[Mpi-forum] cancel in standard

Sun Jun 7 18:23:44 CDT 2015

I think we should avoid MPI-<next>, at least for now. We tried to phase
this out when we decided on MPI 3.1 and I actually thought we removed it,
but I guess not. For now, at least IMHO, it would be good to keep every
active ticket either in MPI 3.1 errata or MPI 4.0. This will make it much
easier if we actually go through with the conversion to git and pull
requests. 

Martin

________________________________________________________________________
Martin Schulz, schulzm at llnl.gov, http://scalability.llnl.gov/
CASC @ Lawrence Livermore National Laboratory, Livermore, USA

On 6/5/15, 10:22 AM, "Jeff Squyres (jsquyres)" <jsquyres at cisco.com> wrote:

>Right now, there is no concept of "MPI-3.2" -- there's only MPI-3.1
>errata and MPI-4.0.
>
>
>> On Jun 5, 2015, at 12:47 PM, Balaji, Pavan <balaji at anl.gov> wrote:
>> 
>> 
>> I've changed the latter two to "MPI-next".
>> 
>> --
>> Pavan Balaji
>> http://www.mcs.anl.gov/~balaji
>> 
>> 
>> 
>> 
>> 
>> 
>> 
>> On 6/5/15, 11:43 AM, "Bland, Wesley" <wesley.bland at intel.com> wrote:
>> 
>>> The version information should probably be updated for the tickets
>>>that you are targeting for MPI 3.2 and MPI 4.0. Right now, they¹re all
>>>MPI 3.1 errata.
>>> 
>>> 
>>> 
>>> On June 5, 2015 at 11:29:58 AM, Balaji, Pavan
>>>(balaji at anl.gov<mailto:balaji at anl.gov>) wrote:
>>> 
>>> FYI, here are the three tickets mentioned below. Feedback is
>>>appreciated. I'll pass this through the pt2pt working group, but I'm
>>>looking for general feedback at this point.
>>> 
>>> https://svn.mpi-forum.org/trac/mpi-forum-web/ticket/478
>>> 
>>> https://svn.mpi-forum.org/trac/mpi-forum-web/ticket/479
>>> 
>>> https://svn.mpi-forum.org/trac/mpi-forum-web/ticket/480
>>> 
>>> 
>>> -- Pavan
>>> 
>>> --
>>> Pavan Balaji
>>> http://www.mcs.anl.gov/~balaji
>>> 
>>> 
>>> 
>>> 
>>> 
>>> 
>>> 
>>> On 6/5/15, 7:08 AM, "mpi-forum on behalf of Balaji, Pavan"
>>><mpi-forum-bounces at lists.mpi-forum.org on behalf of balaji at anl.gov>
>>>wrote:
>>> 
>>>> 
>>>> One of the main reasons cancel send was introduced was for symmetry,
>>>>since there were cancel receives too. But that argument has gone out
>>>>the window with NBC, RMA and I/O requests which are all not cancelable.
>>>> 
>>>> For everyone's information, we are working on three tickets.
>>>> 
>>>> 1. An MPI-3.1 errata ticket to clarify the wording in the MPI
>>>>standard as to what cancel send must do. This will also remove some
>>>>incorrect wording in the standard with respect to the behavior of
>>>>wait/test after a failed cancelation.
>>>> 
>>>> 2. An MPI-3.2 ticket to deprecate send cancel citing all the
>>>>complexities in implementing it efficiently. More importantly the cost
>>>>of supporting send cancel on non-cancel operations.
>>>> 
>>>> 3. An MPI-4.0 ticket to remove send cancel.
>>>> 
>>>> -- Pavan
>>>> 
>>>> Sent from my iPhone
>>>> 
>>>>> On Jun 5, 2015, at 3:16 AM, Supalov, Alexander
>>>>><alexander.supalov at intel.com> wrote:
>>>>> br/>>> Just a note: MPI__Cancel would be next to MPI_ANY SOURCE to
>>>>>go had one introduced assertions. :)
>>>>> br/>>> -----Original Message----- <
>>>>> From: mpi-forum [mailto:mpi-forum-bounces at lists.mpi-forum.org] On
>>>>>Behalf Of William Gropp
>>>>> Sent: Thursday, June 4, 2015 2:24 PM
>>>>> To: Main MPI Forum mailing list
>>>>> Subject: Re: [Mpi-forum] cancel in standard
>>>>> br/>>> But the user did care about it, and theey decided to cancel
>>>>>it. It may not be one message, but one per iteration. And the
>>>>>semantics is clear (even if the standard's language isn't).
>>>>> br/>>> Bill <
>>>>> br/>>>> On Jun 3, 2015, at 10:08 PM, Jeff Hammond
>>>>><jeff.science at gmail.com> wrote:
>>>>>> br/>>>> What's the harm in letting a meessage that the user clearly
>>>>>>does not br/>>>> care about sitt in a queue/buffer forever? I
>>>>>>consider this an br/>>>> aacceptable consequence of a really stupid
>>>>>>feature.
>>>>>> br/>>>> Jeff <
>>>>>> br/>>>>> On Wed, Jun 3, 2015 at 9:550 PM, William Gropp
>>>>>><wgropp at illinois.edu> wrote:
>>>>>>> Pavan is correct about the intent, and we could make this more
>>>>>>>precise. The basic requirement is that a message is *guaranteed* to
>>>>>>>either be canceled (on send) or received. Sitting in a queue or
>>>>>>>buffer somewhere does not count.
>>>>>>> br/>>>>> Bill <
>>>>>>> br/>>>>>> On Jun 3, 2015, att 7:19 PM, Balaji, Pavan
>>>>>>><balaji at anl.gov> wrote:
>>>>>>>> br/>>>>>> br/>>>>>> [[Cc'ed the Forum]
>>>>>>>> br/>>>>>> This wording haas been debated many times at the Forum.
>>>>>>>>Specifically, the "completes normally" and "received at the
>>>>>>>>destination" pieces are ambiguous. For instance, if the data is
>>>>>>>>copied into a bounce buffer, the send would complete normally, but
>>>>>>>>that does not mean that there is a receive posted for that data.
>>>>>>>>In this case, it would be the user's responsibility to post a
>>>>>>>>corresponding receive. If the MPI implementation does not have a
>>>>>>>>bounce buffer (e.g., for large messages), "received at the
>>>>>>>>destination" could mean into an unexpected buffer, not necessarily
>>>>>>>>a user buffer.
>>>>>>>> br/>>>>>> The consensus ffrom MPI-1 veterans (Marc, Bill, etc.)
>>>>>>>>during the last discussion was that if the application does not
>>>>>>>>post a receive, then it is guaranteed that the send should be
>>>>>>>>cancelable. At least that was the intent. The wording, of course,
>>>>>>>>is too vague for such a definition.
>>>>>>>> br/>>>>>> You could try tto weaken that interpretation and say
>>>>>>>>that if the MPI implementation is not able to cancel a send
>>>>>>>>operation, then the user is responsible for posting a matching
>>>>>>>>receive to receive that data. Irrespective of which interpretation
>>>>>>>>we go with, I think it is important to clarify it in the standard.
>>>>>>>>Perhaps an MPI-3.1 errata or MPI-3.2 change instead of waiting for
>>>>>>>>MPI-4? Do you want to create a ticket?
>>>>>>>> br/>>>>>> -- Pavan <
>>>>>>>> br/>>>>>> br/>>>>>> br/>>>>>> br/>>>>>>> On 6/3/15, 7:05 PM,
>>>>>>>>""Sur, Sayantan" <sayantan.sur at intel.com> wrote:
>>>>>>>>> br/>>>>>>> What liines in the standard support the
>>>>>>>>>interpretation that "a br/>>>t;>>>> message that will not be
>>>>>>>>>successfully matched, must be cancelable"?
>>>>>>>>> br/>>>>>>> This iss what I found, and it doesn't support the
>>>>>>>>>stricter interpretation.
>>>>>>>>> br/>>>>>>> br/>>>>>>> br/>>>>>>> br/>>>>>>> br/>>>>>>>
>>>>>>>>>br/>>>>>>> br/>>>>>>> br/>>>>>>> br/>>>>>>> ""If a
>>>>>>>>> send is marked for cancellation, then it must be the case that
>>>>>>>>>br/>>>>>>> either the send ccompletes normally, in which case the
>>>>>>>>>message sent br/>>>>&ggt;>> was received at the destination
>>>>>>>>>process, or that the send is br/>>>>>>> successfully cancelled,
>>>>>>>>>in which case noo part of the message was br/>>>>>>> received at
>>>>>>>>>thhe destination. Then, any matching receive has to be
>>>>>>>>>br/>>>>;>>> satisfied by another send."
>>>>>>>> _______________________________________________
>>>>>>>> mpi-forum mailing list
>>>>>>>> mpi-forum at lists.mpi-forum.org
>>>>>>>> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum
>>>>>>> br/>>>>> ________________________________________________
>>>>>>> mpi-forum mailing list
>>>>>>> mpi-forum at lists.mpi-forum.org
>>>>>>> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum
>>>>>> br/>>>> br/>>>> br/>>>> -- <
>>>>>> Jeff Hammond
>>>>>> jeff.science at gmail.com
>>>>>> http://jeffhammond.github.io/
>>>>>> _______________________________________________
>>>>>> mpi-forum mailing list
>>>>>> mpi-forum at lists.mpi-forum.org
>>>>>> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum
>>>>> br/>>> ________________________________________________
>>>>> mpi-forum mailing list
>>>>> mpi-forum at lists.mpi-forum.org
>>>>> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum
>>>>> Intel GmbH
>>>>> Dornacher Strasse 1
>>>>> 85622 Feldkirchen/Muenchen, Deutschland
>>>>> Sitz der Gesellschaft: Feldkirchen bei Muenchen
>>>>> Geschaeftsfuehrer: Christian Lamprechter, Hannes Schwaderer, Douglas
>>>>>Lusk
>>>>> Registergericht: Muenchen HRB 47456
>>>>> Ust.-IdNr./VAT Registration No.: DE129385895
>>>>> Citibank Frankfurt a.M. (BLZ 502 109 00) 600119052
>>>>> br/>>> ________________________________________________
>>>>> mpi-forum mailing list
>>>>> mpi-forum at lists.mpi-forum.org
>>>>> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum
>>>> _______________________________________________
>>>> mpi-forum mailing list
>>>> mpi-forum at lists.mpi-forum.org
>>>> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum
>>> _______________________________________________
>>> mpi-forum mailing list
>>> mpi-forum at lists.mpi-forum.org
>>> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum
>> _______________________________________________
>> mpi-forum mailing list
>> mpi-forum at lists.mpi-forum.org
>> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum
>
>
>-- 
>Jeff Squyres
>jsquyres at cisco.com
>For corporate legal information go to:
>http://www.cisco.com/web/about/doing_business/legal/cri/
>
>_______________________________________________
>mpi-forum mailing list
>mpi-forum at lists.mpi-forum.org
>http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum