More doc about the new Async feature.

doc/tutorial/using_gpu.txt

@@ -617,3 +617,9 @@ In both case, you can use the NVIDIA driver feature that when
 environment variable CUDA_LAUNCH_BLOCKING=1 is set, all kernal call
 get automatically syncronized. This will restore to the old beavior
 that provide good profiling and error message.
+
+This feature interact with Theano garbage collector of intermediate
+results. To get the most of this feature, you need to disable the gc
+as it insert synchronization point in the graph. Set the Theano flag
+allow_gc=False to get event faster speed! This will raise the memory
+usage.