Merge pull request #4492 from nouiz/fix_crash_gpufft

doc/library/compile/nanguardmode.txt

@@ -63,6 +63,11 @@ kind of abnormal values to monitor. ``nan_is_error`` and ``inf_is_error`` has
 no default values, so they need to be set explicitly, but ``big_is_error`` is
 set to be ``True`` by default.
 
+.. note::
+
+        NanGuardMode significantly slows down computations; only
+        enable as needed.
+
 Reference
 =========
 

doc/tutorial/examples.txt

@@ -545,8 +545,8 @@ It will be used repeatedly.
     training_steps = 10000
 
     # Declare Theano symbolic variables
-    x = T.matrix("x")
-    y = T.vector("y")
+    x = T.dmatrix("x")
+    y = T.dvector("y")
 
     # initialize the weight vector w randomly
     #

theano/sandbox/cuda/opt.py

@@ -1425,8 +1425,9 @@ def _gpu_conv_to_fftconv(node):
     # we import conv2d_fft locally to avoid pycuda warnings
     from theano.sandbox.cuda.fftconv import conv2d_fft
     kwargs = {'border_mode': node.op.border_mode}
-    if (node.op.imshp is not None and node.op.imshp[-1] % 2 == 1 and
-            node.op.imshp[-1] is not None):
+    if (node.op.imshp is not None and
+            node.op.imshp[-1] is not None and
+            node.op.imshp[-1] % 2 == 1):
 
         kwargs['pad_last_dim'] = True
     # If the user supplied the full nonsymbolic image_shape and
@@ -1455,8 +1456,9 @@ def _gpu_conv_to_fftconv(node):
 @local_optimizer([GpuConv])
 def local_conv_fft_valid(node):
     if isinstance(node.op, GpuConv):
-        if (node.op.border_mode == 'valid' and node.op.fft_opt and
-                node.op.subsample == (1, 1)):
+        if (node.op.border_mode == 'valid' and
+                node.op.subsample == (1, 1) and
+                node.op.fft_opt):
 
             return [_gpu_conv_to_fftconv(node)]
         return False
@@ -1465,8 +1467,9 @@ def local_conv_fft_valid(node):
 @local_optimizer([GpuConv])
 def local_conv_fft_full(node):
     if isinstance(node.op, GpuConv):
-        if (node.op.border_mode == 'full' and node.op.fft_opt and
-                node.op.subsample == (1, 1)):
+        if (node.op.border_mode == 'full' and
+                node.op.subsample == (1, 1) and
+                node.op.fft_opt):
 
             return [_gpu_conv_to_fftconv(node)]
         return