Merge branch 'develop' into amd-develop

docs/sphinx/requirements.in

-rocm-docs-core==1.1.1
+rocm-docs-core==1.1.2
 sphinxcontrib-bibtex==2.6.2

docs/sphinx/requirements.txt

@@ -103,7 +103,7 @@ requests==2.31.0
     # via
     #   pygithub
     #   sphinx
-rocm-docs-core==1.1.1
+rocm-docs-core==1.1.2
     # via -r requirements.in
 six==1.16.0
     # via

include/ck/host_utility/flush_cache.hpp

@@ -117,7 +117,7 @@ float launch_and_time_kernel_with_preprocess(const StreamConfig& stream_config,
 #define MEDIAN 1
     if(stream_config.time_kernel_)
     {
-        if(ck::EnvIsEnabled(ENV(CK_LOGGING)))
+        if(ck::EnvIsEnabled(CK_ENV(CK_LOGGING)))
         {
             printf("%s: grid_dim {%u, %u, %u}, block_dim {%u, %u, %u} \n",
                    __func__,
@@ -142,7 +142,7 @@ float launch_and_time_kernel_with_preprocess(const StreamConfig& stream_config,
         {
             return 0.0;
         }
-        if(ck::EnvIsEnabled(ENV(CK_LOGGING)))
+        if(ck::EnvIsEnabled(CK_ENV(CK_LOGGING)))
         {
             printf("Start running %d times...\n", nrepeat);
         }
@@ -186,7 +186,7 @@ float launch_and_time_kernel_with_preprocess(const StreamConfig& stream_config,
             total_time += cur_time;
 #endif
 
-            if(ck::EnvIsEnabled(ENV(CK_LOGGING)))
+            if(ck::EnvIsEnabled(CK_ENV(CK_LOGGING)))
             {
                 std::cout << "i: " << i << " cur_time: " << cur_time << std::endl;
 

include/ck/host_utility/kernel_launch.hpp

@@ -20,7 +20,7 @@ float launch_and_time_kernel(const StreamConfig& stream_config,
 #if CK_TIME_KERNEL
     if(stream_config.time_kernel_)
     {
-        if(ck::EnvIsEnabled(ENV(CK_LOGGING)))
+        if(ck::EnvIsEnabled(CK_ENV(CK_LOGGING)))
         {
             printf("%s: grid_dim {%u, %u, %u}, block_dim {%u, %u, %u} \n",
                    __func__,
@@ -41,7 +41,7 @@ float launch_and_time_kernel(const StreamConfig& stream_config,
         }
 
         const int nrepeat = stream_config.nrepeat_;
-        if(ck::EnvIsEnabled(ENV(CK_LOGGING)))
+        if(ck::EnvIsEnabled(CK_ENV(CK_LOGGING)))
         {
             printf("Start running %d times...\n", nrepeat);
         }
@@ -95,7 +95,7 @@ float launch_and_time_kernel_with_preprocess(const StreamConfig& stream_config,
 #if CK_TIME_KERNEL
     if(stream_config.time_kernel_)
     {
-        if(ck::EnvIsEnabled(ENV(CK_LOGGING)))
+        if(ck::EnvIsEnabled(CK_ENV(CK_LOGGING)))
         {
             printf("%s: grid_dim {%u, %u, %u}, block_dim {%u, %u, %u} \n",
                    __func__,
@@ -117,7 +117,7 @@ float launch_and_time_kernel_with_preprocess(const StreamConfig& stream_config,
         }
 
         const int nrepeat = stream_config.nrepeat_;
-        if(ck::EnvIsEnabled(ENV(CK_LOGGING)))
+        if(ck::EnvIsEnabled(CK_ENV(CK_LOGGING)))
         {
             printf("Start running %d times...\n", nrepeat);
         }

include/ck/tensor_operation/gpu/block/blockwise_gemm_xdlops.hpp

@@ -795,11 +795,6 @@ struct BlockwiseGemmXdlops_v2
                       "wrong!");
     }
 
-    __host__ __device__ BlockwiseGemmXdlops_v2(const BlockwiseGemmXdlops_v2& other)
-        : a_thread_copy_(other.a_origin), b_thread_copy_(other.b_origin)
-    {
-    }
-
     // transposed XDL output supporting C_xdl' = B_xdl' * A_xdl'
     __host__ __device__ static constexpr auto GetCThreadDescriptor_M0_N0_M1_N1_M2_N2_N3_N4()
     {

include/ck/tensor_operation/gpu/device/impl/device_batched_gemm_multiple_d_gemm_multiple_d_xdl_cshuffle.hpp

@@ -587,7 +587,7 @@ struct DeviceBatchedGemmMultipleDGemmMultipleD_Xdl_CShuffle
                                          BatchStrideD1s,
                                          BatchStrideE1}
         {
-            if(ck::EnvIsEnabled(ENV(CK_LOGGING)))
+            if(ck::EnvIsEnabled(CK_ENV(CK_LOGGING)))
             {
                 std::cout << "a0_grid_desc_m_k_{" << a0_grid_desc_m_k_.GetLength(I0) << ", "
                           << a0_grid_desc_m_k_.GetLength(I1) << "}" << std::endl;

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

@@ -658,7 +658,7 @@ struct DeviceBatchedGemmReduce_Xdl_CShuffle : public DeviceGemmReduce<0, ReduceO
 
         float Run(const Argument& arg, const StreamConfig& stream_config = StreamConfig{})
         {
-            if(ck::EnvIsEnabled(ENV(CK_LOGGING)))
+            if(ck::EnvIsEnabled(CK_ENV(CK_LOGGING)))
             {
                 {
                     std::cout << "arg.Batch_ = " << arg.Batch_ << std::endl;

include/ck/tensor_operation/gpu/device/impl/device_batched_gemm_softmax_gemm_permute_xdl_cshuffle.hpp

@@ -719,7 +719,7 @@ struct DeviceBatchedGemmSoftmaxGemmPermute_Xdl_CShuffle
 
     static bool IsSupportedArgument(const Argument& arg)
     {
-        if(ck::EnvIsEnabled(ENV(CK_LOGGING)))
+        if(ck::EnvIsEnabled(CK_ENV(CK_LOGGING)))
         {
             arg.Print();
         }

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

@@ -516,7 +516,7 @@ struct DeviceConv2dBwdDataXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K
             float ave_time = 0;
             for(size_t i = 0; i < arg.a_grid_desc_k0_m_k1_container_.size(); i++)
             {
-                if(ck::EnvIsEnabled(ENV(CK_LOGGING)))
+                if(ck::EnvIsEnabled(CK_ENV(CK_LOGGING)))
                 {
                     {
                         std::cout << "arg.a_grid_desc_k0_m_k1_container_{"

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

@@ -644,7 +644,7 @@ struct
 
         float Run(const Argument& arg, const StreamConfig& stream_config = StreamConfig{})
         {
-            if(ck::EnvIsEnabled(ENV(CK_LOGGING)))
+            if(ck::EnvIsEnabled(CK_ENV(CK_LOGGING)))
             {
                 std::cout << DeviceOp{}.GetTypeString() << std::endl;
                 std::cout << "N " << arg.Conv_N_ << ", "

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

@@ -614,7 +614,7 @@ struct DeviceConv2dFwdXdl_C_Shuffle_Bias_Activation_Input_N_Hi_Wi_C_Weight_K_Y_X
 
         float Run(const Argument& arg, const StreamConfig& stream_config = StreamConfig{})
         {
-            if(ck::EnvIsEnabled(ENV(CK_LOGGING)))
+            if(ck::EnvIsEnabled(CK_ENV(CK_LOGGING)))
             {
                 std::cout << DeviceOp{}.GetTypeString() << std::endl;
                 std::cout << "N " << arg.Conv_N_ << ", "

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

@@ -579,7 +579,7 @@ struct DeviceConv2dFwdXdl_C_Shuffle_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_W
 
         float Run(const Argument& arg, const StreamConfig& stream_config = StreamConfig{})
         {
-            if(ck::EnvIsEnabled(ENV(CK_LOGGING)))
+            if(ck::EnvIsEnabled(CK_ENV(CK_LOGGING)))
             {
                 std::cout << DeviceOp{}.GetTypeString() << std::endl;
                 std::cout << "N " << arg.Conv_N_ << ", "

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

@@ -431,7 +431,7 @@ struct DeviceConv2dFwdXdl_Input_N_Hi_Wi_C_Weight_K_Y_X_C_Output_N_Ho_Wo_K
 
         float Run(const Argument& arg, const StreamConfig& stream_config = StreamConfig{})
         {
-            if(ck::EnvIsEnabled(ENV(CK_LOGGING)))
+            if(ck::EnvIsEnabled(CK_ENV(CK_LOGGING)))
             {
                 std::cout << "arg.a_grid_desc_k0_m_k1_{" << arg.a_grid_desc_k0_m_k1_.GetLength(I0)
                           << ", " << arg.a_grid_desc_k0_m_k1_.GetLength(I1) << ", "

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

@@ -401,7 +401,7 @@ struct DeviceConv3dFwdXdl_Input_N_Di_Hi_Wi_C_Weight_K_Z_Y_X_C_Output_N_Do_Ho_Wo_
 
         float Run(const Argument& arg, const StreamConfig& stream_config = StreamConfig{})
         {
-            if(ck::EnvIsEnabled(ENV(CK_LOGGING)))
+            if(ck::EnvIsEnabled(CK_ENV(CK_LOGGING)))
             {
                 std::cout << "num_batches_of_GEMM = " << arg.num_subbatches_ << std::endl;
                 std::cout << "a_grid_desc_k0_m_k1{" << arg.a_grid_desc_k0_m_k1_.GetLength(I0)

include/ck/tensor_operation/gpu/device/impl/device_convnd_bwd_data_nwc_kxc_nwk_dl.hpp

@@ -1272,7 +1272,7 @@ struct DeviceConvNdBwdDataNwcKxcNwk_Dl
             float ave_time = 0;
             for(size_t i = 0; i < arg.a_grid_desc_k0_m_k1_container_.size(); i++)
             {
-                if(ck::EnvIsEnabled(ENV(CK_LOGGING)))
+                if(ck::EnvIsEnabled(CK_ENV(CK_LOGGING)))
                 {
                     std::cout << "arg.a_grid_desc_k0_m_k1_container_{"
                               << arg.a_grid_desc_k0_m_k1_container_[i].GetLength(I0) << ", "

include/ck/tensor_operation/gpu/device/impl/device_convnd_bwd_data_nwc_kxc_nwk_xdl.hpp

@@ -1220,7 +1220,7 @@ struct DeviceConvNdBwdDataNwcKxcNwk_Xdl
             float ave_time = 0;
             for(size_t i = 0; i < arg.a_grid_desc_k0_m_k1_container_.size(); i++)
             {
-                if(ck::EnvIsEnabled(ENV(CK_LOGGING)))
+                if(ck::EnvIsEnabled(CK_ENV(CK_LOGGING)))
                 {
                     std::cout << "arg.a_grid_desc_k0_m_k1{"
                               << arg.a_grid_desc_k0_m_k1_container_[i].GetLength(I0) << ", "

include/ck/tensor_operation/gpu/device/impl/device_gemm_dl.hpp

@@ -334,7 +334,7 @@ struct DeviceGemmDl : public DeviceGemm<ALayout,
 
         float Run(const Argument& arg, const StreamConfig& stream_config = StreamConfig{})
         {
-            if(ck::EnvIsEnabled(ENV(CK_LOGGING)))
+            if(ck::EnvIsEnabled(CK_ENV(CK_LOGGING)))
             {
                 std::cout << "arg.a_grid_desc_k0_m0_m1_k1_{"
                           << arg.a_grid_desc_k0_m_k1_.GetLength(I0) << ", "

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

@@ -510,7 +510,7 @@ struct DeviceGemmReduce_Xdl_CShuffle : public DeviceGemmReduce<0, ReduceOperatio
 
         float Run(const Argument& arg, const StreamConfig& stream_config = StreamConfig{})
         {
-            if(ck::EnvIsEnabled(ENV(CK_LOGGING)))
+            if(ck::EnvIsEnabled(CK_ENV(CK_LOGGING)))
             {
                 std::cout << "arg.a_grid_desc_ak0_m_ak1_{"
                           << arg.a_grid_desc_ak0_m_ak1_.GetLength(I0) << ", "

include/ck/tensor_operation/gpu/device/impl/device_gemm_xdl_layernorm_cshuffle.hpp

@@ -514,7 +514,7 @@ struct DeviceGemmLayerNorm_Xdl_CShuffle : public BaseOperator
 
         float Run(const Argument& arg, const StreamConfig& stream_config = StreamConfig{})
         {
-            if(ck::EnvIsEnabled(ENV(CK_LOGGING)))
+            if(ck::EnvIsEnabled(CK_ENV(CK_LOGGING)))
             {
                 std::cout << "arg.a_grid_desc_ak0_m_ak1_{"
                           << arg.a_grid_desc_ak0_m_ak1_.GetLength(I0) << ", "

include/ck/tensor_operation/gpu/device/impl/device_gemm_xdl_skip_b_lds.hpp

@@ -299,7 +299,7 @@ struct DeviceGemmXdlSkipBLds : public DeviceGemm<ALayout,
 
         float Run(const Argument& arg, const StreamConfig& stream_config = StreamConfig{})
         {
-            if(ck::EnvIsEnabled(ENV(CK_LOGGING)))
+            if(ck::EnvIsEnabled(CK_ENV(CK_LOGGING)))
             {
                 std::cout << "arg.a_grid_desc_k0_m_k1_{" << arg.a_grid_desc_k0_m_k1_.GetLength(I0)
                           << ", " << arg.a_grid_desc_k0_m_k1_.GetLength(I1) << ", "