 // SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+// Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
 
 #include <cstdlib>
 

library/src/tensor_operation_instance/gpu/batched_gemm_multi_d/device_batched_gemm_multi_d_dl_f16_f16_f16_gmk_gkn_gmn_instance.cpp

 // SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+// Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
 
 #include <cstdlib>
 

library/src/tensor_operation_instance/gpu/batched_gemm_multi_d/device_batched_gemm_multi_d_dl_f16_f16_f16_gmk_gkn_gmn_irregular_instance.cpp

 // SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+// Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
 
 #include <cstdlib>
 

library/src/tensor_operation_instance/gpu/batched_gemm_multi_d/device_batched_gemm_multi_d_dl_f16_f16_f16_gmk_gnk_gmn_instance.cpp

 // SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+// Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
 
 #include <cstdlib>
 

library/src/tensor_operation_instance/gpu/batched_gemm_multi_d/device_batched_gemm_multi_d_dl_f16_f16_f16_gmk_gnk_gmn_irregular_instance.cpp

 // SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+// Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
 
 #include <cstdlib>
 

library/src/tensor_operation_instance/gpu/batched_gemm_multi_d/device_batched_gemm_multi_d_dl_i8_i8_i8_gkm_gkn_gmn_instance.cpp

 // SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+// Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
 
 #include <cstdlib>
 

library/src/tensor_operation_instance/gpu/batched_gemm_multi_d/device_batched_gemm_multi_d_dl_i8_i8_i8_gkm_gkn_gmn_irregular_instance.cpp

 // SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+// Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
 
 #include <cstdlib>
 

library/src/tensor_operation_instance/gpu/batched_gemm_multi_d/device_batched_gemm_multi_d_dl_i8_i8_i8_gkm_gnk_gmn_instance.cpp

 // SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+// Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
 
 #include <cstdlib>
 

library/src/tensor_operation_instance/gpu/batched_gemm_multi_d/device_batched_gemm_multi_d_dl_i8_i8_i8_gkm_gnk_gmn_irregular_instance.cpp

 // SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+// Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
 
 #include <cstdlib>
 

library/src/tensor_operation_instance/gpu/batched_gemm_multi_d/device_batched_gemm_multi_d_dl_i8_i8_i8_gmk_gkn_gmn_instance.cpp

 // SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+// Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
 
 #include <cstdlib>
 

library/src/tensor_operation_instance/gpu/batched_gemm_multi_d/device_batched_gemm_multi_d_dl_i8_i8_i8_gmk_gkn_gmn_irregular_instance.cpp

 // SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+// Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
 
 #include <cstdlib>
 

library/src/tensor_operation_instance/gpu/batched_gemm_multi_d/device_batched_gemm_multi_d_dl_i8_i8_i8_gmk_gnk_gmn_instance.cpp

 // SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+// Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
 
 #include <cstdlib>
 

library/src/tensor_operation_instance/gpu/batched_gemm_multi_d/device_batched_gemm_multi_d_dl_i8_i8_i8_gmk_gnk_gmn_irregular_instance.cpp

 // SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+// Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
 
 #include <cstdlib>
 

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

 add_instance_library(device_gemm_instance
-   device_gemm_xdl_f64_f64_f64_mk_kn_mn_instance.cpp
-   device_gemm_xdl_f64_f64_f64_mk_nk_mn_instance.cpp
-   device_gemm_xdl_f64_f64_f64_km_kn_mn_instance.cpp
-   device_gemm_xdl_f64_f64_f64_km_nk_mn_instance.cpp
-   device_gemm_xdl_f32_f32_f32_mk_kn_mn_instance.cpp
-   device_gemm_xdl_f32_f32_f32_mk_nk_mn_instance.cpp
-   device_gemm_xdl_f32_f32_f32_km_kn_mn_instance.cpp
-   device_gemm_xdl_f32_f32_f32_km_nk_mn_instance.cpp
-   device_gemm_xdl_f16_f16_f16_mk_kn_mn_instance.cpp
-   device_gemm_xdl_f16_f16_f16_mk_nk_mn_instance.cpp
-   device_gemm_xdl_f16_f16_f16_km_kn_mn_instance.cpp
-   device_gemm_xdl_f16_f16_f16_km_nk_mn_instance.cpp
-   device_gemm_xdl_c_shuffle_i8_i8_i8_mk_kn_mn_instance.cpp
-   device_gemm_xdl_c_shuffle_i8_i8_i8_mk_nk_mn_instance.cpp
-   device_gemm_xdl_c_shuffle_i8_i8_i8_km_kn_mn_instance.cpp
-   device_gemm_xdl_c_shuffle_i8_i8_i8_km_nk_mn_instance.cpp
-   device_gemm_xdl_c_shuffle_bf16_bf16_bf16_mk_kn_mn_instance.cpp
-   device_gemm_xdl_c_shuffle_bf16_bf16_bf16_mk_nk_mn_instance.cpp
-   device_gemm_xdl_c_shuffle_bf16_bf16_bf16_km_kn_mn_instance.cpp
-   device_gemm_xdl_c_shuffle_bf16_bf16_bf16_km_nk_mn_instance.cpp
-   device_gemm_xdl_c_shuffle_f16_f16_f16_mk_kn_mn_instance.cpp
-   device_gemm_xdl_c_shuffle_f16_f16_f16_mk_nk_mn_instance.cpp
-   device_gemm_xdl_c_shuffle_f16_f16_f16_km_kn_mn_instance.cpp
-   device_gemm_xdl_c_shuffle_f16_f16_f16_km_nk_mn_instance.cpp
-   device_gemm_xdl_c_shuffle_f32_f32_f32_mk_kn_mn_instance.cpp
-   device_gemm_xdl_c_shuffle_f32_f32_f32_mk_nk_mn_instance.cpp
-   device_gemm_xdl_c_shuffle_f32_f32_f32_km_kn_mn_instance.cpp
-   device_gemm_xdl_c_shuffle_f32_f32_f32_km_nk_mn_instance.cpp
-   device_gemm_xdl_c_shuffle_2_stage_f16_f16_f16_mk_nk_mn_instance.cpp
-   device_gemm_dl_f32_f32_f32_mk_kn_mn_instance.cpp
-   device_gemm_dl_f32_f32_f32_mk_nk_mn_instance.cpp
-   device_gemm_dl_f32_f32_f32_km_kn_mn_instance.cpp
-   device_gemm_dl_f32_f32_f32_km_nk_mn_instance.cpp
-   device_gemm_dl_f16_f16_f16_mk_kn_mn_instance.cpp
-   device_gemm_dl_f16_f16_f16_mk_kn_mn_irregular_instance.cpp
-   device_gemm_dl_f16_f16_f16_mk_nk_mn_instance.cpp
-   device_gemm_dl_f16_f16_f16_mk_nk_mn_irregular_instance.cpp
    device_gemm_dl_f16_f16_f16_km_kn_mn_instance.cpp
    device_gemm_dl_f16_f16_f16_km_kn_mn_irregular_instance.cpp
    device_gemm_dl_f16_f16_f16_km_nk_mn_instance.cpp
    device_gemm_dl_f16_f16_f16_km_nk_mn_irregular_instance.cpp
-   device_gemm_dl_i8_i8_i8_mk_kn_mn_instance.cpp
-   device_gemm_dl_i8_i8_i8_mk_kn_mn_irregular_instance.cpp
-   device_gemm_dl_i8_i8_i8_mk_nk_mn_instance.cpp
-   device_gemm_dl_i8_i8_i8_mk_nk_mn_irregular_instance.cpp
+   device_gemm_dl_f16_f16_f16_mk_kn_mn_instance.cpp
+   device_gemm_dl_f16_f16_f16_mk_kn_mn_irregular_instance.cpp
+   device_gemm_dl_f16_f16_f16_mk_nk_mn_instance.cpp
+   device_gemm_dl_f16_f16_f16_mk_nk_mn_irregular_instance.cpp
+   device_gemm_dl_f32_f32_f32_km_kn_mn_instance.cpp
+   device_gemm_dl_f32_f32_f32_km_nk_mn_instance.cpp
+   device_gemm_dl_f32_f32_f32_mk_kn_mn_instance.cpp
+   device_gemm_dl_f32_f32_f32_mk_nk_mn_instance.cpp
    device_gemm_dl_i8_i8_i8_km_kn_mn_instance.cpp
    device_gemm_dl_i8_i8_i8_km_kn_mn_irregular_instance.cpp
    device_gemm_dl_i8_i8_i8_km_nk_mn_instance.cpp
    device_gemm_dl_i8_i8_i8_km_nk_mn_irregular_instance.cpp
+   device_gemm_dl_i8_i8_i8_mk_kn_mn_instance.cpp
+   device_gemm_dl_i8_i8_i8_mk_kn_mn_irregular_instance.cpp
+   device_gemm_dl_i8_i8_i8_mk_nk_mn_instance.cpp
+   device_gemm_dl_i8_i8_i8_mk_nk_mn_irregular_instance.cpp
+   device_gemm_xdl_c_shuffle_2_stage_f16_f16_f16_mk_nk_mn_instance.cpp
+   device_gemm_xdl_c_shuffle_bf16_bf16_bf16_km_kn_mn_instance.cpp
+   device_gemm_xdl_c_shuffle_bf16_bf16_bf16_km_nk_mn_instance.cpp
+   device_gemm_xdl_c_shuffle_bf16_bf16_bf16_mk_kn_mn_instance.cpp
+   device_gemm_xdl_c_shuffle_bf16_bf16_bf16_mk_nk_mn_instance.cpp
+   device_gemm_xdl_c_shuffle_f16_f16_f16_km_kn_mn_instance.cpp
+   device_gemm_xdl_c_shuffle_f16_f16_f16_km_nk_mn_instance.cpp
+   device_gemm_xdl_c_shuffle_f16_f16_f16_mk_kn_mn_instance.cpp
+   device_gemm_xdl_c_shuffle_f16_f16_f16_mk_nk_mn_instance.cpp
+   device_gemm_xdl_c_shuffle_f32_f32_f32_km_kn_mn_instance.cpp
+   device_gemm_xdl_c_shuffle_f32_f32_f32_km_nk_mn_instance.cpp
+   device_gemm_xdl_c_shuffle_f32_f32_f32_mk_kn_mn_instance.cpp
+   device_gemm_xdl_c_shuffle_f32_f32_f32_mk_nk_mn_instance.cpp
+   device_gemm_xdl_c_shuffle_i8_i8_i8_km_kn_mn_instance.cpp
+   device_gemm_xdl_c_shuffle_i8_i8_i8_km_nk_mn_instance.cpp
+   device_gemm_xdl_c_shuffle_i8_i8_i8_mk_kn_mn_instance.cpp
+   device_gemm_xdl_c_shuffle_i8_i8_i8_mk_nk_mn_instance.cpp
+
+   device_gemm_xdl_f16_f16_f16/km_kn_mn_add_instance.cpp
+   device_gemm_xdl_f16_f16_f16/km_kn_mn_default_pipeline_v1_instance.cpp
+   device_gemm_xdl_f16_f16_f16/km_kn_mn_default_pipeline_v2_instance.cpp
+   device_gemm_xdl_f16_f16_f16/km_kn_mn_default_pipeline_v2_opt_instance.cpp
+   device_gemm_xdl_f16_f16_f16/km_kn_mn_interwave_pipeline_v1_instance.cpp
+   device_gemm_xdl_f16_f16_f16/km_kn_mn_irregular_default_pipeline_v1_instance.cpp
+   device_gemm_xdl_f16_f16_f16/km_kn_mn_irregular_default_pipeline_v2_instance.cpp
+   device_gemm_xdl_f16_f16_f16/km_kn_mn_irregular_interwave_pipeline_v1_instance.cpp
+   device_gemm_xdl_f16_f16_f16/km_nk_mn_add_instance.cpp
+   device_gemm_xdl_f16_f16_f16/km_nk_mn_default_pipeline_v1_instance.cpp
+   device_gemm_xdl_f16_f16_f16/km_nk_mn_default_pipeline_v2_instance.cpp
+   device_gemm_xdl_f16_f16_f16/km_nk_mn_default_pipeline_v2_opt_instance.cpp
+   device_gemm_xdl_f16_f16_f16/km_nk_mn_interwave_pipeline_v1_instance.cpp
+   device_gemm_xdl_f16_f16_f16/km_nk_mn_irregular_default_pipeline_v1_instance.cpp
+   device_gemm_xdl_f16_f16_f16/km_nk_mn_irregular_default_pipeline_v2_instance.cpp
+   device_gemm_xdl_f16_f16_f16/km_nk_mn_irregular_interwave_pipeline_v1_instance.cpp
+   device_gemm_xdl_f16_f16_f16/mk_kn_mn_add_instance.cpp
+   device_gemm_xdl_f16_f16_f16/mk_kn_mn_default_pipeline_v1_instance.cpp
+   device_gemm_xdl_f16_f16_f16/mk_kn_mn_default_pipeline_v2_instance.cpp
+   device_gemm_xdl_f16_f16_f16/mk_kn_mn_default_pipeline_v2_opt_instance.cpp
+   device_gemm_xdl_f16_f16_f16/mk_kn_mn_interwave_pipeline_v1_instance.cpp
+   device_gemm_xdl_f16_f16_f16/mk_kn_mn_irregular_default_pipeline_v1_instance.cpp
+   device_gemm_xdl_f16_f16_f16/mk_kn_mn_irregular_default_pipeline_v2_instance.cpp
+   device_gemm_xdl_f16_f16_f16/mk_kn_mn_irregular_interwave_pipeline_v1_instance.cpp
+   device_gemm_xdl_f16_f16_f16/mk_nk_mn_add_instance.cpp
+   device_gemm_xdl_f16_f16_f16/mk_nk_mn_default_pipeline_v1_instance.cpp
+   device_gemm_xdl_f16_f16_f16/mk_nk_mn_default_pipeline_v2_instance.cpp
+   device_gemm_xdl_f16_f16_f16/mk_nk_mn_default_pipeline_v2_opt_instance.cpp
+   device_gemm_xdl_f16_f16_f16/mk_nk_mn_interwave_pipeline_v1_instance.cpp
+   device_gemm_xdl_f16_f16_f16/mk_nk_mn_irregular_default_pipeline_v1_instance.cpp
+   device_gemm_xdl_f16_f16_f16/mk_nk_mn_irregular_default_pipeline_v2_instance.cpp
+   device_gemm_xdl_f16_f16_f16/mk_nk_mn_irregular_interwave_pipeline_v1_instance.cpp
+
+   device_gemm_xdl_f32_f32_f32_km_kn_mn_instance.cpp
+   device_gemm_xdl_f32_f32_f32_km_nk_mn_instance.cpp
+   device_gemm_xdl_f32_f32_f32_mk_kn_mn_instance.cpp
+   device_gemm_xdl_f32_f32_f32_mk_nk_mn_instance.cpp
+   device_gemm_xdl_f64_f64_f64_km_kn_mn_instance.cpp
+   device_gemm_xdl_f64_f64_f64_km_nk_mn_instance.cpp
+   device_gemm_xdl_f64_f64_f64_mk_kn_mn_instance.cpp
+   device_gemm_xdl_f64_f64_f64_mk_nk_mn_instance.cpp
+)
+
+set(ENABLE_PIPELINE_V2_OPT OFF)
+
+if (ENABLE_PIPELINE_V2_OPT)
+set(MAX_ILP_OPTS
+   -mllvm
+   -amdgpu-enable-max-ilp-scheduling-strategy
+)
+set(WAVES_PER_EU_DEFS
+   CK_USE_WAVES_PER_EU=1
+   CK_MIN_WAVES_PER_EU=1
+   CK_MAX_WAVES_PER_EU=1
+)
+set(IGLP_OPT_DEFS
+   CK_EXPERIMENTAL_PIPELINE_V2_IGLP_OPT=1
 )
+
+# layout=NT
+set_source_files_properties(device_gemm_xdl_f16_f16_f16/km_kn_mn_default_pipeline_v2_opt_instance.cpp PROPERTIES
+   COMPILE_OPTIONS ";;"
+   COMPILE_DEFINITIONS "${WAVES_PER_EU_DEFS};${IGLP_OPT_DEFS}")
+# layout=NN
+set_source_files_properties(device_gemm_xdl_f16_f16_f16/km_nk_mn_default_pipeline_v2_opt_instance.cpp PROPERTIES
+   COMPILE_OPTIONS "${MAX_ILP_OPTS}"
+   COMPILE_DEFINITIONS "${WAVES_PER_EU_DEFS};${IGLP_OPT_DEFS}")
+# layout=TT
+set_source_files_properties(device_gemm_xdl_f16_f16_f16/mk_kn_mn_default_pipeline_v2_opt_instance.cpp PROPERTIES
+   COMPILE_OPTIONS "${MAX_ILP_OPTS}"
+   COMPILE_DEFINITIONS "${WAVES_PER_EU_DEFS}")
+# layout=TN
+set_source_files_properties(device_gemm_xdl_f16_f16_f16/mk_nk_mn_default_pipeline_v2_opt_instance.cpp PROPERTIES
+   COMPILE_OPTIONS "${MAX_ILP_OPTS}"
+   COMPILE_DEFINITIONS "${WAVES_PER_EU_DEFS};${IGLP_OPT_DEFS}")
+endif(ENABLE_PIPELINE_V2_OPT)

library/src/tensor_operation_instance/gpu/gemm/device_gemm_xdl_f16_f16_f16/common.hpp

+// SPDX-License-Identifier: MIT
+// Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
+
+#pragma once
+
+#include "ck/ck.hpp"
+#include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
+#include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
+#include "ck/tensor_operation/gpu/device/impl/device_gemm_xdl.hpp"
+#include "ck/tensor_operation/gpu/device/tensor_layout.hpp"
+#include "ck/utility/data_type.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace instance {
+
+using F16 = ck::half_t;
+using F32 = float;
+
+using Row = ck::tensor_layout::gemm::RowMajor;
+using Col = ck::tensor_layout::gemm::ColumnMajor;
+
+template <ck::index_t... Is>
+using S = ck::Sequence<Is...>;
+
+using PassThrough = ck::tensor_operation::element_wise::PassThrough;
+
+static constexpr auto GemmDefault   = ck::tensor_operation::device::GemmSpecialization::Default;
+static constexpr auto GemmMNPadding = ck::tensor_operation::device::GemmSpecialization::MNPadding;
+
+using InstanceNT = DeviceGemm<Col, Row, Row, F16, F16, F16, PassThrough, PassThrough, PassThrough>;
+using InstanceNN = DeviceGemm<Col, Col, Row, F16, F16, F16, PassThrough, PassThrough, PassThrough>;
+using InstanceTT = DeviceGemm<Row, Row, Row, F16, F16, F16, PassThrough, PassThrough, PassThrough>;
+using InstanceTN = DeviceGemm<Row, Col, Row, F16, F16, F16, PassThrough, PassThrough, PassThrough>;
+
+template <typename Instance>
+using OwnerList = std::vector<std::unique_ptr<Instance>>;
+
+} // namespace instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck