renamed xview to zview since it is a view of z, not x

theano/sandbox/cuda/basic_ops.py

@@ -2457,7 +2457,7 @@ class GpuIncSubtensor(tensor.IncSubtensor, GpuOp):
             This doesn't need to actually set up the view with the
             right indexing; we'll do that manually later.
         """
-        return """CudaNdarray* xview = (CudaNdarray*)
+        return """CudaNdarray* zview = (CudaNdarray*)
                 CudaNdarray_New(%(view_ndim)s)""" % locals()
 
     def get_helper_c_code_args(self):
@@ -2487,41 +2487,41 @@ class GpuIncSubtensor(tensor.IncSubtensor, GpuOp):
     def link_view_array(self, x, fail):
 
         return """
-        if (CudaNdarray_set_device_data(xview, CudaNdarray_DEV_DATA(%(x)s),
+        if (CudaNdarray_set_device_data(zview, CudaNdarray_DEV_DATA(%(x)s),
                                        (PyObject*) NULL))
         {
             PyErr_Format(PyExc_RuntimeError,
                          "GpuSubtensor is not able to set the"
                          " devdata field of the view");
-            Py_XDECREF(xview);
+            Py_XDECREF(zview);
             %(fail)s;
         }
-        cnda_mark_dev_structure_dirty(xview);
+        cnda_mark_dev_structure_dirty(zview);
         """ % locals()
 
     def set_view_base(self, x, fail):
         return """
         //Set the base only now
 
-        if(CudaNdarray_set_device_data(xview, CudaNdarray_DEV_DATA(xview),
+        if(CudaNdarray_set_device_data(zview, CudaNdarray_DEV_DATA(zview),
                                     %(x)s)){
             PyErr_Format(PyExc_RuntimeError,
                          "GpuSubtensor is not able to set"
                          " the base of the view array");
-            Py_XDECREF(xview);
+            Py_XDECREF(zview);
             %(fail)s;
         }""" % locals()
 
-    def add_to_xview(self, x, fail):
+    def add_to_zview(self, x, fail):
 
         return """
 
-        PyObject * add_result =  CudaNdarray_inplace_add((PyObject *) xview,
+        PyObject * add_result =  CudaNdarray_inplace_add((PyObject *) zview,
                                                          (PyObject *) py_%(x)s);
 
         if (! add_result )
         {
-            Py_DECREF(xview);
+            Py_DECREF(zview);
             %(fail)s;
         }
         else

theano/sandbox/cuda/cuda_ndarray.cu

@@ -101,33 +101,42 @@ int device_free(void *ptr)
         // it returns something else I still don't see why we should ignore
         // it.  All we want to do here is reset the flag.
         cudaGetLastError();
-#if COMPUTE_GPU_MEM_USED
-        fprintf(stderr, "Error freeing device pointer %p (%s).%d byte already allocated\n", ptr, cudaGetErrorString(err), _allocated_size);
-#else
-        fprintf(stderr, "Error freeing device pointer %p (%s).\n", ptr, cudaGetErrorString(err));
-#endif
-        PyErr_Format(PyExc_MemoryError, "error freeing device pointer %p (%s)", ptr, cudaGetErrorString(err));
+        #if COMPUTE_GPU_MEM_USED
+            fprintf(stderr,
+                    "Error freeing device pointer %p (%s).%d byte already allocated\n",
+                    ptr, cudaGetErrorString(err), _allocated_size);
+        #else
+            fprintf(stderr,
+                    "Error freeing device pointer %p (%s).\n",
+                    ptr,
+                    cudaGetErrorString(err));
+        #endif
+        PyErr_Format(PyExc_MemoryError,
+                "error freeing device pointer %p (%s)",
+                ptr,
+                cudaGetErrorString(err));
         return -1;
     }
     _outstanding_mallocs[0] -= (ptr != NULL);
-#if COMPUTE_GPU_MEM_USED
-    int i=0;
-    size_t total_freed = 0;
-    for(;i<TABLE_SIZE;i++)
-        if(_alloc_size_table[i].ptr==ptr){
-            _allocated_size -= _alloc_size_table[i].size;
-            total_freed += _alloc_size_table[i].size;
-            _alloc_size_table[i].ptr=0;
-            _alloc_size_table[i].size=0;
+    #if COMPUTE_GPU_MEM_USED
+        int i=0;
+        size_t total_freed = 0;
+        for(;i<TABLE_SIZE;i++)
+            if(_alloc_size_table[i].ptr==ptr){
+                _allocated_size -= _alloc_size_table[i].size;
+                total_freed += _alloc_size_table[i].size;
+                _alloc_size_table[i].ptr=0;
+                _alloc_size_table[i].size=0;
 
-            break;
-        }
-    //if(i==TABLE_SIZE)
-    //    printf("Unallocated unknow size!\n");
-    //fprintf(stderr, "freed %li bytes of device memory (%s). %d already allocated, ptr=%p\n", (long)total_freed, cudaGetErrorString(err),_allocated_size,ptr);
-#endif
+                break;
+            }
+        //if(i==TABLE_SIZE)
+        //    printf("Unallocated unknow size!\n");
+        //fprintf(stderr, "freed %li bytes of device memory (%s). %d already allocated, ptr=%p\n", (long)total_freed, cudaGetErrorString(err),_allocated_size,ptr);
+    #endif
     return 0;
 }
+
 static PyObject *
 outstanding_mallocs(PyObject* self, PyObject * args)
 {