incorporate requested changes for PR:24394 (#661)

Co-authored-by: sfatimar <[email protected]>

Author: 2 people

cmake/onnxruntime_providers_openvino.cmake

@@ -46,7 +46,7 @@
   set_target_properties(onnxruntime_providers_openvino PROPERTIES FOLDER "ONNXRuntime")
 
   target_compile_options(onnxruntime_providers_openvino PRIVATE
-    $<$<NOT:$<CONFIG:Release>>:-DNOT_RELEASE>
+  $<$<CONFIG:Release>:-DRELEASE>
   )
 
   if(NOT MSVC)

onnxruntime/core/providers/openvino/backend_manager.cc

@@ -20,7 +20,6 @@
 #include "core/providers/openvino/ov_interface.h"
 #include "core/providers/openvino/ov_versions/capability.h"
 #include "core/providers/openvino/qdq_transformations/qdq_stripping.h"
-#include "core/providers/openvino/ov_interface.h"
 
 namespace onnxruntime {
 namespace openvino_ep {
@@ -325,7 +324,7 @@ static bool IsQDQGraph(const onnxruntime::GraphViewer& graph_viewer) {
 static void DumpOpenVINOEPModel([[maybe_unused]] const std::filesystem::path& onnx_model_path_name,
                                 [[maybe_unused]] ONNX_NAMESPACE::ModelProto* model_proto,
                                 [[maybe_unused]] const onnxruntime::Node& fused_node) {
-#ifdef  NOT_RELEASE
+#ifndef  RELEASE
   if (openvino_ep::backend_utils::IsDebugEnabled()) {
     auto model_name = onnx_model_path_name.empty() ? "unknown.onnx" : onnx_model_path_name.filename();
 
@@ -385,7 +384,7 @@ BackendManager::GetModelProtoFromFusedNode(const onnxruntime::Node& fused_node,
   if (session_context_.device_type.find("NPU") != std::string::npos &&
       (enable_ovep_qdq_optimizer || session_context_.so_share_ep_contexts)) {
     std::unique_ptr<onnxruntime::Model> model;
-    Status status = CreateModelWithStrippedQDQNodes(subgraph, logger, session_context_.so_share_ep_contexts, model, shared_context_.shared_weights, enable_ovep_qdq_optimizer);
+    Status status = CreateModelWithStrippedQDQNodes(subgraph, logger, session_context_.so_share_ep_contexts, enable_ovep_qdq_optimizer, model, shared_context_.shared_weights);
     auto model_proto = model->ToProto();
     model_proto->set_ir_version(ONNX_NAMESPACE::Version::IR_VERSION);
     print_model_proto_duration();

onnxruntime/core/providers/openvino/qdq_transformations/qdq_stripping.cc

@@ -449,8 +449,8 @@ static bool HandleDoubleQDQ(onnxruntime::Graph& dst_graph, const onnxruntime::Gr
 static void AddStandaloneNodeUnit(onnxruntime::Graph& dst_graph, const onnxruntime::GraphViewer& src_graph,
                                   const NodeUnit& node_unit,
                                   std::set<std::string>& initializers_to_keep,
-                                  const logging::Logger& /* logger */,
-                                  bool IsWeightSharingWithoutOVEPQDQStripping) {
+                                  bool IsWeightSharingWithoutOVEPQDQStripping,
+                                  const logging::Logger& /* logger */) {
   assert(node_unit.UnitType() == NodeUnit::Type::SingleNode);
 
   // this is the scenario where WAI is enabled and ovep stripping is disabled
@@ -521,8 +521,8 @@ static void AddQDQNodeUnit(onnxruntime::Graph& dst_graph,
                            const onnxruntime::GraphViewer& src_graph,
                            const NodeUnit& node_unit,
                            std::set<std::string>& initializers_to_keep,
-                           const logging::Logger& /* logger */,
-                           bool IsWeightSharingWithoutOVEPQDQStripping) {
+                           bool IsWeightSharingWithoutOVEPQDQStripping,
+                           const logging::Logger& /* logger */) {
   assert(node_unit.UnitType() == NodeUnit::Type::QDQGroup);
 
   // Collect inputs coming into the node unit.
@@ -683,8 +683,7 @@ Status CreateModelWithStrippedQDQNodes(const GraphViewer& src_graph,
                                        bool enable_ovep_weight_sharing,
                                        bool enable_ovep_qdq_optimizer,
                                        /*out*/ std::unique_ptr<onnxruntime::Model>& model,
-                                       /*out*/ sw& shared_weights,
-                                       bool enable_ovep_qdq_optimizer) {
+                                       /*out*/ sw& shared_weights) {
   // NOTE: This function is a re-implementation of GraphViewerToProto() in core/graph/graph_proto_serializer.cc
   // with the following differences:
   //   - Uses onnxruntime::Graph APIs instead of onnx::GraphProto APIs.
@@ -778,9 +777,9 @@ Status CreateModelWithStrippedQDQNodes(const GraphViewer& src_graph,
     bool IsWeightSharingWithoutOVEPQDQStripping = enable_ovep_weight_sharing && !enable_ovep_qdq_optimizer;
 
     if (node_unit->UnitType() == NodeUnit::Type::SingleNode) {
-      AddStandaloneNodeUnit(dst_graph, src_graph, *node_unit, initializers_to_keep, logger, IsWeightSharingWithoutOVEPQDQStripping);
+      AddStandaloneNodeUnit(dst_graph, src_graph, *node_unit, initializers_to_keep, IsWeightSharingWithoutOVEPQDQStripping, logger);
     } else {
-      AddQDQNodeUnit(dst_graph, src_graph, *node_unit, initializers_to_keep, logger, IsWeightSharingWithoutOVEPQDQStripping);
+      AddQDQNodeUnit(dst_graph, src_graph, *node_unit, initializers_to_keep, IsWeightSharingWithoutOVEPQDQStripping, logger);
     }
 
     seen_node_units.insert(node_unit);

onnxruntime/core/providers/openvino/qdq_transformations/qdq_stripping.h

@@ -18,8 +18,7 @@ Status CreateModelWithStrippedQDQNodes(const GraphViewer& src_graph,
                                        bool enable_ovep_weight_sharing,
                                        bool enable_ovep_qdq_optimizer,
                                        /*out*/ std::unique_ptr<onnxruntime::Model>& model,
-                                       /*out*/ sw& shared_weights,
-                                       bool enable_ovep_qdq_optimizer);
+                                       /*out*/ sw& shared_weights);
 
 bool dumpMetaDataMapToBinary(const sw::Metadata::Map& shared_weights, const std::string& filename);
 }  // namespace openvino_ep