<html><head></head><body style="word-wrap: break-word; -webkit-nbsp-mode: space; -webkit-line-break: after-white-space; ">An implementation is free to use separate queues for each communicator; some of us have discussed this in the past, in part to permit use of lock-free structures for the queue updates, particularly as this is the only place there are no wild cards, ever. I believe that this is within the existing semantics. It even has benefits for single threaded execution, since the communicator matching is done once, rather than in every query on the queue.<div><br></div><div>In terms of progress, the standard is deliberately vague on the details, and thus I don't believe we have the requirement that you quote. And some of the other interpretations of progress would not be helped by any thread-safety restriction.</div><div><br></div><div>Bill</div><div><br><div>
<span class="Apple-style-span" style="border-collapse: separate; color: rgb(0, 0, 0); font-family: Helvetica; font-style: normal; font-variant: normal; font-weight: normal; letter-spacing: normal; line-height: normal; orphans: 2; text-align: -webkit-auto; text-indent: 0px; text-transform: none; white-space: normal; widows: 2; word-spacing: 0px; -webkit-border-horizontal-spacing: 0px; -webkit-border-vertical-spacing: 0px; -webkit-text-decorations-in-effect: none; -webkit-text-size-adjust: auto; -webkit-text-stroke-width: 0px; font-size: medium; "><span class="Apple-style-span" style="border-collapse: separate; color: rgb(0, 0, 0); font-family: Helvetica; font-style: normal; font-variant: normal; font-weight: normal; letter-spacing: normal; line-height: normal; orphans: 2; text-indent: 0px; text-transform: none; white-space: normal; widows: 2; word-spacing: 0px; -webkit-border-horizontal-spacing: 0px; -webkit-border-vertical-spacing: 0px; -webkit-text-decorations-in-effect: none; -webkit-text-size-adjust: auto; -webkit-text-stroke-width: 0px; font-size: medium; "><div style="word-wrap: break-word; -webkit-nbsp-mode: space; -webkit-line-break: after-white-space; "><div><div style="font-size: 12px; ">William Gropp</div><div style="font-size: 12px; ">Director, Parallel Computing Institute</div><div style="font-size: 12px; ">Deputy Director for Research</div><div style="font-size: 12px; ">Institute for Advanced Computing Applications and Technologies</div></div></div></span><span class="Apple-style-span" style="font-size: 12px; ">Thomas M. Siebel Chair in Computer Science</span><span class="Apple-style-span" style="border-collapse: separate; color: rgb(0, 0, 0); font-family: Helvetica; font-style: normal; font-variant: normal; font-weight: normal; letter-spacing: normal; line-height: normal; orphans: 2; text-indent: 0px; text-transform: none; white-space: normal; widows: 2; word-spacing: 0px; -webkit-border-horizontal-spacing: 0px; -webkit-border-vertical-spacing: 0px; -webkit-text-decorations-in-effect: none; -webkit-text-size-adjust: auto; -webkit-text-stroke-width: 0px; font-size: medium; "><div style="word-wrap: break-word; -webkit-nbsp-mode: space; -webkit-line-break: after-white-space; "><div><div style="font-size: 12px; ">University of Illinois Urbana-Champaign</div></div><div><br></div></div></span><br class="Apple-interchange-newline"></span><br class="Apple-interchange-newline">
</div>
<br><div><div>On Mar 25, 2013, at 4:15 PM, Jeff Hammond wrote:</div><br class="Apple-interchange-newline"><blockquote type="cite"><div>On Mon, Mar 25, 2013 at 3:17 PM, William Gropp <<a href="mailto:wgropp@illinois.edu">wgropp@illinois.edu</a>> wrote:<br><blockquote type="cite">I was only addressing the issue of calling the thread level routines before<br></blockquote><blockquote type="cite">knowing what thread level you had.<br></blockquote><br>Okay, sorry, I cannot tell which tickets people are referring to since<br>I have a bunch of different ones right now.<br><br><blockquote type="cite">I'm not sure what you are looking for. In the case of MPI_THREAD_MULTIPLE,<br></blockquote><blockquote type="cite">an implementation can provide significant concurrency today without any<br></blockquote><blockquote type="cite">change in the MPI standard - that's a major reason for that table (more to<br></blockquote><blockquote type="cite">the point - this table is meant as a guide for not using locks). Can you<br></blockquote><blockquote type="cite">give me an example of something that the current MPI semantics prohibits<br></blockquote><blockquote type="cite">that you'd like to achieve with MPI_THREAD_PER_OBJECT?<br></blockquote><br>It is my understanding of the progress requirements that any call to<br>MPI must make progress on all MPI operations. This means that two<br>threads calling e.g. MPI_Recv must walk all of the message queues. If<br>a thread needs to modify any queue because it matches, then this must<br>be done in a thread-safe way, which presumably requires something<br>resembling mutual exclusion or transactions. If a call to MPI_Recv<br>only had to make progress on its own communicator, then two threads<br>calling MPI_Recv on two different communicators would (1) only have to<br>walk the message queue associated with that communicator and (2)<br>nothing resembling mutual exclusion is required for the thread to<br>update the message queue in the event that matching occurs.<br><br>Forgive me if I've got some of the details wrong. If I've got all of<br>the details and the big picture wrong, then I'll think about it more.<br><br>Jeff<br><br><blockquote type="cite">On Mar 25, 2013, at 2:53 PM, Jeff Hammond wrote:<br></blockquote><blockquote type="cite"><br></blockquote><blockquote type="cite">That doesn't do much for me in terms of enabling greater concurrency<br></blockquote><blockquote type="cite">in performance-critical operations.<br></blockquote><blockquote type="cite"><br></blockquote><blockquote type="cite">I'd like to propose that we try to make all of "Access Only", "Update<br></blockquote><blockquote type="cite">RefCount", "Read of List" and "None" thread safe in all cases. All of<br></blockquote><blockquote type="cite">these are read-only except for "Update RefCount", but this can be done<br></blockquote><blockquote type="cite">with atomics. I am assuming that concurrent reads are only permitted<br></blockquote><blockquote type="cite">to happen after the writing calls on the object have completed. This<br></blockquote><blockquote type="cite">is the essence of MPI_THREAD_PER_OBJECT.<br></blockquote><blockquote type="cite"><br></blockquote><blockquote type="cite">Jeff<br></blockquote><blockquote type="cite"><br></blockquote><blockquote type="cite"><br></blockquote><blockquote type="cite"><br></blockquote><blockquote type="cite">_______________________________________________<br></blockquote><blockquote type="cite">Mpi3-hybridpm mailing list<br></blockquote><blockquote type="cite"><a href="mailto:Mpi3-hybridpm@lists.mpi-forum.org">Mpi3-hybridpm@lists.mpi-forum.org</a><br></blockquote><blockquote type="cite"><a href="http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-hybridpm">http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-hybridpm</a><br></blockquote><br><br><br>-- <br>Jeff Hammond<br>Argonne Leadership Computing Facility<br>University of Chicago Computation Institute<br><a href="mailto:jhammond@alcf.anl.gov">jhammond@alcf.anl.gov</a> / (630) 252-5381<br><a href="http://www.linkedin.com/in/jeffhammond">http://www.linkedin.com/in/jeffhammond</a><br>https://wiki.alcf.anl.gov/parts/index.php/User:Jhammond<br>_______________________________________________<br>Mpi3-hybridpm mailing list<br>Mpi3-hybridpm@lists.mpi-forum.org<br>http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-hybridpm<br></div></blockquote></div><br></div></body></html>