fix guess/time workmem bug

theano/sandbox/cuda/dnn.py

@@ -713,11 +713,11 @@ class GpuDnnConv3d(GpuDnnConv):
 
     def __init__(self, workmem=None, inplace=False):
         """
-        :param workmem: either 'none', 'small', 'large', 'fft', 'time' or
-        'guess'. Default is the value of :attr:`config.dnn.conv.workmem`.
+        :param workmem: either 'none' 'time' or 'guess'.
+        Default is the value of :attr:`config.dnn.conv.workmem`.
         """
-        ### Only workmem = 'none' work with cudnn conv 3d
-        super(GpuDnnConv3d, self).__init__(workmem='none', inplace=inplace)
+        super(GpuDnnConv3d, self).__init__(workmem='guess', inplace=inplace)
+        assert self.workmem in ['none' 'time','guess']
 
     def make_node(self, img, kern, output, desc, alpha=None, beta=None, nb_dim=None):
 

theano/sandbox/cuda/dnn_fwd.c

@@ -36,17 +36,17 @@ APPLY_SPECIFIC(conv_fwd)(CudaNdarray *input, CudaNdarray *kerns,
     void *workspace;
     cudnnConvolutionFwdAlgo_t chosen_algo;
 
+
     if (CHOOSE_ALGO)
     {
-
       // Check if the input and the kernels have the same shape as they have
       // last time the apply node was executed
       bool same_shapes = true;
       for (int i = 0; (i < nb_dim) && same_shapes; i++)
       {
-          same_shapes &= (CudaNdarray_HOST_DIMS(input)[i] !=
+          same_shapes &= (CudaNdarray_HOST_DIMS(input)[i] ==
                           APPLY_SPECIFIC(previous_input_shape)[i]);
-          same_shapes &= (CudaNdarray_HOST_DIMS(kerns)[i] !=
+          same_shapes &= (CudaNdarray_HOST_DIMS(kerns)[i] ==
                           APPLY_SPECIFIC(previous_kerns_shape)[i]);
       }
 
@@ -67,6 +67,7 @@ APPLY_SPECIFIC(conv_fwd)(CudaNdarray *input, CudaNdarray *kerns,
           return 1;
         }
 
+
         // Obtain a convolution algorithm appropriate for the input and kernel
         // shapes. Either by choosing one according to heuristics or by making
         // CuDNN time every implementation and choose the best one.
@@ -131,7 +132,6 @@ APPLY_SPECIFIC(conv_fwd)(CudaNdarray *input, CudaNdarray *kerns,
           // be used here
           chosen_algo = APPLY_SPECIFIC(previous_algo);
       }
-
     }
     else
     {
@@ -179,7 +179,6 @@ APPLY_SPECIFIC(conv_fwd)(CudaNdarray *input, CudaNdarray *kerns,
       }
     }
 
-
     err = cudnnGetConvolutionForwardWorkspaceSize(_handle,
                                                   APPLY_SPECIFIC(input),
                                                   APPLY_SPECIFIC(kerns),
@@ -188,7 +187,6 @@ APPLY_SPECIFIC(conv_fwd)(CudaNdarray *input, CudaNdarray *kerns,
                                                   chosen_algo,
                                                   &worksize);
     if (err != CUDNN_STATUS_SUCCESS) {
-      std::cout << "here" << std::endl;
       PyErr_Format(PyExc_RuntimeError,
                    "GpuDnnConv: error getting worksize: %s",
                    cudnnGetErrorString(err));

theano/sandbox/cuda/dnn_gw.c

@@ -12,11 +12,6 @@ APPLY_SPECIFIC(conv_gw)(CudaNdarray *input, CudaNdarray *output,
     return 1;
   }
 
-  /* if (c_set_tensor4d(input, APPLY_SPECIFIC(input)) == -1) */
-  /*   return 1; */
-  /* if (c_set_tensor4d(output, APPLY_SPECIFIC(output)) == -1) */
-  /*   return 1; */
-
   if (c_set_tensorNd(input, nb_dim, APPLY_SPECIFIC(input)) == -1)
     return 1;
   if (c_set_tensorNd(output, nb_dim, APPLY_SPECIFIC(output)) == -1)
@@ -33,8 +28,6 @@ APPLY_SPECIFIC(conv_gw)(CudaNdarray *input, CudaNdarray *output,
     return 1;
 #endif
 
-  /* if (c_set_filter(*kerns, APPLY_SPECIFIC(kerns)) == -1) */
-  /*   return 1; */
   if (c_set_filterNd(*kerns, nb_dim, APPLY_SPECIFIC(kerns)) == -1)
     return 1;
 
@@ -50,9 +43,9 @@ APPLY_SPECIFIC(conv_gw)(CudaNdarray *input, CudaNdarray *output,
       bool same_shapes = true;
       for (int i = 0; (i < nb_dim) && same_shapes; i++)
       {
-          same_shapes &= (CudaNdarray_HOST_DIMS(input)[i] !=
+          same_shapes &= (CudaNdarray_HOST_DIMS(input)[i] ==
                           APPLY_SPECIFIC(previous_input_shape)[i]);
-          same_shapes &= (CudaNdarray_HOST_DIMS(output)[i] !=
+          same_shapes &= (CudaNdarray_HOST_DIMS(output)[i] ==
                           APPLY_SPECIFIC(previous_output_shape)[i]);
       }