modif needed to tell by theano the version of the kernel that we want for the conv op.

theano/sandbox/conv.py

@@ -34,7 +34,8 @@ class ConvOp(Op):
             imshp_logical=None,
             kshp_logical=None,
             kshp_logical_top_aligned=True,
-            verbose=0):
+            verbose=0,
+            version=-1):
         """
         
 
@@ -48,6 +49,7 @@ class ConvOp(Op):
         unroll_batch - c code generation option
         unroll_kern - c code generation option
         verbose - passed to GpuConv
+        version - passed to GpuConv
 
         The reason that this op does the summation over convolutions within the 'stack' is that
         it allows us to be memory-efficient about how gradients are calculated.  If, for
@@ -79,6 +81,7 @@ class ConvOp(Op):
         self.dx=dx
         self.dy=dy
         self.verbose=verbose
+        self.version=version
         # a triple
         self.imshp_logical = self.imshp if imshp_logical is None else tuple(imshp_logical)
         assert len(self.imshp) == len(self.imshp_logical)