Update nn.Range for GPU backend.

mindspore/nn/layer/math.py

@@ -15,6 +15,7 @@
 """math"""
 import math
 import numpy as np
+import mindspore.context as context
 from mindspore.ops import operations as P
 from mindspore.ops.operations import _inner_ops as inner
 from mindspore.common.tensor import Tensor
@@ -116,7 +117,7 @@ class Range(Cell):
         Tensor, the dtype is int if the dtype of `start`, `limit` and `delta` all are int. Otherwise, dtype is float.
 
     Supported Platforms:
-        ``Ascend`` ``CPU``
+        ``Ascend`` ``GPU`` ``CPU``
 
     Examples:
         >>> net = nn.Range(1, 8, 2)
@@ -127,6 +128,7 @@ class Range(Cell):
 
     def __init__(self, start, limit=None, delta=1):
         super(Range, self).__init__()
+        self.is_gpu = context.get_context("device_target") == "GPU"
         validator.check_value_type("start", start, [int, float], self.cls_name)
         validator.check_value_type("delta", delta, [int, float], self.cls_name)
         if delta == 0:
@@ -155,8 +157,17 @@ class Range(Cell):
             length_input = math.ceil((limit - start) / delta)
         self.input_tensor = Tensor(list(range(length_input)), self.dtype)
 
+        if self.is_gpu:
+            self.start = Tensor(start, self.dtype)
+            self.limit = Tensor(limit, self.dtype)
+            self.delta = Tensor(delta, self.dtype)
+            self.range_gpu = P.Range(length_input)
+
     def construct(self):
-        range_out = self.range_x(self.input_tensor)
+        if self.is_gpu:
+            range_out = self.range_gpu(self.start, self.limit, self.delta)
+        else:
+            range_out = self.range_x(self.input_tensor)
         return range_out
 
 

mindspore/ops/operations/array_ops.py

@@ -4761,7 +4761,7 @@ class Range(PrimitiveWithCheck):
         [0, 4, 8]
 
     Supported Platforms:
-        ``Ascend`` ``GPU``
+        ``GPU``
     """
 
     @prim_attr_register