September 2010 - mpiwg-tools - mailman3.cels.anl.gov

Re: [Mpi3-tools] Updated document
by John DelSignore 16 Sep '10

16 Sep '10

Jeff Squyres wrote: > Here's a copy of the updated document. I made all my changes in red > text; you should be able to flip through it quickly to see my > changes. > > At a minimum, could Martin and John D. reply that it's ok? Looks good to me. > I did *not* change any text about the IP address. Reading the 3 > places where IP addresses are discussed, I'm comfortable leaving it > unspecified as to whether the IP address is IPv4 or IPv6. Do any > existing tools allow MPIR_PROCDESC.host_name to be an IPv6 address? TotalView supports IPv6, and MPIR_PROCDESC.host_name can be an IPv6 address or host name that maps to an IPv6 address. Cheers, John D.

1 0

Re: [Mpi3-tools] Updated document
by Martin Schulz 16 Sep '10

16 Sep '10

Hi Jeff, Thanks for running the session and for making the changes. I am (pleasantly) surprised that they are only that minor :). I looked over the text in red and this all looks good to me. Martin On Sep 16, 2010, at 10:52 AM, Jeff Squyres wrote: > ALERT: There is currently a vulnerability with PDF files. Do not > open any PDF attachments from sources you do not trust or unexpected > PDF files from sources you do trust. > > Here's a copy of the updated document. I made all my changes in red > text; you should be able to flip through it quickly to see my changes. > > At a minimum, could Martin and John D. reply that it's ok? > > I did *not* change any text about the IP address. Reading the 3 > places where IP addresses are discussed, I'm comfortable leaving it > unspecified as to whether the IP address is IPv4 or IPv6. Do any > existing tools allow MPIR_PROCDESC.host_name to be an IPv6 address? > > -- > Jeff Squyres > jsquyres(a)cisco.com > For corporate legal information go to: > http://BLOCKEDwww.BLOCKEDcisco.com/web/about/doing_business/legal/cri/ > <mpi-report.pdf><ATT00001..txt> ________________________________________________________________________ Martin Schulz, schulzm(a)llnl.gov, http://people.llnl.gov/schulzm CASC @ Lawrence Livermore National Laboratory, Livermore, USA

1 0

Re: [Mpi3-tools] URGENT: feedback from first reading of MPIR
by Jeff Squyres 16 Sep '10

16 Sep '10

On Sep 16, 2010, at 6:23 PM, John DelSignore wrote: > No, it doesn't have to be an IPv4 address. There is nothing that I know of in MPIR that excludes IPv6. The "host_name" field of the procdesc structure is a string pointer, so the string can resolve to an IPv4 or IPv6 address. And as it says in the document, on BProc systems, "host_name" is a BProc node number so the string is also open to interpretations other than just IP address. Our emails crossed in the ether, but thankfully your reply supports the same conclusion I came to. :-) -- Jeff Squyres jsquyres(a)cisco.com For corporate legal information go to: http://www.cisco.com/web/about/doing_business/legal/cri/

1 0

[Mpi3-tools] Updated document
by Jeff Squyres 16 Sep '10

16 Sep '10

Here's a copy of the updated document. I made all my changes in red text; you should be able to flip through it quickly to see my changes. At a minimum, could Martin and John D. reply that it's ok? I did *not* change any text about the IP address. Reading the 3 places where IP addresses are discussed, I'm comfortable leaving it unspecified as to whether the IP address is IPv4 or IPv6. Do any existing tools allow MPIR_PROCDESC.host_name to be an IPv6 address? -- 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: [Mpi3-tools] URGENT: feedback from first reading of MPIR
by John DelSignore 16 Sep '10

16 Sep '10

Jeff Squyres wrote: > The reading went generally well. There were a few (legitimate) nit-picks from the Forum that I need to fix tonight and re-present tomorrow. Doing this will still allow us to meet the requirement of "first formal reading" at this Stuttgart Forum meeting -- paving the way for the first vote next meeting, etc. > > Here's a list of the nits: > > - add version number for MPIR document > - It should to be mentioned explicitely that the document is _NOT_ part of MPI 3 and does not have to be implemented for MPI 3 compliance > - Is "IP address" only IPv4 ? No, it doesn't have to be an IPv4 address. There is nothing that I know of in MPIR that excludes IPv6. The "host_name" field of the procdesc structure is a string pointer, so the string can resolve to an IPv4 or IPv6 address. And as it says in the document, on BProc systems, "host_name" is a BProc node number so the string is also open to interpretations other than just IP address. Cheers, John D. > - Message Queue has not been mentioned as a separate document > - MPI_Comm_join is missing on the list of unsupported functionality > - Fix Link to page number in bibliography (It is irritating as they _all_ point to page 1) > - MPIR_debug_state could be set by a compiler default > - My name (Marc-Andre) as Marc-Andr\'{e} > - inconsistent use of names in chapter 2 (MPI APA) vs. rest of document > > I will be fixing all of these in the next hour or two. The only real question is "IP address" -- do we only mean IPv4? Or is IPv6 ok? I'll read the text closely to see what we intended. If anyone has an opinion, please chime in ASAP. Thanks. >

1 0

[Mpi3-tools] URGENT: feedback from first reading of MPIR
by Jeff Squyres 16 Sep '10

16 Sep '10

The reading went generally well. There were a few (legitimate) nit-picks from the Forum that I need to fix tonight and re-present tomorrow. Doing this will still allow us to meet the requirement of "first formal reading" at this Stuttgart Forum meeting -- paving the way for the first vote next meeting, etc. Here's a list of the nits: - add version number for MPIR document - It should to be mentioned explicitely that the document is _NOT_ part of MPI 3 and does not have to be implemented for MPI 3 compliance - Is "IP address" only IPv4 ? - Message Queue has not been mentioned as a separate document - MPI_Comm_join is missing on the list of unsupported functionality - Fix Link to page number in bibliography (It is irritating as they _all_ point to page 1) - MPIR_debug_state could be set by a compiler default - My name (Marc-Andre) as Marc-Andr\'{e} - inconsistent use of names in chapter 2 (MPI APA) vs. rest of document I will be fixing all of these in the next hour or two. The only real question is "IP address" -- do we only mean IPv4? Or is IPv6 ok? I'll read the text closely to see what we intended. If anyone has an opinion, please chime in ASAP. Thanks. -- 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: [Mpi3-tools] Fwd: [MPI Forum] #228: MPIR Specification - An Official Companion Document for the MPI Standard
by Martin Schulz 01 Sep '10

01 Sep '10

Hi Jeff, We overlapped :) - I just sent it out. Martin On Sep 1, 2010, at 3:58 PM, Jeff Squyres wrote: > Did this get sent out to the main Forum mailing list? It's no good > if there's just a ticket for it -- the entire Forum needs to be > notified so that they know to go examine it before going to Stuttgart. > > > > Begin forwarded message: > >> From: "MPI Forum" <mpi-22(a)lists.mpi-forum.org> >> Date: September 1, 2010 12:27:25 AM EDT >> Subject: [MPI Forum] #228: MPIR Specification - An Official >> Companion Document for the MPI Standard >> >> #228: MPIR Specification - An Official Companion Document for the >> MPI Standard >> --------------------------------------------------------- >> +------------------ >> Reporter: schulzm >> | Owner: jsquyres >> Type: Enhancements to standard >> | Status: new >> Priority: Scheduled >> | Milestone: 2010/06/14 California, USA >> Version: MPI 3.0 >> | Keywords: >> Implementation: Unnecessary | >> Author_bill_gropp: 0 >> Author_rich_graham: 0 | >> Author_adam_moody: 0 >> Author_torsten_hoefler: 0 | >> Author_dick_treumann: 0 >> Author_jesper_larsson_traeff: 0 | >> Author_george_bosilca: 0 >> Author_david_solt: 0 | >> Author_bronis_de_supinski: 0 >> Author_rajeev_thakur: 0 | >> Author_jeff_squyres: 0 >> Author_alexander_supalov: 0 | >> Author_rolf_rabenseifner: 0 >> --------------------------------------------------------- >> +------------------ >> = MPIR Specification = >> >> Contact: Martin Schulz and Jeff Squyres >> >> Main document author: John DelSignore >> >> == Description == >> >> The document attached to this ticket describes the MPIR process >> acquisition interface as it is currently used and implemented in a >> variety >> of MPI implementations. The tools WG is proposing to provide this >> document >> as an officially side document approved by the MPI forum. Based on >> the >> discussions during the June MPI forum meeting, such side documents >> should >> go through a similar voting process (details tbd.) as any addition >> to the >> MPI standard itself, but it will be published on the MPI forum >> servers as >> a separate document. >> >> -- >> Ticket URL: <https://*svn.mpi-forum.org/trac/mpi-forum-web/ticket/ >> 228> >> MPI Forum <https://*svn.mpi-forum.org/> >> MPI Forum > > > -- > Jeff Squyres > jsquyres(a)cisco.com > For corporate legal information go to: > http://*www.*cisco.com/web/about/doing_business/legal/cri/ > > > _______________________________________________ > Mpi3-tools mailing list > Mpi3-tools(a)lists.mpi-forum.org > http://*lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-tools > ________________________________________________________________________ Martin Schulz, schulzm(a)llnl.gov, http://people.llnl.gov/schulzm CASC @ Lawrence Livermore National Laboratory, Livermore, USA

1 0

[Mpi3-tools] Fwd: [MPI Forum] #228: MPIR Specification - An Official Companion Document for the MPI Standard
by Jeff Squyres 01 Sep '10

01 Sep '10

Did this get sent out to the main Forum mailing list? It's no good if there's just a ticket for it -- the entire Forum needs to be notified so that they know to go examine it before going to Stuttgart. Begin forwarded message: > From: "MPI Forum" <mpi-22(a)lists.mpi-forum.org> > Date: September 1, 2010 12:27:25 AM EDT > Subject: [MPI Forum] #228: MPIR Specification - An Official Companion Document for the MPI Standard > > #228: MPIR Specification - An Official Companion Document for the MPI Standard > ---------------------------------------------------------+------------------ > Reporter: schulzm | Owner: jsquyres > Type: Enhancements to standard | Status: new > Priority: Scheduled | Milestone: 2010/06/14 California, USA > Version: MPI 3.0 | Keywords: > Implementation: Unnecessary | Author_bill_gropp: 0 > Author_rich_graham: 0 | Author_adam_moody: 0 > Author_torsten_hoefler: 0 | Author_dick_treumann: 0 > Author_jesper_larsson_traeff: 0 | Author_george_bosilca: 0 > Author_david_solt: 0 | Author_bronis_de_supinski: 0 > Author_rajeev_thakur: 0 | Author_jeff_squyres: 0 > Author_alexander_supalov: 0 | Author_rolf_rabenseifner: 0 > ---------------------------------------------------------+------------------ > = MPIR Specification = > > Contact: Martin Schulz and Jeff Squyres > > Main document author: John DelSignore > > == Description == > > The document attached to this ticket describes the MPIR process > acquisition interface as it is currently used and implemented in a variety > of MPI implementations. The tools WG is proposing to provide this document > as an officially side document approved by the MPI forum. Based on the > discussions during the June MPI forum meeting, such side documents should > go through a similar voting process (details tbd.) as any addition to the > MPI standard itself, but it will be published on the MPI forum servers as > a separate document. > > -- > Ticket URL: <https://svn.mpi-forum.org/trac/mpi-forum-web/ticket/228> > MPI Forum <https://svn.mpi-forum.org/> > 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: [Mpi3-tools] Final draft of MPIR Document
by John DelSignore 01 Sep '10

01 Sep '10

The arrows in the picture look right to me. >> I would read the figure (as it is right now) as "The tools starts >> mpiexec". >> whereas the text reads "The tools is started on the mpiexec >> executable". Step 4 is "the tool starts or attached to the mpiexec process". Step 1 is "the tool takes the mpiexec executable name as an argument". The picture and the text are correct as written. I'm in a bit of a hurry at the moment (gotta get to a sunset) so if you need more of an explanation it will need to wait until late when I get back from vacation. For now, I think it is correct as-is. Cheers, John D. Martin Schulz wrote: > Hi Marc-Andre, > > Thanks for the comments - all fixed and in SVN, except for the > comment in the picture, which I will leave to John. > > Martin > > > On Sep 1, 2010, at 2:02 AM, Marc-Andrï¿½ Hermanns wrote: > >> Hi all, >> >> apologies for my late comments. >> >> I refer to the SVN version on Monday morning. >> >> p.1, l.33: >> wasnt -> wasn't >> >> p.1, l.44: >> current state state -> current state >> >> p.2, l.22: >> the tool read symbol ï¿½ and trace ... -> the tool reads symbol ï¿½ and >> traces >> or >> the tools read symbol ï¿½ and trace >> >> p.2, l.43: >> attaches to MPI processes -> attaches to the MPI processes >> >> p.8, l.30: >> attach Colorgreenmode -> attach mode >> >> p.12, l.23 >> (sorry that I obviously missed this in previous drafts) >> In the figure 8.1 directly with the first step, shouldn't the arrow >> be pointing into the other direction? >> I would read the figure (as it is right now) as "The tools starts >> mpiexec". >> whereas the text reads "The tools is started on the mpiexec >> executable". >> Could someone help to clarify this? >> >> p.18, l.41: >> be reported it to -> be reported to >> >> p.21, l.1: >> the tool need only release -> the tool needs only to release >> >> Cheers, >> Marc-Andre >> >> On 28.08.2010, at 01:02, Martin Schulz wrote: >> >>> Hi all, >>> >>> Attached is the final version of the MPIR document, as we are >>> planning to present it to >>> the MPI forum. Please speak up if you have any final comments or >>> concerns. Otherwise, >>> we are planning to submit this version on Monday to the forum for >>> review for the >>> September meeting. >>> >>> Martin >>> >>> <mpi-report.pdf> >>> >>> >>> ________________________________________________________________________ >>> Martin Schulz, schulzm(a)llnl.gov, http://people.llnl.gov/schulzm >>> CASC @ Lawrence Livermore National Laboratory, Livermore, USA >>> >>> >>> >>> _______________________________________________ >>> Mpi3-tools mailing list >>> Mpi3-tools(a)lists.mpi-forum.org >>> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-tools >> -- >> Marc-Andre Hermanns >> German Research School for >> Simulation Sciences GmbH >> c/o Laboratory for Parallel Programming >> 52056 Aachen | Germany >> >> Tel +49 241 80 99753 >> Fax +49 241 80 6 99753 >> Web www.grs-sim.de >> >> Members: Forschungszentrum Jï¿½lich GmbH | RWTH Aachen University >> Registered in the commercial register of the local court of >> Dï¿½ren (Amtsgericht Dï¿½ren) under registration number HRB 5268 >> Registered office: Jï¿½lich >> Executive board: Prof. Dr. Heiner Mï¿½ller-Krumbhaar | Prof. Marek >> Behr, Ph.D. >> >> <smime.p7s><ATT00001..txt> > > ________________________________________________________________________ > Martin Schulz, schulzm(a)llnl.gov, http://people.llnl.gov/schulzm > CASC @ Lawrence Livermore National Laboratory, Livermore, USA > > > > > _______________________________________________ > Mpi3-tools mailing list > Mpi3-tools(a)lists.mpi-forum.org > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-tools >

1 0

Re: [Mpi3-tools] Final draft of MPIR Document
by Martin Schulz 01 Sep '10

01 Sep '10

Hi Marc-Andre, Thanks for the comments - all fixed and in SVN, except for the comment in the picture, which I will leave to John. Martin On Sep 1, 2010, at 2:02 AM, Marc-Andrï¿½ Hermanns wrote: > Hi all, > > apologies for my late comments. > > I refer to the SVN version on Monday morning. > > p.1, l.33: > wasnt -> wasn't > > p.1, l.44: > current state state -> current state > > p.2, l.22: > the tool read symbol ï¿½ and trace ... -> the tool reads symbol ï¿½ and > traces > or > the tools read symbol ï¿½ and trace > > p.2, l.43: > attaches to MPI processes -> attaches to the MPI processes > > p.8, l.30: > attach Colorgreenmode -> attach mode > > p.12, l.23 > (sorry that I obviously missed this in previous drafts) > In the figure 8.1 directly with the first step, shouldn't the arrow > be pointing into the other direction? > I would read the figure (as it is right now) as "The tools starts > mpiexec". > whereas the text reads "The tools is started on the mpiexec > executable". > Could someone help to clarify this? > > p.18, l.41: > be reported it to -> be reported to > > p.21, l.1: > the tool need only release -> the tool needs only to release > > Cheers, > Marc-Andre > > On 28.08.2010, at 01:02, Martin Schulz wrote: > >> Hi all, >> >> Attached is the final version of the MPIR document, as we are >> planning to present it to >> the MPI forum. Please speak up if you have any final comments or >> concerns. Otherwise, >> we are planning to submit this version on Monday to the forum for >> review for the >> September meeting. >> >> Martin >> >> <mpi-report.pdf> >> >> >> ________________________________________________________________________ >> Martin Schulz, schulzm(a)llnl.gov, http://people.llnl.gov/schulzm >> CASC @ Lawrence Livermore National Laboratory, Livermore, USA >> >> >> >> _______________________________________________ >> Mpi3-tools mailing list >> Mpi3-tools(a)lists.mpi-forum.org >> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-tools > > -- > Marc-Andre Hermanns > German Research School for > Simulation Sciences GmbH > c/o Laboratory for Parallel Programming > 52056 Aachen | Germany > > Tel +49 241 80 99753 > Fax +49 241 80 6 99753 > Web www.grs-sim.de > > Members: Forschungszentrum Jï¿½lich GmbH | RWTH Aachen University > Registered in the commercial register of the local court of > Dï¿½ren (Amtsgericht Dï¿½ren) under registration number HRB 5268 > Registered office: Jï¿½lich > Executive board: Prof. Dr. Heiner Mï¿½ller-Krumbhaar | Prof. Marek > Behr, Ph.D. > > <smime.p7s><ATT00001..txt> ________________________________________________________________________ Martin Schulz, schulzm(a)llnl.gov, http://people.llnl.gov/schulzm CASC @ Lawrence Livermore National Laboratory, Livermore, USA

1 0