Update: Addressing feedback from WG - Oct 26

content/nb_collectives_intro.tex

@@ -1,29 +1,29 @@
-An OpenSHMEM nonblocking collective operation, like blocking collective
+An \openshmem nonblocking collective operation, like blocking collective
 operation, is a group communication operation among the
 participants of the team. All participants of the team are required to call the
 collective operation.
 
 \begin{enumerate}
 
-\item Invocation semantics: The non-blocking collective routine initializes the
-buffers, operation type, reduction type, and posts the collective operation. All
-participants of the team should call this routine. The routine returns
-immediately after posting the operation. 
+\item Invocation semantics: Upon invocation of a collective routine interface,
+the operation is posted and returned immediately. All participants of the Team
+should call this routine.
 
-\item Collective Types: Currently, only the nonblocking alltoall, broadcast, and reduction collective
-operations are supported. The reduction operations supported are defined in the
-Table \ref{reducetypes}. 
+\item Collective Types: In the current specification, not all blocking collectives have
+their nonblocking variants. The nonblocking variants supported include alltoall,
+broadcast, and reduction collectives. The reduction types supported
+are defined in the Table \ref{reducetypes}. 
 
-\item Completion semantics: Upon invocation, the collective operations are
-posted and returns immediately. A user can learn the status of the collective operations
-using the \FUNC{shmem\_collective\_test} routine and can be completed using
-the \FUNC{shmem\_collective\_wait} routine.
+\item Completion semantics:  \openshmem programs can learn the status of the collective operations
+using the \FUNC{shmem\_req\_test} routine and can be completed using
+the \FUNC{shmem\_req\_wait} routine.
 
 \item Threads: While using SHMEM\_THREAD\_MULTIPLE, the \openshmem
 programs are allowed to call multiple collective operations on different threads
 and the same Team. The collective operations invoked on different threads
-are ordered by user-provided tag. When the user does not provide the tag, the
-library generates the tag and establishes the order.
+are ordered by user-provided tag. The user may choose to not order the
+collective operations by using the library constant
+\CONST{SHMEM\_COLL\_UNORDERED} instead of specifying the tag.
 
 \end{enumerate}
 

content/shmem_broadcast_nb.tex

@@ -8,7 +8,7 @@
 %% C11
 \begin{C11synopsis}
 int @\FuncDecl{shmem\_broadcast\_nb}@(shmem_team_t team, TYPE *dest, const TYPE
-*source, size_t nelems, int PE_root,uint32_t tag, shmem_req_h *request);
+*source, size_t nelems, int PE_root, uint32_t tag, shmem_req_h *request);
 \end{C11synopsis}
 where \TYPE{} is one of the standard \ac{RMA} types specified by Table \ref{stdrmatypes}.
 
@@ -59,9 +59,16 @@
     \VAR{PE\_root} must be passed by all \acp{PE} participating in the
     collective operation.
 
-    A call to the nonblocking broadcast routine returns immediately without
-    necessarily completing the operation. The operation is completed after a
-    call to \FUNC{shmem\_collective\_test} or \FUNC{shmem\_collective\_wait}.
+    A call to the nonblocking broadcast routine posts the operation and returns
+    immediately without necessarily completing the operation. On the successful
+    post of the operation, an opaque request handle is created and returned. The
+    operation is completed after a call to \FUNC{shmem\_req\_test} or
+    \FUNC{shmem\_req\_wait}. When the operation is complete, the request handle
+    is deallocated and cannot be reused.
+
+    The nonblocking broadcast routine is ordered according to the user-defined tag.
+    If \CONST{SHMEM\_COLL\_UNORDERED} is used instead of a tag, the collective
+    operations are not ordered.
 
     Like blocking broadcast, before any \ac{PE} calls a broadcast routine, the following
     conditions must be ensured:

content/shmem_collective_test.tex

@@ -5,31 +5,30 @@
 \begin{apidefinition}
 
 \begin{Csynopsis}
-int @\FuncDecl{shmem\_collective\_test}@(shmem_req_h request, int *flag);
+int @\FuncDecl{shmem\_req\_test}@(shmem_req_t request);
 \end{Csynopsis}
 
 \begin{apiarguments}
 
   \apiargument{IN}{request}{Request representing a outstanding collective}
-  \apiargument{IN}{flag}{Variable indicating the status of the request}
 
 \end{apiarguments}
 
 \apidescription{
-    A call to \FUNC{shmem\_collective\_test} returns immediately. If the
+    A call to \FUNC{shmem\_req\_test} returns immediately. If the
     collective operation identified by the request is completed, it returns
-    true. The request object is deallocated. If the collective operation is not
-    completed, it returns false.
+    true (non-negative integer). The request object is deallocated. If the collective operation is not
+    completed, it returns zero. 
 
     In a multithreaded environment, the collective and the
-    \FUNC{shmem\_collective\_test} can be
+    \FUNC{shmem\_req\_test} can be
     called by different threads. It is the responsibility of the \openshmem user
     to ensure that the \FUNC{shmem\_collective\_test} operation is called after the
     collective operation.
     }
 
 \apireturnvalues{
-    On success returns zero, otherwise returns a negative integer.
+    On success returns zero or one, otherwise returns a negative integer.
     }
 
 \end{apidefinition}

main_spec.tex

@@ -380,6 +380,7 @@ \subsubsection{\textbf{SHMEM\_COLLECT, SHMEM\_FCOLLECT}}\label{subsec:shmem_coll
 \subsubsection{\textbf{SHMEM\_REDUCTIONS}}\label{subsec:shmem_reductions}
 \input{content/shmem_reductions.tex}
 
+\newpage 
 \subsection{Nonblocking Collective Routines}\label{subsec:nb_coll}
 \input{content/nb_collectives_intro.tex}