Pre Merge pull request !1706 from 王涛/master

ge/graph/load/model_manager/davinci_model.cc

@@ -488,6 +488,8 @@ void DavinciModel::InitRuntimeParams() {
   session_scope_mem_info.memory_size = static_cast<size_t>(ret ? value : 0);
   runtime_param_.memory_infos[kSessionScopeMemory | RT_MEMORY_HBM] = std::move(session_scope_mem_info);
 
+  ret = ge::AttrUtils::GetInt(ge_model_, ATTR_MODEL_ZERO_COPY_MEMORY_SIZE, value);
+  runtime_param_.zero_copy_size = ret ? (uint64_t)value : 0;
   GELOGI("InitRuntimeParams(), %s.", runtime_param_.ToString().c_str());
 }
 

ge/graph/load/model_manager/davinci_model.h

@@ -248,6 +248,13 @@ class DavinciModel {
   // get total mem size
   size_t TotalMemSize() const { return runtime_param_.mem_size; }
 
+  // get total useful size: total mem size - zero copy size;
+  // In known subgraph, no need to allocate zero copy memory during initialization
+  size_t TotalMemSizeExcludeZeroCopy() const {
+    return runtime_param_.mem_size > runtime_param_.zero_copy_size
+             ? runtime_param_.mem_size - runtime_param_.zero_copy_size : 0;
+  }
+
   // model name
   string Name() const { return name_; }
 

ge/graph/load/model_manager/task_info/task_info.h

@@ -49,7 +49,7 @@ struct RuntimeParam {
        << ", label_num:" << label_num << ", logic_mem_base:" << logic_mem_base
        << ", logic_weight_base:" << logic_weight_base << ", logic_var_base:" << logic_var_base
        << ", memory_size:" << mem_size << ", weight_size:" << weight_size << ", var_size:" << var_size
-       << ", ex_memory_info:";
+       << ", zero_copy_size:" << zero_copy_size << ", ex_memory_info:";
     for (auto it : memory_infos) {
       ss << "[memory_type:" << it.first << ", memory_size:" << it.second.memory_size << "]";
     }
@@ -65,6 +65,7 @@ struct RuntimeParam {
   uint64_t var_size = 0;
   uint64_t logic_var_base = 0;
   uint8_t *var_base = nullptr;
+  uint64_t zero_copy_size = 0;
   std::map<uint64_t, MemInfo> memory_infos;
   uint32_t batch_num = 0;
   uint32_t stream_num = 0;

ge/hybrid/node_executor/compiledsubgraph/known_node_executor.cc

@@ -101,10 +101,10 @@ Status KnownNodeTask::Init(TaskContext &context) {
   GE_CHK_STATUS_RET(context.AllocateOutputs(), "[Allocate][Outputs] failed for %s.", context.GetNodeName());
   // allocate mem base
   void *buffer = nullptr;
-  if (davinci_model_->TotalMemSize() != 0) {
+  if (davinci_model_->TotalMemSizeExcludeZeroCopy != 0) {
     RECORD_EXECUTION_EVENT(context.GetExecutionContext(), context.GetNodeName(),
                            "[KnownNodeTask_AllocateWorkspace] Start");
-    GE_CHK_STATUS_RET(context.AllocateWorkspace(davinci_model_->TotalMemSize(), &buffer,
+    GE_CHK_STATUS_RET(context.AllocateWorkspace(davinci_model_->TotalMemSizeExcludeZeroCopy(), &buffer,
                                                 davinci_model_->GetRuntimeParam().mem_base),
                       "[Allocate][Workspace] failed for %s.", context.GetNodeName());
     RECORD_EXECUTION_EVENT(context.GetExecutionContext(), context.GetNodeName(),