- discuss - mailman3.cels.anl.gov

Nemesis mode
by Bob Ilgner 09 May '13

09 May '13

Dear all, I am using a large shared memory machine ( > 100 cores) and want to use mpi in nemesis mode. How do I do that ? There is some material in the faq about using ch3. Where does that come in ? Regards,

1 0

Re: [mpich-discuss] mpich-3.0.4: building shared libraries w/cygwin
by Brad Aagaard 08 May '13

08 May '13

On 05/06/2013 05:45 PM, Pavan Balaji wrote: > > On 05/02/2013 04:36 PM US Central Time, Brad Aagaard wrote: >> What configure/make arguments should I use to build shared libraries on >> cygwin? > > Just use --enable-shared without any additional --enable-sharedlibs > argument. You might also want to give --disable-static if you don't > want static libraries. > Dropping --enable-sharedlibs from configure gives me the same result as when I use --enable-sharedlibs=libtool or --enable-sharedlibs=gcc-cygwin. I end up with static libraries and no shared libraries even though I also use --enable-shared --disable-static. Is there a way to check the Makefile or libtool to see why it doesn't want to build shared libraries? Thanks, Brad

1 0

Re: [mpich-discuss] smpd version mismatch
by Pavan Balaji 08 May '13

08 May '13

Please also tell us the version of mpich you are using, and which OS you are using. Thanks, -- Pavan On 05/08/2013 04:10 PM US Central Time, Antonio J. Peï¿½a wrote: > > > Hi Fernando, > > > > Is "Boussines" the name of your computer? You may have two versions of > MPI in your computer. This may happen, for instance, if you compiled > your own version of MPI, and your system already has an MPI installed. > Could you please share the exact command-line arguments you're using > both for your cpi example which worked, and for that which didn't work. > > > > In addition, this previous discussion may help: > > > > http://lists.mcs.anl.gov/pipermail/mpich-discuss/2009-February/004656.html > > > > -- > > Antonio J. Peï¿½a > > Postdoctoral Appointee > > Mathematics and Computer Science Division > > Argonne National Laboratory > > 9700 South Cass Avenue, Bldg. 240, Of. 3148 > > Argonne, IL 60439-4847 > > apenya(a)mcs.anl.gov > > > > > > > > On Wednesday, May 08, 2013 05:28:06 PM Fernando Barreto wrote: > > HI, my name is Fernando. I'm trying to execute a model in my pc. Mpich2 > is already installed and running (I executed the cpi example with success). > > However when I try to execute my program I get the following error: > Aborting: unable to connect to Boussines, smpd version mismatch > > > > > > > > _______________________________________________ > discuss mailing list discuss(a)mpich.org > To manage subscription options or unsubscribe: > https://lists.mpich.org/mailman/listinfo/discuss > -- Pavan Balaji http://www.mcs.anl.gov/~balaji

1 0

Re: [mpich-discuss] smpd version mismatch
by Antonio J. Peña 08 May '13

08 May '13

Hi Fernando, Is "Boussines" the name of your computer? You may have two versions of MPI in your computer. This may happen, for instance, if you compiled your own version of MPI, and your system already has an MPI installed. Could you please share the exact command-line arguments you're using both for your cpi example which worked, and for that which didn't work. In addition, this previous discussion may help: http://lists.mcs.anl.gov/pipermail/mpich-discuss/2009-February/004656.html -- Antonio J. Peña Postdoctoral Appointee Mathematics and Computer Science Division Argonne National Laboratory 9700 South Cass Avenue, Bldg. 240, Of. 3148 Argonne, IL 60439-4847 apenya(a)mcs.anl.gov On Wednesday, May 08, 2013 05:28:06 PM Fernando Barreto wrote: HI, my name is Fernando. I'm trying to execute a model in my pc. Mpich2 is already installed and running (I executed the cpi example with success). However when I try to execute my program I get the following error: Aborting: unable to connect to Boussines, smpd version mismatch

1 0

smpd version mismatch
by Fernando Barreto 08 May '13

08 May '13

HI, my name is Fernando. I'm trying to execute a model in my pc. Mpich2 is already installed and running (I executed the cpi example with success). However when I try to execute my program I get the following error: Aborting: unable to connect to Boussines, smpd version mismatch

1 0

Re: [mpich-discuss] BLCR kernel module not present
by Pavan Balaji 06 May '13

06 May '13

On 05/01/2013 01:08 PM US Central Time, basma a.azeem wrote: > it works but i need to run these commands every time i shut down > i think that there is a method to have them installed permanently > > i am new to Linux and parallel programming also > > i need to find long running c/mpi applications to test blcr , the > applications i will use > to test blcr should have an enough long execution time , i found many > applications but with a short execution time . Please only use this mailing list for mpich related questions. -- Pavan -- Pavan Balaji http://www.mcs.anl.gov/~balaji

1 0

Re: [mpich-discuss] Issue regarding MPI_Comm_spawn_multiple
by Pavan Balaji 06 May '13

06 May '13

It's impossible to tell the problem from your description below. Can you please send us a small test program that reproduces your problem? -- Pavan On 05/06/2013 06:45 AM US Central Time, Mahesh Doijade wrote: > Hi, > I am using MPI_Comm_spawn_multiple for dynamic process creation > and passing different arguments to each process through this function, > but the code runs perfectly fine for some inputs and throws SEGFAULT for > some other inputs the code sample is > > > /*Arguments generation starts*/ > char ***procs_argv = procs_argv = (char ***) malloc(sizeof(char **) > * no_processes_to_launch); > > for(i=0; i < no_processes_to_launch; i++) > { > procs_argv[i] = ( char **)malloc(sizeof(char *) * 2); > procs_argv[i][0] = ( char *) malloc(sizeof(char)*4); > procs_argv[i][1] = ( char *) malloc(sizeof(char)*10); > sprintf(procs_argv[i][0], "%d", ppstat_ptr[i].no_of_pivots); > //Total no. of pivot variables > sprintf(procs_argv[i][1], "%d", > all_sizes_elems_fep[i+1]);//Total bytes of all sizes data. > } > > /*Arguments generation finish*/ > and process is launched as -- > MPI_Comm_spawn_multiple(no_processes_to_launch, cmds, procs_argv, > nprocs, infos, 0, MPI_COMM_WORLD, &worker_comm, errcodes); > > Error given when debugged through gdb : > > in _IO_vfprintf_internal (s=<value optimized out>, format=<value > optimized out>, ap=<value optimized out>) at vfprintf.c : 1603 , > process_string_arg(((struct printf_spec *)NULL)); > > > > > > -- > > Regards, > -- Mahesh Doijade > > > > > > _______________________________________________ > discuss mailing list discuss(a)mpich.org > To manage subscription options or unsubscribe: > https://lists.mpich.org/mailman/listinfo/discuss > -- Pavan Balaji http://www.mcs.anl.gov/~balaji

1 0

Re: [mpich-discuss] mpich-3.0.4: building shared libraries w/cygwin
by Pavan Balaji 06 May '13

06 May '13

On 05/02/2013 04:36 PM US Central Time, Brad Aagaard wrote: > What configure/make arguments should I use to build shared libraries on > cygwin? Just use --enable-shared without any additional --enable-sharedlibs argument. You might also want to give --disable-static if you don't want static libraries. -- Pavan Balaji http://www.mcs.anl.gov/~balaji

1 0

Re: [mpich-discuss] where is two-phase I/O implemented
by Matthieu Dorier 06 May '13

06 May '13

Thanks Rob. Good to know that it's the messiest part, I managed to find everything I need so I'm confident with the rest of ROMIO :) Matthieu ----- Mail original ----- > De: "Rob Latham" <robl(a)mcs.anl.gov> > À: discuss(a)mpich.org > Envoyé: Lundi 6 Mai 2013 17:38:11 > Objet: Re: [mpich-discuss] where is two-phase I/O implemented > > On Wed, May 01, 2013 at 11:41:52AM +0200, Matthieu Dorier wrote: > > Btw how does Mpich knows which ADIO backend to use, in general? > > (and > > more specifically how does ADIOI_GEN_ReadStridedColl knows that it > > has to use ADIOI_BGL_ReadContig?) Does it have something to do with > > the fact that for instance with PVFS2 when the kernel module is not > > used, the filename should be prefixed by pvfs2:// ? If I want to > > develop a new ADIO backend, how can I make it such that filenames > > prefixed with myfilesystem:// will use this backend? (Maybe there's > > a documentation for that somewhere?) > > To add a bit upon what Dries answered > > Yes, if you prefix the file name, ROMIO will parse that and select > the > right set of function pointers. > > ROMIO will also try to guess the file system type (usually with a > 'statfs' call.) See the functions in adio/common/ad_fstype.c. > > It's a bit of a mess because on some platforms one calls statfs, > others one calls fstatfs, etc. If you squint around the preprocessor > directives I hope it eventually makes sense. Don't hesitate to ask > me > questions -- it's probably the messiest part of ROMIO ... > > ==rob > > > Thanks, > > > > Matthieu > > > > ----- Mail original ----- > > > De: "Pavan Balaji" <balaji(a)mcs.anl.gov> > > > À: discuss(a)mpich.org > > > Cc: "Matthieu Dorier" <matthieu.dorier(a)irisa.fr> > > > Envoyé: Mardi 30 Avril 2013 18:01:11 > > > Objet: Re: [mpich-discuss] where is two-phase I/O implemented > > > > > > > > > Please don't call it mpich3; it's just "mpich" :-). > > > > > > -- Pavan > > > > > > On 04/30/2013 10:32 AM US Central Time, Matthieu Dorier wrote: > > > > Hi, > > > > > > > > A question more for the mpich3 developers: > > > > > > > > If I were to implement an ADIO interface to a file system in > > > > mpich3, > > > > would I have to implement optimizations like two-phase I/O > > > > myself > > > > or are > > > > these optimizations implemented outside ADIO, independently of > > > > the > > > > file > > > > system? > > > > > > > > Thanks, > > > > > > > > Matthieu Dorier > > > > PhD student at ENS Cachan Brittany and IRISA > > > > http://people.irisa.fr/Matthieu.Dorier > > > > > > > > > > > > _______________________________________________ > > > > discuss mailing list discuss(a)mpich.org > > > > To manage subscription options or unsubscribe: > > > > https://lists.mpich.org/mailman/listinfo/discuss > > > > > > > > > > -- > > > Pavan Balaji > > > http://www.mcs.anl.gov/~balaji > > > > > _______________________________________________ > > discuss mailing list discuss(a)mpich.org > > To manage subscription options or unsubscribe: > > https://lists.mpich.org/mailman/listinfo/discuss > > -- > Rob Latham > Mathematics and Computer Science Division > Argonne National Lab, IL USA > _______________________________________________ > discuss mailing list discuss(a)mpich.org > To manage subscription options or unsubscribe: > https://lists.mpich.org/mailman/listinfo/discuss >

1 0

Re: [mpich-discuss] where is two-phase I/O implemented
by Rob Latham 06 May '13

06 May '13

On Wed, May 01, 2013 at 11:41:52AM +0200, Matthieu Dorier wrote: > Btw how does Mpich knows which ADIO backend to use, in general? (and > more specifically how does ADIOI_GEN_ReadStridedColl knows that it > has to use ADIOI_BGL_ReadContig?) Does it have something to do with > the fact that for instance with PVFS2 when the kernel module is not > used, the filename should be prefixed by pvfs2:// ? If I want to > develop a new ADIO backend, how can I make it such that filenames > prefixed with myfilesystem:// will use this backend? (Maybe there's > a documentation for that somewhere?) To add a bit upon what Dries answered Yes, if you prefix the file name, ROMIO will parse that and select the right set of function pointers. ROMIO will also try to guess the file system type (usually with a 'statfs' call.) See the functions in adio/common/ad_fstype.c. It's a bit of a mess because on some platforms one calls statfs, others one calls fstatfs, etc. If you squint around the preprocessor directives I hope it eventually makes sense. Don't hesitate to ask me questions -- it's probably the messiest part of ROMIO ... ==rob > Thanks, > > Matthieu > > ----- Mail original ----- > > De: "Pavan Balaji" <balaji(a)mcs.anl.gov> > > ï¿½: discuss(a)mpich.org > > Cc: "Matthieu Dorier" <matthieu.dorier(a)irisa.fr> > > Envoyï¿½: Mardi 30 Avril 2013 18:01:11 > > Objet: Re: [mpich-discuss] where is two-phase I/O implemented > > > > > > Please don't call it mpich3; it's just "mpich" :-). > > > > -- Pavan > > > > On 04/30/2013 10:32 AM US Central Time, Matthieu Dorier wrote: > > > Hi, > > > > > > A question more for the mpich3 developers: > > > > > > If I were to implement an ADIO interface to a file system in > > > mpich3, > > > would I have to implement optimizations like two-phase I/O myself > > > or are > > > these optimizations implemented outside ADIO, independently of the > > > file > > > system? > > > > > > Thanks, > > > > > > Matthieu Dorier > > > PhD student at ENS Cachan Brittany and IRISA > > > http://people.irisa.fr/Matthieu.Dorier > > > > > > > > > _______________________________________________ > > > discuss mailing list discuss(a)mpich.org > > > To manage subscription options or unsubscribe: > > > https://lists.mpich.org/mailman/listinfo/discuss > > > > > > > -- > > Pavan Balaji > > http://www.mcs.anl.gov/~balaji > > > _______________________________________________ > discuss mailing list discuss(a)mpich.org > To manage subscription options or unsubscribe: > https://lists.mpich.org/mailman/listinfo/discuss -- Rob Latham Mathematics and Computer Science Division Argonne National Lab, IL USA

1 0