Theano reshape with -1 does not work on GPU in case of tensor3

gistfile1.py

from theano import tensor as T
import theano
import numpy as np

x = T.tensor3('')
xf = x.flatten()
xr1 = x.reshape((x.shape[0] * x.shape[1], x.shape[-1]))
xr2 = x.reshape((-1, x.shape[-1]))

ff = theano.function([x], xf)
fr1 = theano.function([x], xr1)
fr2 = theano.function([x], xr2)

ff(np.zeros((81, 243, 16)).astype('float32'))
fr1(np.zeros((81, 243, 16)).astype('float32'))
fr2(np.zeros((81, 243, 16)).astype('float32'))

gistfile2.txt

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-8-fcf5c7edf3ee> in <module>()
      1 ff(np.zeros((81, 243, 16)).astype('float32'))
      2 fr1(np.zeros((81, 243, 16)).astype('float32'))
----> 3 fr2(np.zeros((81, 243, 16)).astype('float32'))

/uhome/jbayer/devel/Theano/theano/compile/function_module.pyc in __call__(self, *args, **kwargs)
    586                     # For the CVM
    587                     gof.vm.raise_with_op(self.fn.nodes[self.fn.position_of_error],
--> 588                                          self.fn.thunks[self.fn.position_of_error])
    589                 else:
    590                     # For the c linker

/uhome/jbayer/devel/Theano/theano/compile/function_module.pyc in __call__(self, *args, **kwargs)
    577         t0_fn = time.time()
    578         try:
--> 579             outputs = self.fn()
    580         except Exception:
    581             if hasattr(self.fn, 'position_of_error'):

/uhome/jbayer/devel/Theano/theano/gof/op.pyc in rval(p, i, o, n)
    642 
    643         def rval(p=p, i=node_input_storage, o=node_output_storage, n=node):
--> 644             r = p(n, [x[0] for x in i], o)
    645             for o in node.outputs:
    646                 compute_map[o][0] = True

/uhome/jbayer/devel/Theano/theano/sandbox/cuda/basic_ops.py in perform(self, node, inp, out_)
   2202                              ' has incorrect length %i'
   2203                              ', should be %i' % (len(shp), self.ndim), shp)
-> 2204         out[0] = x.reshape(tuple(shp))
   2205 
   2206 

ValueError: size must remain unchanged, changed from 314928 to -16
Apply node that caused the error: GpuReshape{2}(GpuFromHost.0, MakeVector.0)
Inputs types: [CudaNdarrayType(float32, 3D), TensorType(int64, vector)]
Inputs shapes: [(81, 243, 16), (2,)]
Inputs strides: [(3888, 16, 1), (8,)]
Inputs scalar values: ['not scalar', 'not scalar']

HINT: Re-running with most Theano optimization disabled could give you a back-traces when this node was created. This can be done with by setting the Theano flags optimizer=fast_compile
HINT: Use the Theano flag 'exception_verbosity=high' for a debugprint of this apply node.