December 2013 - discuss - mailman3.cels.anl.gov

Re: [mpich-discuss] Mpich + Siesta erro
by Pavan Balaji 06 Dec '13

06 Dec '13

Hi Julio, There are two steps needed for this: 1. You need to tell your MPI application to return errors instead of aborting. 2. Tell the process manager to not clean up your remaining processes when one of the processes dies. Details on both these steps are listed in the "Fault Tolerance” section of the MPICH README. Please try it out and let us know how it goes. — Pavan On Dec 6, 2013, at 6:54 PM, Julio Henrique <juliohenrique(a)msn.com> wrote: > > I am using mpich-3.0.4 on cluster with 7 nodes running the latest version of siesta. My problem is that when a one node goes down, the siesta and mpich stops running and giveserror. > How do I get when a node falls, siesta and mpich continue to run? > Thank's. > Julio. > > > > _______________________________________________ > 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

Mpich + Siesta erro
by Julio Henrique 06 Dec '13

06 Dec '13

I am using mpich-3.0.4 on cluster with 7 nodes running the latest version of siesta. My problem is that when a one node goes down, the siesta and mpich stops running and gives error. How do I get when a node falls, siesta and mpich continue to run? Thank's. Julio.

1 0

Re: [mpich-discuss] I have a question about using MPICH2 1.4.1p1 for Windows
by Pavan Balaji 06 Dec '13

06 Dec '13

Absolutely. MS-MPI provides full support for Windows. In fact, very recently, Microsoft provided several updates to their stack including making it freely available and distributable. They are working on adding more features as well. Windows is a first-class supported platform for Intel MPI as well. MPICH doesn’t support windows due to lack of developer bandwidth. That’s got nothing to do with how widely the platform is used or supported. — Pavan On Dec 6, 2013, at 4:10 PM, Jeff Squyres (jsquyres) <jsquyres(a)cisco.com> wrote: > On Dec 5, 2013, at 10:55 AM, Wesley Bland <wbland(a)mcs.anl.gov> wrote: > >> The biggest recommendation would be to not use Windows. While there is certainly a community of people who do use Windows for scientific computing, it’s shrinking all of the time. Even Microsoft is de-prioritizing this kind of work. > > Not to put too fine a point on it, but I don't think you should be speaking for Microsoft like that. > > Indeed, if I had to guess, I think Microsoft would vigorously disagree with your assessment. > > -- > Jeff Squyres > jsquyres(a)cisco.com > For corporate legal information go to: http://www.cisco.com/web/about/doing_business/legal/cri/ > > _______________________________________________ > 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] mpifort wrapper compiler?
by Jeff Squyres (jsquyres) 06 Dec '13

06 Dec '13

On Dec 6, 2013, at 11:16 AM, Jeff Hammond <jeff.science(a)gmail.com> wrote: >> Is this MPICH, or IBM MPI? > > That dichotomy is dated, but not particularly relevant for this discussion. Ah -- forgive my ignorance; I thought/assumed IBM periodically seeded from upstream MPICH and was an actual separate MPI implementation distribution. My bad. >> When I install MPICH 3.x on my Linux commodity cluster, I only see mpif77 and mpif90. Does MPICH do something different when installed on a BG system? > > IBM deploys MPICH for BG with a bunch of wrappers, one for each bgxl* > compiler (including an _r variant for each as well). They are named > mpixl* to prevent confusion with the GCC wrappers, at least when some > variants of MPI are used. See > https://wiki.alcf.anl.gov/parts/index.php/Mira_MPI_Documentation#Explanatio… > for details. If you have further questions, request an ALCF account > and you can dig around yourself :-) Good enough. But my original point remains: those wrappers are not in MPICH itself -- they come from IBM. > MPICH does not have these, but I don't see why it would today given > that it only supports the F77 and F90 interfaces. I'm going to be pedantic again, because this is a critical point: there are no such things as "F77" and "F90" interfaces for MPI. There is mpif.h, the mpi module, and the mpi_f08 module. Labeling them as F77 and F90 is incorrect. 1st case in point: mpif.h never worked for Fortran 77. 2nd case in point: you can use mpif.h with any version of Fortran that supports MPI (e.g., 90, 95, 03, 08), and you can use the mpi module with any version of Fortran that supports MPI and modules (e.g., 90, 95, 03, 08). The mpi_f08 module works with F08 and later, obviously (it uses some newer features that debuted in Fortran 2008). My point: They are not F77 and F90 interfaces, since they can be used with many different Fortran versions. Also, MPI states that you should be able to use all 3 interfaces in a single compilation unit. So other than the backwards compatibility issue, a single "mpifort" wrapper that makes all 3 interfaces available makes perfect sense. I don't think there can be any expectation that "mpif77" provides a true, honest, and only Fortran 77 compiler (because MPI never worked with Fortran 77). "mpif90" might have a reasonable expectation that it only provides Fortran 90 syntax, and potentially errors out if you use F03/F08 syntax. But that's kinda weak. -- 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: [mpich-discuss] I have a question about using MPICH2 1.4.1p1 for Windows
by Wesley Bland 06 Dec '13

06 Dec '13

I should back-pedal on some of what I said here as it's not exactly true and wasn't representative of the MPICH project. Microsoft does still have an active MPI team. You can find more information about it from the link on our downloads page. If you want to use Windows and MPI together, you will need to use the MS-MPI version that we have linked from there and contact them for further support. If in the future, you are on a system that we support, you can continue to ask questions here. Thanks, Wesley > On Dec 5, 2013, at 9:55 AM, Wesley Bland <wbland(a)mcs.anl.gov> wrote: > >> On Dec 5, 2013, at 9:33 AM, Ярослав Иванов <ivanov.jaroslaw(a)gmail.com> wrote: >> >> Do you mean http://www.microsoft.com/en-us/download/details.aspx?id=39961 ? > > Yes, that’s what I was referring to. > >> >> Additional question, excuse me. I want to create cluster on common PC's local network. May be I should use PVM instead MPI? Thank you for your reply! > > I wouldn’t recommend it. PVM is a dead project that is very rarely used anymore. If you want support for that there is *nobody* left to ask. MPI is still the most popular way to do internode communication. The biggest recommendation would be to not use Windows. While there is certainly a community of people who do use Windows for scientific computing, it’s shrinking all of the time. Even Microsoft is de-prioritizing this kind of work. You’d be far better served using Linux as that’s where all of the packages you need are available. There are many websites out there that can give you information about how to build a cluster out of common computers. Your favorite search engine is your friend here. > > Wesley > >> >> >> 2013/12/5 Wesley Bland <wbland(a)mcs.anl.gov> >>> Sorry. There is no version that we support for Windows anymore. 1.4.1p “should” work for Windows, but none of us are Windows developers and therefore can’t provide any solutions if a problem arises. There are alternate solutions available on the MPICH downloads website including an MPI implementation written by Microsoft that might be able to do what you’re looking for if the version we provided isn’t working. >>> >>> Wesley >>> >>>> On Dec 5, 2013, at 9:16 AM, Ярослав Иванов <ivanov.jaroslaw(a)gmail.com> wrote: >>>> >>>> So. Can I use http://www.mpich.org/static/downloads/3.0.4/hydra-3.0.4.tar.gz for Windows? I read in ReadMe, that it possible to install "hydra" on Windows (some solution exist in instruction). Do Hydra works on Windows correct? Thank you for your reply, Pavan! >>>> >>>> >>>> 2013/12/5 Pavan Balaji <balaji(a)mcs.anl.gov> >>>>> >>>>> You are welcome to download and use the windows binaries, of course. >>>>> >>>>> What I meant is that if it doesn’t work for you, you’ll need to figure it out on your own. We can’t help you with it. >>>>> >>>>> — Pavan >>>>> >>>>> On Dec 5, 2013, at 7:44 AM, Ярослав Иванов <ivanov.jaroslaw(a)gmail.com> wrote: >>>>> >>>>> > Hello, Pavan. But I use MPICH2 1.4.1p1 for Windows - http://www.mpich.org/static/tarballs/1.4.1p1/mpich2-1.4.1p1-win-ia32.msi >>>>> > >>>>> > I think, this version is for me (if I use Windows). Or http://www.mpich.org/static/tarballs/1.4.1p1/mpich2-1.4.1p1-win-ia32.msi don't support Windows? >>>>> > >>>>> > >>>>> > 2013/12/5 Pavan Balaji <balaji(a)mcs.anl.gov> >>>>> > Hello, >>>>> > >>>>> > Sorry, windows is not supported in mpich anymore. >>>>> > >>>>> > http://wiki.mpich.org/mpich/index.php/Frequently_Asked_Questions#Q:_Why_can… >>>>> > >>>>> > — Pavan >>>>> > >>>>> > On Dec 5, 2013, at 7:10 AM, Ярослав Иванов <ivanov.jaroslaw(a)gmail.com> wrote: >>>>> > >>>>> > > Hello! My name is Jaroslaw, I am student of CSD (Tomsk State University). I am using MPICH2 1.4.1p1 for Windows. It is for my exam work. I have trouble, but I don't know, how to question about my problem on http://lists.mpich.org >>>>> > > >>>>> > > I read: >>>>> > > Using discuss >>>>> > > To post a message to all the list members, send email to discuss(a)mpich.org. >>>>> > > You can subscribe to the list, or change your existing subscription, in the sections below. >>>>> > > >>>>> > > So, do I just send you my question and it display in http://lists.mpich.org ? Or no? >>>>> > > >>>>> > > So, message about problem: >>>>> > > >>>>> > > "Hello! >>>>> > > >>>>> > > I have a question about using MPICH2 1.4.1p1 for Windows. >>>>> > > >>>>> > > http://www.mpich.org/static/tarballs/1.4.1p1/mpich2-1.4.1p1-win-ia32.msi >>>>> > > >>>>> > > I recompilated examples (Visual Studio 2010) about calculating pi. So. If I exec program on pure C - it works fine on network of 2 machines. But, If I exec program on C++ - it don't work on 2 machines (but work on 1 machine). >>>>> > > >>>>> > > When I try to exec C++ version on 2 machines, console return me next issue: >>>>> > > http://pastebin.com/JimcBW2N >>>>> > > >>>>> > > Thank you for your help and I sorry, if I badly use English - it is no my native language." >>>>> > > _______________________________________________ >>>>> > > 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 >>>>> > >>>>> > >>>>> > >>>>> > -- >>>>> > С уважением, Иванов Ярослав >>>>> > _______________________________________________ >>>>> > 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 >>>> >>>> >>>> >>>> -- >>>> С уважением, Иванов Ярослав >>>> _______________________________________________ >>>> discuss mailing list discuss(a)mpich.org >>>> To manage subscription options or unsubscribe: >>>> https://lists.mpich.org/mailman/listinfo/discuss >>> >>> >>> _______________________________________________ >>> discuss mailing list discuss(a)mpich.org >>> To manage subscription options or unsubscribe: >>> https://lists.mpich.org/mailman/listinfo/discuss >> >> >> >> -- >> С уважением, Иванов Ярослав >> _______________________________________________ >> 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] I have a question about using MPICH2 1.4.1p1 for Windows
by Jeff Squyres (jsquyres) 06 Dec '13

06 Dec '13

On Dec 5, 2013, at 10:55 AM, Wesley Bland <wbland(a)mcs.anl.gov> wrote: > The biggest recommendation would be to not use Windows. While there is certainly a community of people who do use Windows for scientific computing, it’s shrinking all of the time. Even Microsoft is de-prioritizing this kind of work. Not to put too fine a point on it, but I don't think you should be speaking for Microsoft like that. Indeed, if I had to guess, I think Microsoft would vigorously disagree with your assessment. -- 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: [mpich-discuss] mpifort wrapper compiler?
by Jeff Hammond 06 Dec '13

06 Dec '13

> Is this MPICH, or IBM MPI? That dichotomy is dated, but not particularly relevant for this discussion. > When I install MPICH 3.x on my Linux commodity cluster, I only see mpif77 and mpif90. Does MPICH do something different when installed on a BG system? IBM deploys MPICH for BG with a bunch of wrappers, one for each bgxl* compiler (including an _r variant for each as well). They are named mpixl* to prevent confusion with the GCC wrappers, at least when some variants of MPI are used. See https://wiki.alcf.anl.gov/parts/index.php/Mira_MPI_Documentation#Explanatio… for details. If you have further questions, request an ALCF account and you can dig around yourself :-) >> By default, XLF does not enforce strict language compliance, but it >> will enforce a wide range of syntax requirements that cause user codes >> to fail to compile. I recall more than one case where Fortran 2003 >> and 2008 differ just slightly and users have to use the compiler >> associated with precisely the language that they are trying to use. > > Fair enough. > > But I'm trying to determine if MPICH really has mpif03 and mpif08, or whether you were using IBM's MPI, or whether you used MPICH's "mpif90" with some appropriate flag to force a particular syntax/dialect. MPICH does not have these, but I don't see why it would today given that it only supports the F77 and F90 interfaces. Whether or not mpif08 is required depends on whether or not the Fortran compiler is indifferent to all syntax between 1990 and 2008+. Jeff -- Jeff Hammond jeff.science(a)gmail.com

1 0

Re: [mpich-discuss] mpifort wrapper compiler?
by Jeff Squyres (jsquyres) 06 Dec '13

06 Dec '13

On Dec 6, 2013, at 9:47 AM, Jeff Hammond <jeff.science(a)gmail.com> wrote: > On Blue Gene: > > #!/bin/bash > echo "mpixlf77 points to bgxlf, which assumes fixed source form and > thus will not compile Fortran 90+ programs i.e. free source form." > for n in 90 95 2003 2008; do > echo "mpixlf$n points bgxlf$n, which assumes free source form and > will likely fail to compile syntax incompatible with Fortran $n" > done Is this MPICH, or IBM MPI? When I install MPICH 3.x on my Linux commodity cluster, I only see mpif77 and mpif90. Does MPICH do something different when installed on a BG system? > By default, XLF does not enforce strict language compliance, but it > will enforce a wide range of syntax requirements that cause user codes > to fail to compile. I recall more than one case where Fortran 2003 > and 2008 differ just slightly and users have to use the compiler > associated with precisely the language that they are trying to use. Fair enough. But I'm trying to determine if MPICH really has mpif03 and mpif08, or whether you were using IBM's MPI, or whether you used MPICH's "mpif90" with some appropriate flag to force a particular syntax/dialect. -- 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: [mpich-discuss] mpifort wrapper compiler?
by Jeff Hammond 06 Dec '13

06 Dec '13

On Fri, Dec 6, 2013 at 8:31 AM, Jeff Squyres (jsquyres) <jsquyres(a)cisco.com> wrote: > On Dec 6, 2013, at 9:16 AM, Jeff Hammond <jeff.science(a)gmail.com> wrote: > >> integer JeffSquyresIsUnfortunatelyRightSometimes > > LOL! > >> The IBM XLF compiler can compile any valid Fortran code by specifying >> the language requested explicitly with a flag. I suppose I can just >> do the following: >> >> alias mpif77='mpifort -gimme-f77' >> alias mpif90='mpifort -gimme-f90' >> alias mpif95='mpifort -gimme-f95' >> alias mpif03='mpifort -gimme-f03' >> alias mpif08='mpifort -gimme-f08' >> >> I just find this annoying since it adds a previously unnecessary step >> to users' workflow and breaks backwards compatibility for anyone that >> relies upon their MPI implementation to provide mpif77. > > A fair point. > > Heh, but you could look at this as your MPI implementation gently prodding you to stop using ancient Fortran dialects. ;-) (yes, I know, I know... it's just a joke) > > However, keep in mind that OMPI (and MPICH) only have mpif77 and mpif90 -- not mpif95, mpif03, or mpif08. > > Just to be clear: are you saying that, for example, on IBM platforms, MPICH's mpif77 and mpif90 enforce F77 and F90 language compliance? > On Blue Gene: #!/bin/bash echo "mpixlf77 points to bgxlf, which assumes fixed source form and thus will not compile Fortran 90+ programs i.e. free source form." for n in 90 95 2003 2008; do echo "mpixlf$n points bgxlf$n, which assumes free source form and will likely fail to compile syntax incompatible with Fortran $n" done By default, XLF does not enforce strict language compliance, but it will enforce a wide range of syntax requirements that cause user codes to fail to compile. I recall more than one case where Fortran 2003 and 2008 differ just slightly and users have to use the compiler associated with precisely the language that they are trying to use. The free/fixed and comment syntax issues are by far the biggest problem, but they are not the only ones. Jeff -- Jeff Hammond jeff.science(a)gmail.com

1 0

Re: [mpich-discuss] mpifort wrapper compiler?
by Jeff Squyres (jsquyres) 06 Dec '13

06 Dec '13

On Dec 6, 2013, at 9:16 AM, Jeff Hammond <jeff.science(a)gmail.com> wrote: > integer JeffSquyresIsUnfortunatelyRightSometimes LOL! > The IBM XLF compiler can compile any valid Fortran code by specifying > the language requested explicitly with a flag. I suppose I can just > do the following: > > alias mpif77='mpifort -gimme-f77' > alias mpif90='mpifort -gimme-f90' > alias mpif95='mpifort -gimme-f95' > alias mpif03='mpifort -gimme-f03' > alias mpif08='mpifort -gimme-f08' > > I just find this annoying since it adds a previously unnecessary step > to users' workflow and breaks backwards compatibility for anyone that > relies upon their MPI implementation to provide mpif77. A fair point. Heh, but you could look at this as your MPI implementation gently prodding you to stop using ancient Fortran dialects. ;-) (yes, I know, I know... it's just a joke) However, keep in mind that OMPI (and MPICH) only have mpif77 and mpif90 -- not mpif95, mpif03, or mpif08. Just to be clear: are you saying that, for example, on IBM platforms, MPICH's mpif77 and mpif90 enforce F77 and F90 language compliance? -- Jeff Squyres jsquyres(a)cisco.com For corporate legal information go to: http://www.cisco.com/web/about/doing_business/legal/cri/

1 0