Improved passage on immediates and their undefined behaviour. (pytorch#152)

* Added distributed tutorial.

* Updated undefined behaviour when using immediates

Author: seba-1511

intermediate_source/dist_tuto.rst

@@ -138,12 +138,18 @@ return a ``DistributedRequest`` object upon which we can choose to
             # Receive tensor from process 0
             req = dist.irecv(tensor=tensor, src=0)
             print('Rank 1 started receiving')
-            print('Rank 1 has data ', tensor[0])
         req.wait()
         print('Rank ', rank, ' has data ', tensor[0])
 
-Running the above function might result in process 1 still having 0.0
-while having already started receiving. However, after ``req.wait()``
+When using immediates we have to be careful about with our usage of the sent and received tensors.
+Since we do not know when the data will be communicated to the other process,
+we should not modify the sent tensor nor access the received tensor before ``req.wait()`` has completed.
+In other words, 
+
+-  writing to ``tensor`` after ``dist.isend()`` will result in undefined behaviour.
+-  reading from ``tensor`` after ``dist.irecv()`` will result in undefined behaviour. 
+
+However, after ``req.wait()``
 has been executed we are guaranteed that the communication took place,
 and that the value stored in ``tensor[0]`` is 1.0.