hack to lower the number of compilation on the GPU.

theano/gof/cc.py

@@ -342,8 +342,18 @@ def get_c_extract(r, name, sub):
     """Wrapper around c_extract that initializes py_name from storage."""
     if any([getattr(c.op, 'check_input', config.check_input) for (c, _) in
             r.clients]):
-
-        c_extract = r.type.c_extract(name, sub, True)
+        # check_broadcast is just an hack to easily remove just the
+        # broadcast check on the old GPU back-end. THis check isn't
+        # done in the new GPU back-end or on the CPU.
+        if hasattr(c.op, 'check_broadcast'):
+            try:
+                c_extract = r.type.c_extract(
+                    name, sub, True,
+                    check_broadcast=c.op.check_broadcast)
+            except TypeError, e:
+                c_extract = r.type.c_extract(name, sub, True)
+        else:
+            c_extract = r.type.c_extract(name, sub, True)
     else:
         c_extract = r.type.c_extract(name, sub, False)
 

theano/sandbox/cuda/dnn.py

@@ -244,6 +244,7 @@ class GpuDnnConvDesc(GpuOp):
 
 class GpuDnnConvBase(DnnBase):
     __props__ = ()
+    check_broadcast = False
 
     def c_support_code_struct(self, node, struct_id):
         return """

theano/sandbox/cuda/type.py

@@ -280,7 +280,8 @@ class CudaNdarrayType(Type):
     def c_init(self, name, sub):
         return "%(name)s = NULL;" % locals()
 
-    def c_extract(self, name, sub, check_input=True):
+    def c_extract(self, name, sub, check_input=True,
+                  check_broadcast=True):
         sio = StringIO()
         fail = sub['fail']
         nd = self.ndim
@@ -307,7 +308,7 @@ class CudaNdarrayType(Type):
                 //std::cerr << "c_extract " << %(name)s << " nd check passed\\n";
             """ % locals()
             for i, b in enumerate(self.broadcastable):
-                if b:
+                if b and check_broadcast:
                     print >> sio, """
                 if (CudaNdarray_HOST_DIMS(%(name)s)[%(i)s] != 1)
                 {