Fixed caching for fallback, warnings

theano/gpuarray/c_code/dnn_conv_base.c

@@ -197,8 +197,8 @@ static std::string dnn_conv_shape(cudnnTensorDescriptor_t inputDesc, PyGpuArrayO
           (PyGpuArray_DIMS(output)[1] / groups  != expected_output_dims[1]) ||
           (PyGpuArray_DIMS(output)[2] != expected_output_dims[2]) ||
           (PyGpuArray_DIMS(output)[3] != expected_output_dims[3])) {
-        PyErr_Format(PyExc_ValueError, "impossible convolution output dim: expected %ldx%ldx%ldx%ld"
-                     " but received gradient with shape %dx%dx% dx%d",
+        PyErr_Format(PyExc_ValueError, "impossible convolution output dim: expected %dx%dx%dx%d"
+                     " but received gradient with shape %ldx%ldx%ldx%ld",
                      expected_output_dims[0], expected_output_dims[1] / groups,
                      expected_output_dims[2], expected_output_dims[3],
                      PyGpuArray_DIMS(output)[0], PyGpuArray_DIMS(output)[1],
@@ -211,7 +211,7 @@ static std::string dnn_conv_shape(cudnnTensorDescriptor_t inputDesc, PyGpuArrayO
           (PyGpuArray_DIMS(output)[2] != expected_output_dims[2]) ||
           (PyGpuArray_DIMS(output)[3] != expected_output_dims[3]) ||
           (PyGpuArray_DIMS(output)[4] != expected_output_dims[4])) {
-        PyErr_Format(PyExc_ValueError, "impossible convolution output dim: expected %ldx%ldx%ldx%ldx%ld"
+        PyErr_Format(PyExc_ValueError, "impossible convolution output dim: expected %dx%dx%dx%dx%d"
                      " but received gradient with shape %ldx%ldx%ldx%ldx%ld",
                      expected_output_dims[0], expected_output_dims[1],
                      expected_output_dims[2], expected_output_dims[3],

theano/gpuarray/c_code/dnn_fwd.c

@@ -231,8 +231,9 @@ APPLY_SPECIFIC(conv_fwd)(PyGpuArrayObject *input, PyGpuArrayObject *kerns,
     cuda_exit(c->ctx);
     return 1;
   }
-  // save worksize for next time/cache
+  // save for next time/cache
   prev_algo.wsSize = worksize;
+  prev_algo.algo = algo;
 
   // Add to the cache, even if this node use *_once algo
   // (in case the user specify the algo per layer and not globally).

theano/gpuarray/c_code/dnn_gi.c

@@ -247,8 +247,9 @@ APPLY_SPECIFIC(conv_gi)(PyGpuArrayObject *kerns, PyGpuArrayObject *output,
       cuda_exit(c->ctx);
       return 1;
     }
-    // save worksize for next time/cache
+    // save for next time/cache
     prev_algo.wsSize = worksize;
+    prev_algo.algo = algo;
 
     // Add to the cache
     if (params->choose_algo)

theano/gpuarray/c_code/dnn_gw.c

@@ -219,8 +219,9 @@ APPLY_SPECIFIC(conv_gw)(PyGpuArrayObject *input, PyGpuArrayObject *output,
         }
       }
 
-      // save worksize for next time/cache
+      // save for next time/cache
       prev_algo.wsSize = worksize;
+      prev_algo.algo = algo;
 
       // Add to the cache
       if (params->choose_algo)