June 2011 - mpi-forum - mailman3.cels.anl.gov

Re: [Mpi-forum] (Continued) Latex woes
by Jeff Squyres 10 Jun '11

10 Jun '11

Thanks! Is there an SVN permissions problem such that you can't commit? (assumedly to the /minutiae branch, since the MPIreplace* (etc.) macros are not on the /approved branch...?) On Jun 9, 2011, at 8:23 PM, William Gropp wrote: > Attached is the diff of the fix. I can't commit it, so one of you will need to. > > Bill > > <fix> > On Jun 9, 2011, at 12:20 PM, William Gropp wrote: > >> Floats lost is often due to a limitation in LaTeX. One possible issue is that the the change markers use floats, and so do tables. LaTeX isn't very happy with floats within floats. I'm in a meeting today and can't look at it until later, but one solution may be to inline the ticket info in these cases. >> >> Bill >> >> On Jun 9, 2011, at 11:30 AM, Jeff Squyres wrote: >> >>> Bill -- >>> >>> We're still having latex woes in ticket 278. Can you help? >>> >>> We replaced all the \MPIreplace macros with deletebegin/end and updatebegin/end, but we're still getting "Float(s) lost". It seems to have to do with a table in datatype.tex. The table looks like this: >>> >>> ----- >>> \begin{table}[ht] >>> \begin{center} >>> \begin{constlist} >>> \constitem{MPI\_COMBINER\_NAMED}{a named predefined datatype} >>> ...etc. >>> \MPIdeleteBegin{3.0}{278} >>> \constitem{MPI\_COMBINER\_HVECTOR\_INTEGER}{\mpifunc{MPI\_TYPE\_HVECTOR} from Fortran} >>> \constitem{MPI\_COMBINER\_HVECTOR}{\mpifunc{MPI\_TYPE\_HVECTOR} from C or C++} >>> \constskipitem{}{\ \ and in some case Fortran} >>> \constskipitem{}{\ \ or \mpifunc{MPI\_TYPE\_CREATE\_HVECTOR}} >>> \MPIdeleteEnd{3.0} >>> \MPIupdateBegin{3.0}{278} >>> \constitem{MPI\_COMBINER\_HVECTOR}{\mpifunc{MPI\_TYPE\_CREATE\_HVECTOR}} >>> \MPIupdateEnd{3.0} >>> ...etc. >>> ----- >>> >>> If we remove all the \MPIdelete* and \MPIupdate* macros, it builds ok. If we put even one of them back, we get "Float(s) lost" and pdflatex aborts. But other than that, we're totally stumped (we've wasted 2+ hours trying to solve this so far :-( ). >>> >>> What should we do? >>> >>> -- >>> Jeff Squyres >>> jsquyres(a)cisco.com >>> For corporate legal information go to: >>> http://www.cisco.com/web/about/doing_business/legal/cri/ >>> >>> >>> _______________________________________________ >>> mpi-forum mailing list >>> mpi-forum(a)lists.mpi-forum.org >>> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum >> >> William Gropp >> Director, Parallel Computing Institute >> Deputy Director for Research >> Institute for Advanced Computing Applications and Technologies >> Paul and Cynthia Saylor Professor of Computer Science >> University of Illinois Urbana-Champaign >> >> >> >> _______________________________________________ >> mpi-forum mailing list >> mpi-forum(a)lists.mpi-forum.org >> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum > > William Gropp > Director, Parallel Computing Institute > Deputy Director for Research > Institute for Advanced Computing Applications and Technologies > Paul and Cynthia Saylor Professor of Computer Science > University of Illinois Urbana-Champaign > > > > _______________________________________________ > mpi-forum mailing list > mpi-forum(a)lists.mpi-forum.org > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum -- Jeff Squyres jsquyres(a)cisco.com For corporate legal information go to: http://www.cisco.com/web/about/doing_business/legal/cri/

1 0

Re: [Mpi-forum] (Continued) Latex woes
by William Gropp 09 Jun '11

09 Jun '11

Attached is the diff of the fix. I can't commit it, so one of you will need to. Bill On Jun 9, 2011, at 12:20 PM, William Gropp wrote: > Floats lost is often due to a limitation in LaTeX. One possible > issue is that the the change markers use floats, and so do tables. > LaTeX isn't very happy with floats within floats. I'm in a meeting > today and can't look at it until later, but one solution may be to > inline the ticket info in these cases. > > Bill > > On Jun 9, 2011, at 11:30 AM, Jeff Squyres wrote: > >> Bill -- >> >> We're still having latex woes in ticket 278. Can you help? >> >> We replaced all the \MPIreplace macros with deletebegin/end and >> updatebegin/end, but we're still getting "Float(s) lost". It seems >> to have to do with a table in datatype.tex. The table looks like >> this: >> >> ----- >> \begin{table}[ht] >> \begin{center} >> \begin{constlist} >> \constitem{MPI\_COMBINER\_NAMED}{a named predefined datatype} >> ...etc. >> \MPIdeleteBegin{3.0}{278} >> \constitem{MPI\_COMBINER\_HVECTOR\_INTEGER}{\mpifunc{MPI\_TYPE >> \_HVECTOR} from Fortran} >> \constitem{MPI\_COMBINER\_HVECTOR}{\mpifunc{MPI\_TYPE\_HVECTOR} >> from C or C++} >> \constskipitem{}{\ \ and in some case Fortran} >> \constskipitem{}{\ \ or \mpifunc{MPI\_TYPE\_CREATE\_HVECTOR}} >> \MPIdeleteEnd{3.0} >> \MPIupdateBegin{3.0}{278} >> \constitem{MPI\_COMBINER\_HVECTOR}{\mpifunc{MPI\_TYPE\_CREATE >> \_HVECTOR}} >> \MPIupdateEnd{3.0} >> ...etc. >> ----- >> >> If we remove all the \MPIdelete* and \MPIupdate* macros, it builds >> ok. If we put even one of them back, we get "Float(s) lost" and >> pdflatex aborts. But other than that, we're totally stumped (we've >> wasted 2+ hours trying to solve this so far :-( ). >> >> What should we do? >> >> -- >> Jeff Squyres >> jsquyres(a)cisco.com >> For corporate legal information go to: >> http://www.cisco.com/web/about/doing_business/legal/cri/ >> >> >> _______________________________________________ >> mpi-forum mailing list >> mpi-forum(a)lists.mpi-forum.org >> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum > > William Gropp > Director, Parallel Computing Institute > Deputy Director for Research > Institute for Advanced Computing Applications and Technologies > Paul and Cynthia Saylor Professor of Computer Science > University of Illinois Urbana-Champaign > > > > _______________________________________________ > mpi-forum mailing list > mpi-forum(a)lists.mpi-forum.org > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum William Gropp Director, Parallel Computing Institute Deputy Director for Research Institute for Advanced Computing Applications and Technologies Paul and Cynthia Saylor Professor of Computer Science University of Illinois Urbana-Champaign

1 0

Re: [Mpi-forum] 2012 tentative meeting agenda
by Underwood, Keith D 09 Jun '11

09 Jun '11

The thought was that the later meetings would be less frequent and one day longer... > -----Original Message----- > From: mpi-forum-bounces(a)lists.mpi-forum.org [mailto:mpi-forum- > bounces(a)lists.mpi-forum.org] On Behalf Of Jeff Squyres > Sent: Thursday, June 09, 2011 9:26 AM > To: Main MPI Forum mailing list > Subject: Re: [Mpi-forum] 2012 tentative meeting agenda > > Rich -- > > Did you mean June 25-27 and Dec 3-5? (you said June 25-28 and Dec 3-6) > > Also, if you put the meetings in Portland and Chicago, then Microsoft > will be hosting *ALL* of them. Was that intended? Will Microsoft be > willing to do that? Will we all be willing to put up with intermittent > wifi at every meeting? (sorry, it's true...) > > I'd be willing to keep hosting at Cisco in San Jose, CA. > > > > On Jun 8, 2011, at 11:05 AM, Graham, Richard L. wrote: > > > At the last meeting we had general agreement to continue the current > pace of meetings until MPI 3.0 is finalized, and then go down to 4 > meetings a year, but longer meetings. The time frame we have set for > MPI 3.0 means that the first 3 meetings in 2012 will continue at the > current pace, and then we will cut back to quarterly. Here is a first > suggestion for meeting dates - please send any conflicts that you are > aware of. Also, there has been a push to move the west coast meetings > to Portland, OR - are people OK with that ? > > > > Thanks, > > Rich > > > > Jan 9-11: Portland, OR > > March 5-7: Chicago, IL > > April 30 - May 2: Portland, OR - Final MPI 3.0 vote > > June 25-28: Chicago, IL > > Sept: In conjunction with EuroMPI > > Dec 3-6: Portland, OR > > _______________________________________________ > > mpi-forum mailing list > > mpi-forum(a)lists.mpi-forum.org > > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum > > > -- > Jeff Squyres > jsquyres(a)cisco.com > For corporate legal information go to: > http://www.cisco.com/web/about/doing_business/legal/cri/ > > > _______________________________________________ > mpi-forum mailing list > mpi-forum(a)lists.mpi-forum.org > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum

1 0

Re: [Mpi-forum] 2012 tentative meeting agenda
by Fab Tillier 09 Jun '11

09 Jun '11

Rolf Rabenseifner wrote on Thu, 9 Jun 2011 at 05:19:43 > 2) Location Portland, OR: > ------------------------- > For Tokyo - USA, and Frankfurt - USA, direct flights exists > - to Chicago (ORD) with many airlines, but not with Delta > - to Portland OR (PDX) with Delta As far as I know, the direct flight from FRA to PDX has been cancelled. There is still a direct from AMS to PDX though. > - to Seattle WA (SEA) with many airlines > > Especially for crossed tickets (the cheap way for HLRS) > the same flight company is essential. > Therefore, the pair Chicago & Seattle would be work > with Star Alliance flights significantly better for me, > and the same or better for Tokyo. > > I do not know, who can offer a good place in Seattle > for our meetings. I don't have any interest in hosting in Seattle, as it is no better for me than San Jose. Cheers, -Fab > > Rolf > > > ----- Original Message ----- >> From: "Anthony Skjellum" <tony(a)cis.uab.edu> >> To: "Main MPI Forum mailing list" <mpi-forum(a)lists.mpi-forum.org> >> Sent: Thursday, June 9, 2011 12:53:15 AM >> Subject: Re: [Mpi-forum] 2012 tentative meeting agenda >> May I suggest Atlanta for January 2012 meeting, it is too cold and too >> far to go to Portland :-) >> We never have a meeting in the Southeast :-) >> >> ----- Original Message ----- >> From: "Bronis R. de Supinski" <bronis(a)llnl.gov> >> To: "Main MPI Forum mailing list" <mpi-forum(a)lists.mpi-forum.org> >> Sent: Wednesday, June 8, 2011 1:12:10 PM >> Subject: Re: [Mpi-forum] 2012 tentative meeting agenda >> >> >> Rich: >> >> My 2012 clalendar is not fully populated yet so >> it is hard to say for certain although I do not >> see any conflicts for most of the dates. However, >> June 25-28 is bad. It conflicts with ICS and most >> probably with the SC PC meeting (its dates will >> be finalized tomorrow). In general June 2012 is >> already full with significant events each week. >> I suggest we push that meeting out to mid to late >> July 2012. >> >> Bronis >> >> >> On Wed, 8 Jun 2011, Graham, Richard L. wrote: >> >>> At the last meeting we had general agreement to continue the current >>> pace of meetings until MPI 3.0 is finalized, and then go down to 4 >>> meetings a year, but longer meetings. The time frame we have set for >>> MPI 3.0 means that the first 3 meetings in 2012 will continue at the >>> current pace, and then we will cut back to quarterly. Here is a >>> first suggestion for meeting dates – please send any conflicts that >>> you are aware of. Also, there has been a push to move the west coast >>> meetings to Portland, OR – are people OK with that ? >>> >>> Thanks, >>> Rich >>> >>> Jan 9-11: Portland, OR >>> March 5-7: Chicago, IL >>> April 30 – May 2: Portland, OR – Final MPI 3.0 vote >>> June 25-28: Chicago, IL >>> Sept: In conjunction with EuroMPI >>> Dec 3-6: Portland, OR >>> >> _______________________________________________ >> mpi-forum mailing list >> mpi-forum(a)lists.mpi-forum.org >> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum >> >> -- >> >> >> >> >> Anthony Skjellum, PhD >> Professor and Chair >> Dept. of Computer and Information Sciences >> College of Arts and Sciences (CAS) >> University of Alabama at Birmingham >> +1-(205)934-8657; FAX: +1- (205)934-5473 >> ___________________________________________ >> CONFIDENTIALITY: This e-mail and any attachments are confidential and >> may be privileged. If you are not a named recipient, please notify >> the >> sender immediately and do not disclose the contents to another person, >> use it for any purpose or store or copy the information in any medium. >> >> Please consider the environment before printing this e-mail ☺ >> >> _______________________________________________ >> mpi-forum mailing list >> mpi-forum(a)lists.mpi-forum.org >> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum >

1 0

Re: [Mpi-forum] 2012 tentative meeting agenda
by Fab Tillier 09 Jun '11

09 Jun '11

Jeff Squyres wrote on Thu, 9 Jun 2011 at 06:25:51 > Rich -- > > Did you mean June 25-27 and Dec 3-5? (you said June 25-28 and Dec 3-6) My understanding is that once we go quarterly, the meetings will be longer. > Also, if you put the meetings in Portland and Chicago, then Microsoft will be > hosting *ALL* of them. Was that intended? It's not a requirement that Microsoft host in Chicago, someone else could. Currently, it's mostly out of convenience as the downtown location of the Microsoft office is easily accessible, and consensus was that this outweighed the wifi issues. > Will Microsoft be willing to do that? The Chicago meetings are a bit of a PITA for me to attend, so I was hoping to skip out on some and let someone else host. Hosting in Portland is obviously ideal for me: good coffee, good beer, good food, and I live here. Assuming I attend the Chicago meetings I don't foresee any problems continuing to host them at our downtown facility. > Will we all be willing to put up with intermittent wifi at every meeting? > (sorry, it's true...) I would hope the guest internet access is better than the Chicago one, though I don't have a system to test it with. It would certainly be newer, and newer things are supposed to be better... > I'd be willing to keep hosting at Cisco in San Jose, CA. I suggest we pick just one meeting in Portland at first, see how the location works out, and then the Forum can decided whether to do more here or stick with San Jose. If we did the January meeting in Portland, we could leave the December meeting open and then pick either PDX or SJC based on how it goes? Cheers, -Fab > On Jun 8, 2011, at 11:05 AM, Graham, Richard L. wrote: > >> At the last meeting we had general agreement to continue the current > pace of meetings until MPI 3.0 is finalized, and then go down to 4 meetings a > year, but longer meetings. The time frame we have set for MPI 3.0 means > that the first 3 meetings in 2012 will continue at the current pace, and then > we will cut back to quarterly. Here is a first suggestion for meeting dates - > please send any conflicts that you are aware of. Also, there has been a push > to move the west coast meetings to Portland, OR - are people OK with that ? >> >> Thanks, >> Rich >> >> Jan 9-11: Portland, OR >> March 5-7: Chicago, IL >> April 30 - May 2: Portland, OR - Final MPI 3.0 vote >> June 25-28: Chicago, IL >> Sept: In conjunction with EuroMPI >> Dec 3-6: Portland, OR >> _______________________________________________ >> mpi-forum mailing list >> mpi-forum(a)lists.mpi-forum.org >> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum > >

1 0

Re: [Mpi-forum] Ticket 279, making C++ bindings optional
by Fab Tillier 09 Jun '11

09 Jun '11

For those wanting a full standard document with changelog update, I just attached it ticket 279 (https://svn.mpi-forum.org/trac/mpi-forum-web/attachment/ticket/279/ticket-2…) Search for 'ticket279' to find the changes. Cheers, -Fab Fab Tillier wrote on Thu, 9 Jun 2011 at 10:51:24 > Hi Folks, > > In ticket 279 (https://svn.mpi-forum.org/trac/mpi-forum-web/ticket/279) I > added a single sentence to allow C++ bindings to be optional. The wording is: > > "A compliant MPI implementation providing C++ language bindings must > provide the entire set defined in this document." > > This language is similar to that of the Fortran bindings in section 16.2.1 > > The goal is for Jeff to do the first reading of this at the next meeting, and this > seems pretty straight forward to me. Still, I wanted to find out if it gave > anyone heartburn. > > Thanks, > -Fab > > _______________________________________________ > mpi-forum mailing list > mpi-forum(a)lists.mpi-forum.org > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum

1 0

First reading for MPI Timer Request ticket 277
by Fab Tillier 09 Jun '11

09 Jun '11

Hi Folks, In ticket 277 (https://svn.mpi-forum.org/trac/mpi-forum-web/ticket/277) I formalized the timer functionality discussed at the March meeting (PPT slides from that meeting attached to the ticket). I have just attached to ticket 277 two PDF files. The first (https://svn.mpi-forum.org/trac/mpi-forum-web/attachment/ticket/277/ticket-2…) is just the two pages changed in chapter 8 which contain the new function definitions. The second (https://svn.mpi-forum.org/trac/mpi-forum-web/attachment/ticket/277/ticket-2…) is the full standard document, including the changelog entry. Jeff will be conducting a first reading at the upcoming meeting in July. Please have a look and provide feedback. Thanks, -Fab

1 0

Ticket 279, making C++ bindings optional
by Fab Tillier 09 Jun '11

09 Jun '11

Hi Folks, In ticket 279 (https://svn.mpi-forum.org/trac/mpi-forum-web/ticket/279) I added a single sentence to allow C++ bindings to be optional. The wording is: "A compliant MPI implementation providing C++ language bindings must provide the entire set defined in this document." This language is similar to that of the Fortran bindings in section 16.2.1 The goal is for Jeff to do the first reading of this at the next meeting, and this seems pretty straight forward to me. Still, I wanted to find out if it gave anyone heartburn. Thanks, -Fab

1 0

Re: [Mpi-forum] (Continued) Latex woes
by William Gropp 09 Jun '11

09 Jun '11

Floats lost is often due to a limitation in LaTeX. One possible issue is that the the change markers use floats, and so do tables. LaTeX isn't very happy with floats within floats. I'm in a meeting today and can't look at it until later, but one solution may be to inline the ticket info in these cases. Bill On Jun 9, 2011, at 11:30 AM, Jeff Squyres wrote: > Bill -- > > We're still having latex woes in ticket 278. Can you help? > > We replaced all the \MPIreplace macros with deletebegin/end and > updatebegin/end, but we're still getting "Float(s) lost". It seems > to have to do with a table in datatype.tex. The table looks like > this: > > ----- > \begin{table}[ht] > \begin{center} > \begin{constlist} > \constitem{MPI\_COMBINER\_NAMED}{a named predefined datatype} > ...etc. > \MPIdeleteBegin{3.0}{278} > \constitem{MPI\_COMBINER\_HVECTOR\_INTEGER}{\mpifunc{MPI\_TYPE > \_HVECTOR} from Fortran} > \constitem{MPI\_COMBINER\_HVECTOR}{\mpifunc{MPI\_TYPE\_HVECTOR} from > C or C++} > \constskipitem{}{\ \ and in some case Fortran} > \constskipitem{}{\ \ or \mpifunc{MPI\_TYPE\_CREATE\_HVECTOR}} > \MPIdeleteEnd{3.0} > \MPIupdateBegin{3.0}{278} > \constitem{MPI\_COMBINER\_HVECTOR}{\mpifunc{MPI\_TYPE\_CREATE > \_HVECTOR}} > \MPIupdateEnd{3.0} > ...etc. > ----- > > If we remove all the \MPIdelete* and \MPIupdate* macros, it builds > ok. If we put even one of them back, we get "Float(s) lost" and > pdflatex aborts. But other than that, we're totally stumped (we've > wasted 2+ hours trying to solve this so far :-( ). > > What should we do? > > -- > Jeff Squyres > jsquyres(a)cisco.com > For corporate legal information go to: > http://www.cisco.com/web/about/doing_business/legal/cri/ > > > _______________________________________________ > mpi-forum mailing list > mpi-forum(a)lists.mpi-forum.org > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-forum William Gropp Director, Parallel Computing Institute Deputy Director for Research Institute for Advanced Computing Applications and Technologies Paul and Cynthia Saylor Professor of Computer Science University of Illinois Urbana-Champaign

1 0

(Continued) Latex woes
by Jeff Squyres 09 Jun '11

09 Jun '11

Bill -- We're still having latex woes in ticket 278. Can you help? We replaced all the \MPIreplace macros with deletebegin/end and updatebegin/end, but we're still getting "Float(s) lost". It seems to have to do with a table in datatype.tex. The table looks like this: ----- \begin{table}[ht] \begin{center} \begin{constlist} \constitem{MPI\_COMBINER\_NAMED}{a named predefined datatype} ...etc. \MPIdeleteBegin{3.0}{278} \constitem{MPI\_COMBINER\_HVECTOR\_INTEGER}{\mpifunc{MPI\_TYPE\_HVECTOR} from Fortran} \constitem{MPI\_COMBINER\_HVECTOR}{\mpifunc{MPI\_TYPE\_HVECTOR} from C or C++} \constskipitem{}{\ \ and in some case Fortran} \constskipitem{}{\ \ or \mpifunc{MPI\_TYPE\_CREATE\_HVECTOR}} \MPIdeleteEnd{3.0} \MPIupdateBegin{3.0}{278} \constitem{MPI\_COMBINER\_HVECTOR}{\mpifunc{MPI\_TYPE\_CREATE\_HVECTOR}} \MPIupdateEnd{3.0} ...etc. ----- If we remove all the \MPIdelete* and \MPIupdate* macros, it builds ok. If we put even one of them back, we get "Float(s) lost" and pdflatex aborts. But other than that, we're totally stumped (we've wasted 2+ hours trying to solve this so far :-( ). What should we do? -- Jeff Squyres jsquyres(a)cisco.com For corporate legal information go to: http://www.cisco.com/web/about/doing_business/legal/cri/

1 0