make sure that in speed tests for RNN everything runs with CVM.

I wanted to make sure that I get the right timings. It seems that CVM is faster then what James suggested for now on CPU, and slower on GPU(but can be fixed). Note that a full implementation of what James suggested will be slower then what this text predicts ( this is as fast as it will ever get ), namely scan needs to do a bit more bookeeping that the cython right now does.

make sure that in speed tests for RNN everything runs with CVM.
83c76420 · Razvan Pascanu · 1f00a3ca · 83c76420
--- a/theano/scan_module/tests/test_scan.py
+++ b/theano/scan_module/tests/test_scan.py
@@ -2460,7 +2460,8 @@ def test_speed():
        s_r = tensor.matrix()
        s_y, updates = theano.scan(fn=lambda ri, rii:ri+rii,
                sequences=[s_r[1:]],
-                outputs_info=tensor.constant(r[0]))
+                outputs_info=tensor.constant(r[0]),
+                mode = theano.Mode(linker='cvm')  )
        assert not updates
        f = theano.function([s_r], s_y)
@@ -2530,9 +2531,10 @@ def test_speed_rnn():
        s_r = tensor.matrix()
        s_y, updates = theano.scan(fn=lambda ri, rii:tensor.tanh(tensor.dot(rii, w)),
                sequences=[s_r[1:]],
-                outputs_info=tensor.constant(r[0]))
+                outputs_info=tensor.constant(r[0]),
+                    mode = theano.Mode(linker='cvm'))
        assert not updates
-        f = theano.function([s_r], s_y, mode=theano.Mode(linker='c|py_nogc'))
+        f = theano.function([s_r], s_y, mode=theano.Mode(linker='cvm'))
        t2 = time.time()
        f(r)