diff --git a/doc/array.rst b/doc/array.rst
index 3ece17dc..f26ac4d7 100644
--- a/doc/array.rst
+++ b/doc/array.rst
@@ -148,7 +148,7 @@ The :class:`GPUArray` Array Class
         size (not necessarily shape) and dtype. If it is not given,
         a *page-locked* array is newly allocated.
 
-    .. method :: copy()
+    .. method :: copy(stream=None)
 
         .. versionadded :: 2013.1
 
@@ -187,7 +187,7 @@ The :class:`GPUArray` Array Class
 
     .. method :: astype(dtype, stream=None)
 
-        Return *self*, cast to *dtype*.
+        Return copy of the array, cast to a specified type.
 
     .. method :: any(stream=None, allocator=None)
 
diff --git a/pycuda/gpuarray.py b/pycuda/gpuarray.py
index fedd4bf0..309c5f94 100644
--- a/pycuda/gpuarray.py
+++ b/pycuda/gpuarray.py
@@ -396,9 +396,9 @@ def get(self, ary=None, pagelocked=False, async_=False, stream=None, **kwargs):
     def get_async(self, stream=None, ary=None):
         return self.get(ary=ary, async_=True, stream=stream)
 
-    def copy(self):
+    def copy(self, stream=None):
         new = GPUArray(self.shape, self.dtype, self.allocator, strides=self.strides)
-        _memcpy_discontig(new, self)
+        _memcpy_discontig(new, self, **{"async_": True, "stream": stream} if stream is not None else {})
         return new
 
     def __str__(self):
@@ -926,11 +926,11 @@ def reverse(self, stream=None):
     def astype(self, dtype, stream=None):
         if not self.flags.forc:
             raise RuntimeError(
-                "only contiguous arrays may " "be used as arguments to this operation"
+                "only contiguous arrays may be used as arguments to this operation"
             )
 
         if dtype == self.dtype:
-            return self.copy()
+            return self.copy(stream=stream)
 
         result = self._new_like_me(dtype=dtype)