Add method for checking memory address changes in the zero tensor

Signed-off-by: Bogdan Pereanu <[email protected]>

src/plugins/intel_npu/src/backend/include/zero_infer_request.hpp

@@ -15,14 +15,10 @@
 #include "zero_pipeline.hpp"
 #include "zero_profiling.hpp"
 #include "zero_remote_tensor.hpp"
+#include "zero_tensor.hpp"
 
 namespace intel_npu {
 
-struct TensorInfo {
-    bool tensorCreatedLocally;
-    uint64_t originalMemoryId;
-};
-
 class ZeroInferRequest final : public SyncInferRequest {
 public:
     explicit ZeroInferRequest(const std::shared_ptr<ZeroInitStructsHolder>& initStructs,
@@ -67,12 +63,9 @@ class ZeroInferRequest final : public SyncInferRequest {
     std::shared_ptr<ov::ITensor>& get_level_zero_input(size_t index, size_t tensorNo = 0) const;
     std::vector<std::shared_ptr<ov::ITensor>>& get_level_zero_inputs(size_t index) const;
 
-    std::shared_ptr<ov::ITensor> allocate_tensor(
-        const IODescriptor& descriptor,
-        const size_t index,
-        const bool isInput,
-        const ov::Allocator& allocator = {},
-        const std::optional<std::size_t> batchSize = std::nullopt) const override;
+    std::shared_ptr<ov::ITensor> create_tensor(ov::element::Type type,
+                                               const ov::Shape& shape,
+                                               const ov::Allocator& allocator = {}) const override;
 
     const std::shared_ptr<ZeroInitStructsHolder> _initStructs;
     const std::shared_ptr<IGraph> _graph;
@@ -84,9 +77,6 @@ class ZeroInferRequest final : public SyncInferRequest {
     mutable std::vector<std::vector<std::shared_ptr<ov::ITensor>>> _levelZeroInputTensors;
     mutable std::vector<std::shared_ptr<ov::ITensor>> _levelZeroOutputTensors;
 
-    mutable std::vector<TensorInfo> _levelZeroInputTensorInfo;
-    mutable std::vector<TensorInfo> _levelZeroOutputTensorInfo;
-
     ze_device_properties_t _properties = {};
     std::shared_ptr<const zeroMemory::HostMemAllocator> _inputAllocator;
     std::shared_ptr<const zeroMemory::HostMemAllocator> _outputAllocator;

src/plugins/intel_npu/src/backend/include/zero_pipeline.hpp

@@ -7,22 +7,22 @@
 #include "intel_npu/common/igraph.hpp"
 #include "intel_npu/utils/zero/zero_utils.hpp"
 #include "intel_npu/utils/zero/zero_wrappers.hpp"
-#include "openvino/runtime/itensor.hpp"
 #include "zero_memory.hpp"
 #include "zero_profiling.hpp"
+#include "zero_tensor.hpp"
 
 namespace intel_npu {
 
 struct Pipeline {
 public:
     Pipeline(const Config& config,
-             const std::shared_ptr<ZeroInitStructsHolder>& initStructs,
+             const std::shared_ptr<ZeroInitStructsHolder>& init_structs,
              const std::shared_ptr<IGraph>& graph,
              zeroProfiling::ProfilingPool& profiling_pool,
              zeroProfiling::ProfilingQuery& profiling_query,
              const std::shared_ptr<zeroProfiling::NpuInferProfiling>& npu_profiling,
-             const std::vector<std::vector<std::shared_ptr<ov::ITensor>>>& inputTensorsData,
-             const std::vector<std::shared_ptr<ov::ITensor>>& outputTensorsData,
+             const std::vector<std::vector<std::shared_ptr<ov::ITensor>>>& input_tensors_data,
+             const std::vector<std::shared_ptr<ov::ITensor>>& output_tensors_data,
              uint32_t group_ordinal);
 
     Pipeline(const Pipeline&) = delete;
@@ -33,8 +33,8 @@ struct Pipeline {
     void pull();
     void reset() const;
 
-    void updateCommandList(const void* data, size_t byte_size, uint32_t index);
-    void updateCommandList(const void* data, uint32_t index, size_t commandListIndex);
+    void updateCommandList(uint32_t arg_index, const void* arg_data, size_t byte_size);
+    void updateCommandListIndex(uint32_t arg_index, const void* arg_data, size_t command_list_index);
 
 protected:
     std::shared_ptr<IGraph> _graph;

src/plugins/intel_npu/src/backend/include/zero_tensor.hpp

@@ -15,14 +15,20 @@
 
 namespace intel_npu {
 
+/**
+ * @brief Constructs Tensor using element type and shape. Allocate internal host storage using custom allocator.
+ * @details The implementation is simillar with the AllocatedTensor class from OV repository.
+ * @note Set_shape method throw an error in case re-allocation is needed but this is not supported by the driver.
+ * There are two extra methods to notify the consumer if memorey changed or not and to reset the flag.
+ */
 class ZeroTensor final : public ov::ITensor {
 public:
     ZeroTensor(const std::shared_ptr<ZeroInitStructsHolder>& init_structs,
                const ov::element::Type element_type,
                const ov::Shape& shape,
                const ov::Allocator& allocator);
 
-    void* data(const ov::element::Type& element_type) const override;
+    void* data(const ov::element::Type& type = {}) const override;
 
     const ov::element::Type& get_element_type() const override;
 
@@ -32,6 +38,9 @@ class ZeroTensor final : public ov::ITensor {
 
     const ov::Strides& get_strides() const override;
 
+    bool memory_address_changed();
+    void reset_memory_flag();
+
     ~ZeroTensor();
 
 private:
@@ -51,6 +60,7 @@ class ZeroTensor final : public ov::ITensor {
     mutable std::once_flag _strides_once;
     ov::Allocator _allocator;
     void* _ptr = nullptr;
+    bool _reset_tensor_memory = false;
 };
 
 }  // namespace intel_npu