Newsgroups: comp.parallel.pvm
From: papadopo@cs.utk.edu (Philip Papadopoulos)
Subject: Re: pvm message passing timing
Organization: CS Department, University of Tennessee, Knoxville
Date: 11 Aug 1995 08:52:27 -0400
Message-ID: <40fjqbINNbfe@duncan.cs.utk.edu>

In article <40f30l$1fho@unixfe.rl.ac.uk> rff@inf.rl.ac.uk writes:
[previous stuff deleted]
>
>If RouteDirect is used it seems quite possible to get blocking in the
>send if the receiver is busy.
>
>Ron Fowler
This is correct. On direct-connected tasks on workstations, for example,
a TCP/IP socket is used. If the sender (writer on the socket) sends more
information than the protocol can buffer, then the sender is blocked until
the socket has been sufficiently drained by the receiver (reader). 
This can have the effect that the sender is blocked until the receiver
actually gets to a receive that would drain the socket.  This may not
be exactly what you wanted, but is consisentent with the notion that 
when the "send returns, the send buffer is safe for re-use."


-Phil



