Doc corrections for new gpu backend

doc/install.txt

@@ -524,8 +524,8 @@ You can also set these options in the .theanorc file's ``[global]`` section:
 Note that:
 
     * If your computer has multiple GPUs and you use 'device=cuda', the driver
-      selects the one to use (usually gpu0).
-    * You can use the program nvida-smi to change this policy.
+      selects the one to use (usually cuda0).
+    * You can use the program ``nvidia-smi`` to change this policy.
     * You can choose one specific GPU by specifying 'device=cudaX', with X the
       the corresponding GPU index (0, 1, 2, ...)
     * By default, when ``device`` indicates preference for GPU computations,

doc/library/config.txt

@@ -103,13 +103,16 @@ import theano and print the config variable, as in:
 .. attribute:: device
 
     String value: either ``'cpu'``, ``'cuda'``, ``'cuda0'``, ``'cuda1'``,
-    ``'opencl0:0'``, or ``'opencl0:1'`` ...
+    ``'opencl0:0'``, ``'opencl0:1'``, ``'gpu'``, ``'gpu0'`` ...
 
     Default device for computations. If ``'cuda*``, change the default to try
     to move computation to the GPU using CUDA libraries. If ``'opencl*'``,
     the openCL libraries will be used. To let the driver select the device,
-    use ``'cuda'`` or ``'opencl'``. If we are not able to use the GPU,
-    either we fall back on the CPU, or an error is raised, depending on the :attr:`force_device` flag.
+    use ``'cuda'`` or ``'opencl'``. If ``'gpu*'``, the old gpu backend will
+    be used, although users are encouraged to migrate to the new GpuArray 
+    backend. If we are not able to use the GPU,
+    either we fall back on the CPU, or an error is raised, depending
+    on the :attr:`force_device` flag.
 
     This flag's value cannot be modified during the program execution.
 
@@ -135,11 +138,11 @@ import theano and print the config variable, as in:
 .. attribute:: init_gpu_device
 
     String value: either ``''``, ``'cuda'``, ``'cuda0'``, ``'cuda1'``,
-    ``'opencl0:0'``, or ``'opencl0:1'`` ...
+    ``'opencl0:0'``, ``'opencl0:1'``, ``'gpu'``, ``'gpu0'`` ...
     
     Initialize the gpu device to use.
-    When its value is cuda* or opencl*, the theano flag :attr:`device` must
-    be ``"cpu"``.
+    When its value is ``'cuda*'``, ``'opencl*'`` or ``'gpu*'``, the theano
+    flag :attr:`device` must be ``'cpu'``.
     Unlike :attr:`device`, setting this flag to a specific GPU will not
     try to use this device by default, in particular it will **not** move
     computations, nor shared variables, to the specified GPU.

doc/tutorial/using_gpu.txt

@@ -133,7 +133,7 @@ the GPU object directly.  The following code is modified to do just that.
 
   rng = numpy.random.RandomState(22)
   x = shared(numpy.asarray(rng.rand(vlen), config.floatX))
-  f = function([], tensor.exp(x).transfer('dev0'))
+  f = function([], tensor.exp(x).transfer(None))
   print(f.maker.fgraph.toposort())
   t0 = time.time()
   for i in range(iters):
@@ -148,7 +148,7 @@ the GPU object directly.  The following code is modified to do just that.
   else:
       print('Used the gpu')
 
-Here ``tensor.exp(x).transfer('None')`` means "copy ``exp(x)`` to the GPU",
+Here ``tensor.exp(x).transfer(None)`` means "copy ``exp(x)`` to the GPU",
 with ``None`` the default GPU context when not explicitly given.
 For information on how to set GPU contexts, see :ref:`tut_using_multi_gpu`. 
 
@@ -158,12 +158,15 @@ The output is
    :hide:
    :options: +ELLIPSIS, +SKIP
 
-   Using device cuda0: ...
-   [GpuElemwise{exp,no_inplace}(<GpuArray<float64>>)]
-   Looping 1000 times took ... seconds
-   Result is ...
+   $ THEANO_FLAGS=device=cuda0 python gpu_tutorial2.py
+   Mapped name None to device cuda0: GeForce GTX 680 (cuDNN version 5004)
+   [GpuElemwise{exp,no_inplace}(<GpuArrayType<None>(float64, (False,))>)]
+   Looping 1000 times took 0.088381 seconds
+   Result is [ 1.23178032  1.61879341  1.52278065 ...,  2.20771815  2.29967753
+     1.62323285]
    Used the gpu
 
+
 .. code-block:: none
 
   $ THEANO_FLAGS=device=cuda0 python gpu_tutorial2.py
@@ -212,7 +215,7 @@ double (float64) or small (less than 32 bits like int16) data types.
 You will get an error at compile time or runtime if this is the case.
 
 By default all inputs will get transferred to GPU. You can prevent an
-input from getting transferred by setting its tag.target attribute to
+input from getting transferred by setting its ``tag.target`` attribute to
 'cpu'.
 
 Complex support is untested and most likely completely broken.
@@ -225,9 +228,10 @@ Tips for Improving Performance on GPU
 * The GPU backend supports *float64* variables, but they are still slower
   to compute than *float32*. The more *float32*, the better GPU performance
   you will get. 
-* Prefer constructors like ``matrix``, ``vector`` and ``scalar`` to
-  ``dmatrix``, ``dvector`` and ``dscalar`` because the former will give
-  you *float32* variables and ignore the type given to ``floatX``.
+* Prefer constructors like ``matrix``, ``vector`` and ``scalar`` (which 
+  follow the type set in ``floatX``) to ``dmatrix``, ``dvector`` and
+  ``dscalar``. The latter enforce double precision (*float64* on most 
+  machines), which slows down GPU computations on current hardware.
 * Minimize transfers to the GPU device by using ``shared`` variables
   to store frequently-accessed data (see :func:`shared()<shared.shared>`).
   When using the GPU, tensor ``shared`` variables are stored on