!15552 use host shape instead of device shape for debugger

From: @john_tzanakakis
Reviewed-by: @yelihua,@pandoublefeng
Signed-off-by: @pandoublefeng

mindspore/ccsrc/debug/debugger/debugger.cc

@@ -1157,10 +1157,7 @@ void Debugger::LoadSingleAnfnode(const AnfNodePtr &anf_node, const size_t output
   }
   auto format = kOpFormat_DEFAULT;
   string tensor_name = node_name + ':' + "0";
-  ShapeVector int_shapes;
-  auto shape = AnfAlgo::GetOutputDeviceShape(anf_node, output_index);
-  (void)std::transform(shape.begin(), shape.end(), std::back_inserter(int_shapes),
-                       [](size_t inner_item) { return SizeToInt(inner_item); });
+  ShapeVector int_shapes = trans::GetRuntimePaddingShape(anf_node, output_index);
   bool keep_prev;
   if (anf_node->isa<Parameter>()) {
     keep_prev = true;
@@ -1222,10 +1219,7 @@ void Debugger::LoadGraphOutputs() {
       }
       auto format = kOpFormat_DEFAULT;
       string tensor_name = kernel_name + ':' + std::to_string(j);
-      ShapeVector int_shapes;
-      auto shape = AnfAlgo::GetOutputDeviceShape(node, j);
-      (void)std::transform(shape.begin(), shape.end(), std::back_inserter(int_shapes),
-                           [](size_t inner_item) { return SizeToInt(inner_item); });
+      ShapeVector int_shapes = trans::GetRuntimePaddingShape(node, j);
       auto ret = addr->LoadMemToHost(tensor_name, exec_order, format, int_shapes, type, j, false);
       if (!ret) {
         MS_LOG(ERROR) << "LoadMemToHost:"

mindspore/ccsrc/debug/debugger/debugger.h

@@ -25,6 +25,7 @@
 #include "backend/session/kernel_graph.h"
 #include "debug/debugger/grpc_client.h"
 #include "debug/debug_services.h"
+#include "common/trans.h"
 
 using debugger::Chunk;
 using debugger::DataType;

mindspore/ccsrc/runtime/device/gpu/gpu_kernel_runtime.cc

@@ -157,10 +157,7 @@ void LoadKernelData(Debugger *debugger, const CNodePtr &kernel,
       auto format = kOpFormat_DEFAULT;
       auto gpu_addr = std::make_unique<GPUDeviceAddress>(addr->addr, addr->size, format, type);
       string input_tensor_name = input_kernel_name + ':' + "0";
-      ShapeVector int_shapes;
-      auto shape = AnfAlgo::GetOutputDeviceShape(input_kernel, PARAMETER_OUTPUT_INDEX);
-      (void)std::transform(shape.begin(), shape.end(), std::back_inserter(int_shapes),
-                           [](size_t inner_item) { return SizeToInt(inner_item); });
+      ShapeVector int_shapes = trans::GetRuntimePaddingShape(input_kernel, PARAMETER_OUTPUT_INDEX);
       auto ret = gpu_addr->LoadMemToHost(input_tensor_name, exec_order, format, int_shapes, type, 0, true);
       if (!ret) {
         MS_LOG(ERROR) << "LoadMemToHost:"
@@ -187,10 +184,7 @@ void LoadKernelData(Debugger *debugger, const CNodePtr &kernel,
       auto format = kOpFormat_DEFAULT;
       auto gpu_addr = std::make_unique<GPUDeviceAddress>(addr->addr, addr->size, format, type);
       string tensor_name = kernel_name + ':' + std::to_string(j);
-      ShapeVector int_shapes;
-      auto shape = AnfAlgo::GetOutputDeviceShape(kernel, j);
-      (void)std::transform(shape.begin(), shape.end(), std::back_inserter(int_shapes),
-                           [](size_t inner_item) { return SizeToInt(inner_item); });
+      ShapeVector int_shapes = trans::GetRuntimePaddingShape(kernel, j);
       auto ret = gpu_addr->LoadMemToHost(tensor_name, exec_order, format, int_shapes, type, j, false);
       if (!ret) {
         MS_LOG(ERROR) << "LoadMemToHost:"