Gilles is right, the script shows only what is used right after MPI_Init,
and it will disregard some of the less mainstreams type of modules, the
ones that are dynamically loaded as needed during the execution. It also
only shows only what is related to libmpi, and ignores everything related
to ORTE that is not in use inside the MPI library. However, it does allow
you to define a list of necessary modules, that you can then use during
configure to limit the size of your MPI library.
1. If your goal is to limit the size of the library for a limited set of
applications you can do the following. Instead of generating an app, use
the output of the script to generate a function. You can then link it with
your application(s). Calling the function right before your MPI_Finalize
will allow you to dump the entire list of used modules in your
application(s).
2. During configure use the option --enable-mca-no-build="list" to remove
all unnecessary modules from the build process. The configure will ignore
them, and therefore they will not endup in your libmpi.so
3. Some of the framework are dynamically selected for each communicator or
peer process (e.g. collective and BTL), so it might be difficult and error
prone to trim then down more.
George.
On Wed, Nov 2, 2016 at 12:28 AM, Gilles Gouaillardet <gil...@rist.or.jp>
wrote:
> Did you strip the libraries already ?
>
>
> the script will show the list of frameworks and components used by MPI
> helloworld.
>
> from that, you can deduce a list of components that are not required,
> exclude them via the configure command line, and rebuild a trimmed Open MPI.
>
> note this is pretty painful and incomplete. for example, the ompi/io
> components are not explicitly required by MPI helloworld, but they are
> required
>
> if your app uses MPI-IO (e.g. MPI_File_xxx)
>
> some more components might be dynamically required by realworld MPI app.
>
>
> may i ask why you are focusing on reducing the lib size ?
>
> reducing the lib size by excluding (allegedly) useless components is a
> long and painful process, and you might end up having to debug
>
> new problems by your own ...
>
> as far as i am concerned, if a few MB libs is too big (filesystem ? memory
> ?), i do not see how a real world application can even run on your arm node
>
>
> Cheers,
>
>
> Gilles
> On 11/2/2016 12:49 PM, Mahesh Nanavalla wrote:
>
> HI George,
> Thanks for reply,
>
> By that above script ,how can i reduce* libmpi.so* size.
>
>
>
> On Tue, Nov 1, 2016 at 11:27 PM, George Bosilca <bosi...@icl.utk.edu>
> wrote:
>
>> Let's try to coerce OMPI to dump all modules that are still loaded after
>> MPI_Init. We are still having a superset of the needed modules, but at
>> least everything unnecessary in your particular environment has been
>> trimmed as during a normal OMPI run.
>>
>> George.
>>
>> PS: It's a shell script that needs ag to run. You need to provide the
>> OMPI source directory. You will get a C file (named tmp.c) in the current
>> directory that contain the code necessary to dump all active modules. You
>> will have to fiddle with the compile line to get it to work, as you will
>> need to specify both source and build header files directories. For the
>> sake of completeness here is my compile line
>>
>> mpicc -o tmp -g tmp.c -I. -I../debug/opal/include -I../debug/ompi/include
>> -Iompi/include -Iopal/include -Iopal/mca/event/libevent2022/libevent
>> -Iorte/include -I../debug/opal/mca/hwloc/hwloc1113/hwloc/include
>> -Iopal/mca/hwloc/hwloc1113/hwloc/include -Ioshmem/include -I../debug/
>> -lopen-rte -l open-pal
>>
>>
>>
>> On Tue, Nov 1, 2016 at 7:12 AM, Jeff Squyres (jsquyres) <
>> jsquy...@cisco.com> wrote:
>>
>>> Run ompi_info; it will tell you all the plugins that are installed.
>>>
>>> > On Nov 1, 2016, at 2:13 AM, Mahesh Nanavalla <
>>> mahesh.nanavalla...@gmail.com> wrote:
>>> >
>>> > Hi Jeff Squyres,
>>> >
>>> > Thank you for your reply...
>>> >
>>> > My problem is i want to reduce library size by removing unwanted
>>> plugin's.
>>> >
>>> > Here libmpi.so.12.0.3 size is 2.4MB.
>>> >
>>> > How can i know what are the pluggin's included to build the
>>> libmpi.so.12.0.3 and how can remove.
>>> >
>>> > Thanks&Regards,
>>> > Mahesh N
>>> >
>>> > On Fri, Oct 28, 2016 at 7:09 PM, Jeff Squyres (jsquyres) <
>>> jsquy...@cisco.com> wrote:
>>> > On Oct 28, 2016, at 8:12 AM, Mahesh Nanavalla <
>>> mahesh.nanavalla...@gmail.com> wrote:
>>> > >
>>> > > i have configured as below for arm
>>> > >
>>> > > ./configure --enable-orterun-prefix-by-default
>>> --prefix="/home/nmahesh/Workspace/ARM_MPI/openmpi"
>>> CC=arm-openwrt-linux-muslgnueabi-gcc CXX=arm-openwrt-linux-muslgnueabi-g++
>>> --host=arm-openwrt-linux-muslgnueabi --enable-script-wrapper-compilers
>>> --disable-mpi-fortran --enable-dlopen --enable-shared --disable-vt
>>> --disable-java --disable-libompitrace --disable-static
>>> >
>>> > Note that there is a tradeoff here: --enable-dlopen will reduce the
>>> size of libmpi.so by splitting out all the plugins into separate DSOs
>>> (dynamic shared objects -- i.e., individual .so plugin files). But note
>>> that some of plugins are quite small in terms of code. I mention this
>>> because when you dlopen a DSO, it will load in DSOs in units of pages. So
>>> even if a DSO only has 1KB of code, it will use <page_size> of bytes in
>>> your running process (e.g., 4KB -- or whatever the page size is on your
>>> system).
>>> >
>>> > On the other hand, if you --disable-dlopen, then all of Open MPI's
>>> plugins are slurped into libmpi.so (and friends). Meaning: no DSOs, no
>>> dlopen, no page-boundary-loading behavior. This allows the compiler/linker
>>> to pack in all the plugins into memory more efficiently (because they'll be
>>> compiled as part of libmpi.so, and all the code is packed in there -- just
>>> like any other library). Your total memory usage in the process may be
>>> smaller.
>>> >
>>> > Sidenote: if you run more than one MPI process per node, then
>>> libmpi.so (and friends) will be shared between processes. You're assumedly
>>> running in an embedded environment, so I don't know if this factor matters
>>> (i.e., I don't know if you'll run with ppn>1), but I thought I'd mention it
>>> anyway.
>>> >
>>> > On the other hand (that's your third hand, for those at home
>>> counting...), you may not want to include *all* the plugins. I.e., there
>>> may be a bunch of plugins that you're not actually using, and therefore if
>>> they are compiled in as part of libmpi.so (and friends), they're consuming
>>> space that you don't want/need. So the dlopen mechanism might actually be
>>> better -- because Open MPI may dlopen a plugin at run time, determine that
>>> it won't be used, and then dlclose it (i.e., release the memory that would
>>> have been used for it).
>>> >
>>> > On the other (fourth!) hand, you can actually tell Open MPI to *not*
>>> build specific plugins with the --enable-dso-no-build=LIST configure
>>> option. I.e., if you know exactly what plugins you want to use, you can
>>> negate the ones that you *don't* want to use on the configure line, use
>>> --disable-static and --disable-dlopen, and you'll likely use the least
>>> amount of memory. This is admittedly a bit clunky, but Open MPI's
>>> configure process was (obviously) not optimized for this use case -- it's
>>> much more optimized to the "build everything possible, and figure out which
>>> to use at run time" use case.
>>> >
>>> > If you really want to hit rock bottom on MPI process size in your
>>> embedded environment, you can do some experimentation to figure out exactly
>>> which components you need. You can use repeated runs with "mpirun --mca
>>> ABC_base_verbose 100 ...", where "ABC" is each of Open MPI's framework
>>> names ("framework" = collection of plugins of the same type). This verbose
>>> output will show you exactly which components are opened, which ones are
>>> used, and which ones are discarded. You can build up a list of all the
>>> discarded components and --enable-mca-no-build them.
>>> >
>>> > > While i am running the using mpirun
>>> > > am getting following errror..
>>> > > root@OpenWrt:~# /usr/bin/mpirun --allow-run-as-root -np 1
>>> /usr/bin/openmpiWiFiBulb
>>> > > ------------------------------------------------------------
>>> --------------
>>> > > Sorry! You were supposed to get help about:
>>> > > opal_init:startup:internal-failure
>>> > > But I couldn't open the help file:
>>> > >
>>> > > /home/nmahesh/Workspace/ARM_MPI/openmpi/share/openmpi/help-opal-runtime.txt:
>>> No such file or directory. Sorry!
>>> >
>>> > So this is really two errors:
>>> >
>>> > 1. The help message file is not being found.
>>> > 2. Something is obviously going wrong during opal_init() (which is one
>>> of Open MPI's startup functions).
>>> >
>>> > For #1, when I do a default build of Open MPI 1.10.3, that file *is*
>>> installed. Are you trimming the installation tree, perchance? If so, if
>>> you can put at least that one file back in its installation location (it's
>>> in the Open MPI source tarball), it might reveal more information on
>>> exactly what is failing.
>>> >
>>> > Additionally, I wonder if shared memory is not getting setup right.
>>> Try running with "mpirun --mca shmem_base_verbose 100 ..." and see if it's
>>> reporting an error.
>>> >
>>> > --
>>> > Jeff Squyres
>>> > jsquy...@cisco.com
>>> > For corporate legal information go to: http://www.cisco.com/web/about
>>> /doing_business/legal/cri/
>>> >
>>> > _______________________________________________
>>> > users mailing list
>>> > users@lists.open-mpi.org
>>> > https://rfd.newmexicoconsortium.org/mailman/listinfo/users
>>> >
>>> > _______________________________________________
>>> > users mailing list
>>> > users@lists.open-mpi.org
>>> > https://rfd.newmexicoconsortium.org/mailman/listinfo/users
>>>
>>>
>>> --
>>> Jeff Squyres
>>> jsquy...@cisco.com
>>> For corporate legal information go to: http://www.cisco.com/web/about
>>> /doing_business/legal/cri/
>>>
>>> _______________________________________________
>>> users mailing list
>>> users@lists.open-mpi.org
>>> https://rfd.newmexicoconsortium.org/mailman/listinfo/users
>>>
>>
>>
>> _______________________________________________
>> users mailing list
>> users@lists.open-mpi.org
>> https://rfd.newmexicoconsortium.org/mailman/listinfo/users
>>
>
>
>
> _______________________________________________
> users mailing
> listus...@lists.open-mpi.orghttps://rfd.newmexicoconsortium.org/mailman/listinfo/users
>
>
>
> _______________________________________________
> users mailing list
> users@lists.open-mpi.org
> https://rfd.newmexicoconsortium.org/mailman/listinfo/users
>
_______________________________________________
users mailing list
users@lists.open-mpi.org
https://rfd.newmexicoconsortium.org/mailman/listinfo/users
