use from_numpy and add do_copy option

mindspore/dataset/engine/datasets.py

@@ -1255,7 +1255,7 @@ class Dataset:
         del api_tree
 
     @check_tuple_iterator
-    def create_tuple_iterator(self, columns=None, num_epochs=-1, output_numpy=False):
+    def create_tuple_iterator(self, columns=None, num_epochs=-1, output_numpy=False, do_copy=True):
         """
         Create an iterator over the dataset. The data retrieved will be a list of ndarrays of data.
 
@@ -1269,6 +1269,8 @@ class Dataset:
                 (default=-1, iterator can be iterated infinite number of epochs)
             output_numpy (bool, optional): Whether or not to output NumPy datatype.
                 If output_numpy=False, iterator will output MSTensor (default=False).
+            do_copy (bool, optional): when output data type is mindspore.Tensor,
+                use this param to select the conversion method, only take False for better performance (default=True).
 
         Returns:
             Iterator, list of ndarrays.
@@ -1290,7 +1292,7 @@ class Dataset:
 
         if Dataset._noop_mode():
             return DummyIterator(self, 'tuple')
-        return TupleIterator(self, columns, num_epochs, output_numpy)
+        return TupleIterator(self, columns, num_epochs, output_numpy, do_copy)
 
     @check_dict_iterator
     def create_dict_iterator(self, num_epochs=-1, output_numpy=False):
@@ -2788,7 +2790,7 @@ class TransferDataset(Dataset):
     def create_dict_iterator(self, num_epochs=-1, output_numpy=False):
         raise RuntimeError("TransferDataset is not iterable.")
 
-    def create_tuple_iterator(self, columns=None, num_epochs=-1, output_numpy=False):
+    def create_tuple_iterator(self, columns=None, num_epochs=-1, output_numpy=False, do_copy=True):
         raise RuntimeError("TransferDataset is not iterable.")
 
     def __iter__(self):

mindspore/dataset/engine/iterators.py

@@ -63,7 +63,7 @@ class Iterator:
         dataset: Dataset to be iterated over
     """
 
-    def __init__(self, dataset, num_epochs=-1, output_numpy=False):
+    def __init__(self, dataset, num_epochs=-1, output_numpy=False, do_copy=True):
         self._col_names = None
 
         # create a copy of tree and work on it.
@@ -80,7 +80,10 @@ class Iterator:
 
         self._transform_tensor = lambda t: t.as_array()
         if not output_numpy:
-            self._transform_tensor = lambda t: Tensor(t.as_array())
+            if do_copy:
+                self._transform_tensor = lambda t: Tensor(t.as_array())
+            else:
+                self._transform_tensor = lambda t: Tensor.from_numpy(t.as_array())
         self._index = 0
 
         # todo remove next when ContextManager is done
@@ -179,13 +182,13 @@ class TupleIterator(Iterator):
     The derived class of Iterator with list type.
     """
 
-    def __init__(self, dataset, columns=None, num_epochs=-1, output_numpy=False):
+    def __init__(self, dataset, columns=None, num_epochs=-1, output_numpy=False, do_copy=True):
         if columns is not None:
             if not isinstance(columns, list):
                 columns = [columns]
             # todo: move next to IR
             dataset = dataset.project(columns)
-        super().__init__(dataset, num_epochs, output_numpy)
+        super().__init__(dataset, num_epochs, output_numpy, do_copy)
 
     def _get_next(self):
         """

mindspore/dataset/engine/validators.py

@@ -298,7 +298,7 @@ def check_tuple_iterator(method):
 
     @wraps(method)
     def new_method(self, *args, **kwargs):
-        [columns, num_epochs, _], param_dict = parse_user_args(method, *args, **kwargs)
+        [columns, num_epochs, _, _], param_dict = parse_user_args(method, *args, **kwargs)
         nreq_param_bool = ['output_numpy']
         validate_dataset_param_value(nreq_param_bool, param_dict, bool)
         if num_epochs is not None:

mindspore/train/dataset_helper.py

@@ -394,7 +394,7 @@ class _DatasetIterNormal:
         self.dataset = dataset
         self.device_num = _get_device_num()
         self.global_rank = _get_global_rank()
-        self.iter = self.dataset.create_tuple_iterator(num_epochs=epoch_num)
+        self.iter = self.dataset.create_tuple_iterator(num_epochs=epoch_num, do_copy=False)
 
     def __iter__(self):
         return self

tests/dataset_mock.py

@@ -55,7 +55,7 @@ class MindData:
         self.send_epoch_end = send_epoch_end
         return self
 
-    def create_tuple_iterator(self, num_epochs=-1):
+    def create_tuple_iterator(self, num_epochs=-1, do_copy=True):
         return self.__iter__()
 
     def send(self, num_epochs=-1):

tests/st/auto_parallel/optimizer_parallel.py

@@ -125,7 +125,7 @@ class FakeData:
     def set_label_onehot(self, is_onehot=True):
         self.is_onehot = is_onehot
 
-    def create_tuple_iterator(self, num_epochs=-1):
+    def create_tuple_iterator(self, num_epochs=-1, do_copy=True):
         _ = num_epochs
         return self
 

tests/st/auto_parallel/parallel_strategy_search.py

@@ -128,7 +128,7 @@ class FakeData:
     def set_label_onehot(self, is_onehot=True):
         self.is_onehot = is_onehot
 
-    def create_tuple_iterator(self, num_epochs=-1):
+    def create_tuple_iterator(self, num_epochs=-1, do_copy=True):
         _ = num_epochs
         return self
 

tests/st/pynative/loss_scale/test_loss_scale.py

@@ -60,7 +60,7 @@ class MindData:
     def output_shapes(self):
         return self._output_shapes
 
-    def create_tuple_iterator(self, num_epochs=-1):
+    def create_tuple_iterator(self, num_epochs=-1, do_copy=True):
         return self
 
     @property

tests/ut/python/parallel/test_auto_parallel_double_subgraphs.py

@@ -152,7 +152,7 @@ class DatasetLenet():
     def get_repeat_count(self):
         return 1
 
-    def create_tuple_iterator(self, num_epochs=-1):
+    def create_tuple_iterator(self, num_epochs=-1, do_copy=True):
         return self
 
 def test_double_subgraphs_train():

tests/ut/python/parallel/test_auto_parallel_resnet.py

@@ -275,7 +275,7 @@ class DatasetLenet():
     def get_repeat_count(self):
         return 1
 
-    def create_tuple_iterator(self, num_epochs=-1):
+    def create_tuple_iterator(self, num_epochs=-1, do_copy=True):
         return self
 
 

tests/ut/python/parallel/test_bias_add.py

@@ -61,7 +61,7 @@ class DatasetLenet():
     def get_repeat_count(self):
         return 1
 
-    def create_tuple_iterator(self, num_epochs=-1):
+    def create_tuple_iterator(self, num_epochs=-1, do_copy=True):
         return self
 
 

tests/ut/python/parallel/test_gather_v2_primitive.py

@@ -59,7 +59,7 @@ class Dataset():
     def get_repeat_count(self):
         return 1
 
-    def create_tuple_iterator(self, num_epochs=-1):
+    def create_tuple_iterator(self, num_epochs=-1, do_copy=True):
         return self
 
 

tests/ut/python/parallel/test_pipeline_split.py

@@ -51,7 +51,7 @@ class DatasetLenet():
     def get_batch_size(self):
         return 32
 
-    def create_tuple_iterator(self, num_epochs=1):
+    def create_tuple_iterator(self, num_epochs=1, do_copy=True):
         return self