Fix formatting

example/11_conv2d_bwd_weight/conv2d_bwd_weight_xdl.cpp

@@ -72,8 +72,13 @@ using DeviceConvBwdWeightInstance = ck::tensor_operation::device::
         8>;                               // CBlockTransferScalarPerVector_NWaveNPerXdl
 // clang-format on
 
-using ReferenceConvBwdWeightInstance = ck::tensor_operation::host::
-    ReferenceConvBwdWeight<InDataType, WeiDataType, OutDataType, InElementOp, WeiElementOp, OutElementOp>;
+using ReferenceConvBwdWeightInstance =
+    ck::tensor_operation::host::ReferenceConvBwdWeight<InDataType,
+                                                       WeiDataType,
+                                                       OutDataType,
+                                                       InElementOp,
+                                                       WeiElementOp,
+                                                       OutElementOp>;
 
 int main(int argc, char* argv[])
 {

include/ck/tensor_operation/gpu/device/device_base.hpp

@@ -22,7 +22,10 @@ struct BaseInvoker
     BaseInvoker(const BaseInvoker&) = default;
     BaseInvoker& operator=(const BaseInvoker&) = default;
 
-    virtual float Run(const BaseArgument*, int = 1, hipStream_t = nullptr, bool = false){return -1;}
+    virtual float Run(const BaseArgument*, int = 1, hipStream_t = nullptr, bool = false)
+    {
+        return -1;
+    }
 
     virtual ~BaseInvoker() {}
 };
@@ -33,8 +36,8 @@ struct BaseOperator
     BaseOperator(const BaseOperator&) = default;
     BaseOperator& operator=(const BaseOperator&) = default;
 
-    virtual bool IsSupportedArgument(const BaseArgument*){return false;}
-    virtual std::string GetTypeString() const            {return "";}
+    virtual bool IsSupportedArgument(const BaseArgument*) { return false; }
+    virtual std::string GetTypeString() const { return ""; }
 
     virtual ~BaseOperator() {}
 };

include/ck/tensor_operation/gpu/device/device_batched_gemm_reduce_xdl_cshuffle.hpp

@@ -693,7 +693,10 @@ struct DeviceBatchedGemmReduce_Xdl_CShuffle : public DeviceGemmReduce<AElementwi
     {
         using Argument = DeviceOp::Argument;
 
-        float Run(const Argument& arg, int  nrepeat  = 1, hipStream_t stream_id = nullptr, bool measure_time = false)
+        float Run(const Argument& arg,
+                  int nrepeat           = 1,
+                  hipStream_t stream_id = nullptr,
+                  bool measure_time     = false)
         {
 #if 0
             {
@@ -752,7 +755,9 @@ struct DeviceBatchedGemmReduce_Xdl_CShuffle : public DeviceGemmReduce<AElementwi
                     remove_reference_t<Block2CTileMap>,
                     true>;
 
-                elapsed_time = launch_and_time_kernel(kernel,nrepeat,
+                elapsed_time =
+                    launch_and_time_kernel(kernel,
+                                           nrepeat,
                                            dim3(grid_size),
                                            dim3(BlockSize),
                                            0,
@@ -796,7 +801,8 @@ struct DeviceBatchedGemmReduce_Xdl_CShuffle : public DeviceGemmReduce<AElementwi
                     remove_reference_t<Block2CTileMap>,
                     false>;
 
-                elapsed_time = launch_and_time_kernel(kernel,
+                elapsed_time =
+                    launch_and_time_kernel(kernel,
                                            nrepeat,
                                            dim3(grid_size),
                                            dim3(BlockSize),
@@ -826,7 +832,10 @@ struct DeviceBatchedGemmReduce_Xdl_CShuffle : public DeviceGemmReduce<AElementwi
         }
 
         // polymorphic
-        float Run(const BaseArgument* p_arg, int nrepeat = 1, hipStream_t stream_id = nullptr, bool measure_time = false) override
+        float Run(const BaseArgument* p_arg,
+                  int nrepeat           = 1,
+                  hipStream_t stream_id = nullptr,
+                  bool measure_time     = false) override
         {
             return Run(*dynamic_cast<const Argument*>(p_arg), nrepeat, stream_id, measure_time);
         }

include/ck/tensor_operation/gpu/device/device_batched_gemm_xdl.hpp

@@ -402,7 +402,10 @@ struct DeviceBatchedGemmXdl
     {
         using Argument = DeviceBatchedGemmXdl::Argument;
 
-        float Run(const Argument& arg, int nrepeat = 1, hipStream_t stream_id = nullptr, bool measure_time = false)
+        float Run(const Argument& arg,
+                  int nrepeat           = 1,
+                  hipStream_t stream_id = nullptr,
+                  bool measure_time     = false)
         {
             {
                 std::cout << "arg.a_grid_desc_k0_m_k1_{" << arg.a_grid_desc_k0_m_k1_.GetLength(I0)
@@ -513,7 +516,10 @@ struct DeviceBatchedGemmXdl
         }
 
         // polymorphic
-        float Run(const BaseArgument* p_arg, int nrepeat = 1, hipStream_t stream_id = nullptr, bool measure_time = false) override
+        float Run(const BaseArgument* p_arg,
+                  int nrepeat           = 1,
+                  hipStream_t stream_id = nullptr,
+                  bool measure_time     = false) override
         {
             return Run(*dynamic_cast<const Argument*>(p_arg), nrepeat, stream_id, measure_time);
         }

include/ck/tensor_operation/gpu/device/device_conv2d_backward_weight_xdl_c_shuffle_nhwc_kyxc_nhwk.hpp

@@ -415,7 +415,10 @@ struct DeviceConv2dBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_
                       << arg.c_grid_desc_m_n_.GetLength(I1) << "}" << std::endl;
         }
 
-        float Run(const Argument& arg, int nrepeat = 1, hipStream_t stream_id = nullptr, bool measure_time = false)
+        float Run(const Argument& arg,
+                  int nrepeat           = 1,
+                  hipStream_t stream_id = nullptr,
+                  bool measure_time     = false)
         {
             ShowInfo(arg);
             if(!GridwiseGemm::CheckValidity(arg.a_grid_desc_kbatch_k0_m_k1_,
@@ -563,7 +566,10 @@ struct DeviceConv2dBwdWeightXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_
             return ave_time;
         }
 
-        float Run(const BaseArgument* p_arg, int nrepeat = 1, hipStream_t stream_id = nullptr, bool measure_time = false) override
+        float Run(const BaseArgument* p_arg,
+                  int nrepeat           = 1,
+                  hipStream_t stream_id = nullptr,
+                  bool measure_time     = false) override
         {
             return Run(*dynamic_cast<const Argument*>(p_arg), nrepeat, stream_id, measure_time);
         }

include/ck/tensor_operation/gpu/device/device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk.hpp

@@ -531,7 +531,10 @@ struct DeviceConv2dBwdDataXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K
     {
         using Argument = DeviceOp::Argument;
 
-        float Run(const Argument& arg, int nrepeat = 1, hipStream_t stream_id = nullptr, bool measure_time = false)
+        float Run(const Argument& arg,
+                  int nrepeat           = 1,
+                  hipStream_t stream_id = nullptr,
+                  bool measure_time     = false)
         {
             float ave_time = 0;
             for(size_t i = 0; i < arg.a_grid_desc_k0_m_k1_container_.size(); i++)
@@ -660,7 +663,10 @@ struct DeviceConv2dBwdDataXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K
             return ave_time;
         }
 
-        float Run(const BaseArgument* p_arg, int nrepeat = 1, hipStream_t stream_id = nullptr, bool measure_time = false) override
+        float Run(const BaseArgument* p_arg,
+                  int nrepeat           = 1,
+                  hipStream_t stream_id = nullptr,
+                  bool measure_time     = false) override
         {
             return Run(*dynamic_cast<const Argument*>(p_arg), nrepeat, stream_id, measure_time);
         }

include/ck/tensor_operation/gpu/device/device_conv2d_fwd_xdl_c_shuffle_bias_activation_add_nhwc_kyxc_nhwk.hpp

@@ -642,7 +642,10 @@ struct
     {
         using Argument = DeviceOp::Argument;
 
-        float Run(const Argument& arg, int nrepeat = 1, hipStream_t stream_id = nullptr, bool measure_time = false)
+        float Run(const Argument& arg,
+                  int nrepeat           = 1,
+                  hipStream_t stream_id = nullptr,
+                  bool measure_time     = false)
         {
 #if 0
             {
@@ -800,7 +803,10 @@ struct
             return ave_time;
         }
 
-        float Run(const BaseArgument* p_arg, int nrepeat = 1, hipStream_t stream_id = nullptr, bool measure_time = false) override
+        float Run(const BaseArgument* p_arg,
+                  int nrepeat           = 1,
+                  hipStream_t stream_id = nullptr,
+                  bool measure_time     = false) override
         {
             return Run(*dynamic_cast<const Argument*>(p_arg), nrepeat, stream_id, measure_time);
         }

include/ck/tensor_operation/gpu/device/device_conv2d_fwd_xdl_c_shuffle_bias_activation_nhwc_kyxc_nhwk.hpp

@@ -607,7 +607,10 @@ struct DeviceConv2dFwdXdl_C_Shuffle_Bias_Activation_Input_N_Hi_Wi_C_Weight_K_Y_X
     {
         using Argument = DeviceOp::Argument;
 
-        float Run(const Argument& arg, int nrepeat = 1, hipStream_t stream_id = nullptr, bool measure_time = false)
+        float Run(const Argument& arg,
+                  int nrepeat           = 1,
+                  hipStream_t stream_id = nullptr,
+                  bool measure_time     = false)
         {
 #if 0
             {
@@ -752,7 +755,10 @@ struct DeviceConv2dFwdXdl_C_Shuffle_Bias_Activation_Input_N_Hi_Wi_C_Weight_K_Y_X
             return ave_time;
         }
 
-        float Run(const BaseArgument* p_arg, int nrepeat = 1, hipStream_t stream_id = nullptr, bool measure_time = false) override
+        float Run(const BaseArgument* p_arg,
+                  int nrepeat           = 1,
+                  hipStream_t stream_id = nullptr,
+                  bool measure_time     = false) override
         {
             return Run(*dynamic_cast<const Argument*>(p_arg), nrepeat, stream_id, measure_time);
         }

include/ck/tensor_operation/gpu/device/device_conv2d_fwd_xdl_c_shuffle_nhwc_kyxc_nhwk.hpp

@@ -568,7 +568,10 @@ struct DeviceConv2dFwdXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_W
     {
         using Argument = DeviceOp::Argument;
 
-        float Run(const Argument& arg, int nrepeat = 1, hipStream_t stream_id = nullptr, bool measure_time = false)
+        float Run(const Argument& arg,
+                  int nrepeat           = 1,
+                  hipStream_t stream_id = nullptr,
+                  bool measure_time     = false)
         {
 #if 0
             {
@@ -722,7 +725,10 @@ struct DeviceConv2dFwdXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_W
             return ave_time;
         }
 
-        float Run(const BaseArgument* p_arg, int nrepeat = 1, hipStream_t stream_id = nullptr, bool measure_time = false) override
+        float Run(const BaseArgument* p_arg,
+                  int nrepeat           = 1,
+                  hipStream_t stream_id = nullptr,
+                  bool measure_time     = false) override
         {
             return Run(*dynamic_cast<const Argument*>(p_arg), nrepeat, stream_id, measure_time);
         }

include/ck/tensor_operation/gpu/device/device_conv2d_fwd_xdl_nhwc_kyxc_nhwk.hpp

@@ -450,7 +450,10 @@ struct DeviceConv2dFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K
     {
         using Argument = DeviceOp::Argument;
 
-        float Run(const Argument& arg, int nrepeat = 1, hipStream_t stream_id = nullptr, bool measure_time = false)
+        float Run(const Argument& arg,
+                  int nrepeat           = 1,
+                  hipStream_t stream_id = nullptr,
+                  bool measure_time     = false)
         {
 #if 0
             {
@@ -554,7 +557,10 @@ struct DeviceConv2dFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K
             return ave_time;
         }
 
-        float Run(const BaseArgument* p_arg, int nrepeat = 1, hipStream_t stream_id = nullptr, bool measure_time = false) override
+        float Run(const BaseArgument* p_arg,
+                  int nrepeat           = 1,
+                  hipStream_t stream_id = nullptr,
+                  bool measure_time     = false) override
         {
             return Run(*dynamic_cast<const Argument*>(p_arg), nrepeat, stream_id, measure_time);
         }

include/ck/tensor_operation/gpu/device/device_conv3d_fwd_naive_ndhwc_kzyxc_ndhwk.hpp

@@ -92,7 +92,10 @@ struct DeviceConv3dFwdNaive_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_W
     {
         using Argument = DeviceOp::Argument;
 
-        float Run(const Argument& arg, int nrepeat = 1, hipStream_t stream_id = nullptr, bool measure_time = false)
+        float Run(const Argument& arg,
+                  int nrepeat           = 1,
+                  hipStream_t stream_id = nullptr,
+                  bool measure_time     = false)
         {
             const auto naive_conv3d_fwd =
                 ref::naive_conv_fwd_ndhwc_kzyxc_ndhwk<InDataType,
@@ -139,7 +142,10 @@ struct DeviceConv3dFwdNaive_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_W
         }
 
         // polymorphic
-        float Run(const BaseArgument* p_arg, int nrepeat = 1, hipStream_t stream_id = nullptr, bool measure_time = false) override
+        float Run(const BaseArgument* p_arg,
+                  int nrepeat           = 1,
+                  hipStream_t stream_id = nullptr,
+                  bool measure_time     = false) override
         {
             return Run(*dynamic_cast<const Argument*>(p_arg), nrepeat, stream_id, measure_time);
         }

include/ck/tensor_operation/gpu/device/device_conv3d_fwd_xdl_ndhwc_kzyxc_ndhwk.hpp

@@ -435,7 +435,10 @@ struct DeviceConv3dFwdXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_
     {
         using Argument = DeviceOp::Argument;
 
-        float Run(const Argument& arg, int nrepeat = 1, hipStream_t stream_id = nullptr, bool measure_time = false)
+        float Run(const Argument& arg,
+                  int nrepeat           = 1,
+                  hipStream_t stream_id = nullptr,
+                  bool measure_time     = false)
         {
             {
                 std::cout << "num_batches_of_GEMM = " << arg.num_subbatches_ << std::endl;
@@ -548,7 +551,10 @@ struct DeviceConv3dFwdXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_
         }
 
         // polymorphic
-        float Run(const BaseArgument* p_arg, int nrepeat = 1, hipStream_t stream_id = nullptr, bool measure_time = false) override
+        float Run(const BaseArgument* p_arg,
+                  int nrepeat           = 1,
+                  hipStream_t stream_id = nullptr,
+                  bool measure_time     = false) override
         {
             return Run(*dynamic_cast<const Argument*>(p_arg), nrepeat, stream_id, measure_time);
         }

include/ck/tensor_operation/gpu/device/device_convnd_bwd_data_xdl_ndhwc_kzyxc_ndhwk.hpp

@@ -1241,7 +1241,10 @@ struct DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho
     {
         using Argument = DeviceOp::Argument;
 
-        float Run(const Argument& arg, int nrepeat = 1, hipStream_t stream_id = nullptr, bool measure_time = false)
+        float Run(const Argument& arg,
+                  int nrepeat           = 1,
+                  hipStream_t stream_id = nullptr,
+                  bool measure_time     = false)
         {
             float ave_time = 0;
             for(size_t i = 0; i < arg.a_grid_desc_k0_m_k1_container_.size(); i++)
@@ -1374,7 +1377,10 @@ struct DeviceConvndBwdDataXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho
             return ave_time;
         }
 
-        float Run(const BaseArgument* p_arg, int nrepeat = 1, hipStream_t stream_id = nullptr, bool measure_time = false) override
+        float Run(const BaseArgument* p_arg,
+                  int nrepeat           = 1,
+                  hipStream_t stream_id = nullptr,
+                  bool measure_time     = false) override
         {
             return Run(*dynamic_cast<const Argument*>(p_arg), nrepeat, stream_id, measure_time);
         }

include/ck/tensor_operation/gpu/device/device_convnd_fwd_xdl_nhwc_kyxc_nhwk.hpp

@@ -747,7 +747,10 @@ struct DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K
     {
         using Argument = DeviceOp::Argument;
 
-        float Run(const Argument& arg, int nrepeat = 1, hipStream_t stream_id = nullptr, bool measure_time = false)
+        float Run(const Argument& arg,
+                  int nrepeat           = 1,
+                  hipStream_t stream_id = nullptr,
+                  bool measure_time     = false)
         {
 #if 0
             {
@@ -851,7 +854,10 @@ struct DeviceConvNDFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K
             return ave_time;
         }
 
-        float Run(const BaseArgument* p_arg, int nrepeat = 1, hipStream_t stream_id = nullptr, bool measure_time = false) override
+        float Run(const BaseArgument* p_arg,
+                  int nrepeat           = 1,
+                  hipStream_t stream_id = nullptr,
+                  bool measure_time     = false) override
         {
             return Run(*dynamic_cast<const Argument*>(p_arg), nrepeat, stream_id, measure_time);
         }

include/ck/tensor_operation/gpu/device/device_gemm_reduce_xdl_cshuffle.hpp

@@ -500,7 +500,10 @@ struct DeviceGemmReduce_Xdl_CShuffle : public DeviceGemmReduce<AElementwiseOpera
     {
         using Argument = DeviceOp::Argument;
 
-        float Run(const Argument& arg, int  nrepeat  = 1, hipStream_t stream_id = nullptr, bool measure_time = false)
+        float Run(const Argument& arg,
+                  int nrepeat           = 1,
+                  hipStream_t stream_id = nullptr,
+                  bool measure_time     = false)
         {
 #if 0
             {
@@ -554,7 +557,8 @@ struct DeviceGemmReduce_Xdl_CShuffle : public DeviceGemmReduce<AElementwiseOpera
                     typename GridwiseGemm::DefaultBlock2CTileMap,
                     true>;
 
-                elapsed_time = launch_and_time_kernel(kernel,
+                elapsed_time =
+                    launch_and_time_kernel(kernel,
                                            nrepeat,
                                            dim3(grid_size),
                                            dim3(BlockSize),
@@ -596,7 +600,8 @@ struct DeviceGemmReduce_Xdl_CShuffle : public DeviceGemmReduce<AElementwiseOpera
                     typename GridwiseGemm::DefaultBlock2CTileMap,
                     false>;
 
-                elapsed_time = launch_and_time_kernel(kernel,
+                elapsed_time =
+                    launch_and_time_kernel(kernel,
                                            nrepeat,
                                            dim3(grid_size),
                                            dim3(BlockSize),
@@ -624,7 +629,10 @@ struct DeviceGemmReduce_Xdl_CShuffle : public DeviceGemmReduce<AElementwiseOpera
         }
 
         // polymorphic
-        float Run(const BaseArgument* p_arg, int nrepeat = 1, hipStream_t stream_id = nullptr, bool measure_time = false) override
+        float Run(const BaseArgument* p_arg,
+                  int nrepeat           = 1,
+                  hipStream_t stream_id = nullptr,
+                  bool measure_time     = false) override
         {
             return Run(*dynamic_cast<const Argument*>(p_arg), nrepeat, stream_id, measure_time);
         }

include/ck/tensor_operation/gpu/device/device_gemm_xdl.hpp

@@ -290,7 +290,10 @@ struct DeviceGemmXdl
     {
         using Argument = DeviceGemmXdl::Argument;
 
-        float Run(const Argument& arg, int nrepeat = 1, hipStream_t stream_id = nullptr, bool measure_time = false)
+        float Run(const Argument& arg,
+                  int nrepeat           = 1,
+                  hipStream_t stream_id = nullptr,
+                  bool measure_time     = false)
         {
             {
                 std::cout << "arg.a_grid_desc_k0_m_k1_{" << arg.a_grid_desc_k0_m_k1_.GetLength(I0)
@@ -394,7 +397,10 @@ struct DeviceGemmXdl
         }
 
         // polymorphic
-        float Run(const BaseArgument* p_arg, int nrepeat = 1, hipStream_t stream_id = nullptr, bool measure_time = false) override
+        float Run(const BaseArgument* p_arg,
+                  int nrepeat           = 1,
+                  hipStream_t stream_id = nullptr,
+                  bool measure_time     = false) override
         {
             return Run(*dynamic_cast<const Argument*>(p_arg), nrepeat, stream_id, measure_time);
         }

include/ck/tensor_operation/gpu/device/device_gemm_xdl_c_shuffle.hpp

@@ -249,7 +249,10 @@ struct DeviceGemmXdl_C_Shuffle
     {
         using Argument = DeviceGemmXdl_C_Shuffle::Argument;
 
-        float Run(const Argument& arg, int nrepeat = 1, hipStream_t stream_id = nullptr, bool measure_time = false)
+        float Run(const Argument& arg,
+                  int nrepeat           = 1,
+                  hipStream_t stream_id = nullptr,
+                  bool measure_time     = false)
         {
             {
                 std::cout << "arg.a_grid_desc_k0_m_k1_{" << arg.a_grid_desc_k0_m_k1_.GetLength(I0)
@@ -359,7 +362,10 @@ struct DeviceGemmXdl_C_Shuffle
         }
 
         // polymorphic
-        float Run(const BaseArgument* p_arg, int nrepeat = 1, hipStream_t stream_id = nullptr, bool measure_time = false) override
+        float Run(const BaseArgument* p_arg,
+                  int nrepeat           = 1,
+                  hipStream_t stream_id = nullptr,
+                  bool measure_time     = false) override
         {
             return Run(*dynamic_cast<const Argument*>(p_arg), nrepeat, stream_id, measure_time);
         }

include/ck/tensor_operation/gpu/device/device_gemm_xdl_c_shuffle_bias_2d.hpp

@@ -266,7 +266,10 @@ struct DeviceGemmXdl_C_Shuffle_Bias_2d
     {
         using Argument = DeviceGemmXdl_C_Shuffle_Bias_2d::Argument;
 
-        float Run(const Argument& arg, int nrepeat = 1, hipStream_t stream_id = nullptr, bool measure_time = false)
+        float Run(const Argument& arg,
+                  int nrepeat           = 1,
+                  hipStream_t stream_id = nullptr,
+                  bool measure_time     = false)
         {
             {
                 std::cout << "arg.a_grid_desc_k0_m_k1_{" << arg.a_grid_desc_k0_m_k1_.GetLength(I0)
@@ -389,7 +392,10 @@ struct DeviceGemmXdl_C_Shuffle_Bias_2d
         }
 
         // polymorphic
-        float Run(const BaseArgument* p_arg, int nrepeat = 1, hipStream_t stream_id = nullptr, bool measure_time = false) override
+        float Run(const BaseArgument* p_arg,
+                  int nrepeat           = 1,
+                  hipStream_t stream_id = nullptr,
+                  bool measure_time     = false) override
         {
             return Run(*dynamic_cast<const Argument*>(p_arg), nrepeat, stream_id, measure_time);
         }

include/ck/tensor_operation/gpu/device/device_gemm_xdl_c_shuffle_bias_activation.hpp

@@ -273,7 +273,10 @@ struct DeviceGemmXdl_C_Shuffle_Bias_Activation
     {
         using Argument = DeviceOp::Argument;
 
-        float Run(const Argument& arg, int nrepeat = 1, hipStream_t stream_id = nullptr, bool measure_time = false)
+        float Run(const Argument& arg,
+                  int nrepeat           = 1,
+                  hipStream_t stream_id = nullptr,
+                  bool measure_time     = false)
         {
             {
                 std::cout << "arg.a_grid_desc_k0_m_k1_{" << arg.a_grid_desc_k0_m_k1_.GetLength(I0)
@@ -396,7 +399,10 @@ struct DeviceGemmXdl_C_Shuffle_Bias_Activation
         }
 
         // polymorphic
-        float Run(const BaseArgument* p_arg, int nrepeat = 1, hipStream_t stream_id = nullptr, bool measure_time = false) override
+        float Run(const BaseArgument* p_arg,
+                  int nrepeat           = 1,
+                  hipStream_t stream_id = nullptr,
+                  bool measure_time     = false) override
         {
             return Run(*dynamic_cast<const Argument*>(p_arg), nrepeat, stream_id, measure_time);
         }

include/ck/tensor_operation/gpu/device/device_gemm_xdl_c_shuffle_bias_activation_add.hpp

@@ -312,7 +312,10 @@ struct DeviceGemmXdl_C_Shuffle_Bias_Activation_Add
     {
         using Argument = DeviceOp::Argument;
 
-        float Run(const Argument& arg, int nrepeat = 1, hipStream_t stream_id = nullptr, bool measure_time = false)
+        float Run(const Argument& arg,
+                  int nrepeat           = 1,
+                  hipStream_t stream_id = nullptr,
+                  bool measure_time     = false)
         {
             {
                 std::cout << "arg.a_grid_desc_k0_m_k1_{" << arg.a_grid_desc_k0_m_k1_.GetLength(I0)
@@ -448,7 +451,10 @@ struct DeviceGemmXdl_C_Shuffle_Bias_Activation_Add
         }
 
         // polymorphic
-        float Run(const BaseArgument* p_arg, int nrepeat = 1, hipStream_t stream_id = nullptr, bool measure_time = false) override
+        float Run(const BaseArgument* p_arg,
+                  int nrepeat           = 1,
+                  hipStream_t stream_id = nullptr,
+                  bool measure_time     = false) override
         {
             return Run(*dynamic_cast<const Argument*>(p_arg), nrepeat, stream_id, measure_time);
         }