Re: [OMPI users] OpenMPI 4 and pmi2 support

[Jeff Squyres (jsquyres) via users]

Noam --

I believe we fixed this issue after v4.0.0 was released.  Can you try the 
v4.0.1rc3 tarball that was just released today?

   https://www.open-mpi.org/software/ompi/v4.0/



> On Mar 22, 2019, at 6:07 PM, Noam Bernstein via users 
>  wrote:
> 
> Hi - I'm trying to compile openmpi 4.0.0 with srun support, so I'm trying to 
> tell openmpi's configure where to find the relevant files by doing
>   $ ./configure --with-verbs --with-ofi --with-pmi=/usr/include/slurm 
> --with-pmi-libdir=/usr/lib64 --prefix=/share/apps/mpi/openmpi/4.0.0/ib/gnu
> verbs and ofi are there from past versions (recycling the commands from 
> openmpi 3.1.X), although I think it's actually using ucx for our IB 
> interfaces.
> 
> The directories I'm specifying are correct:
> [bernadm@tin openmpi-4.0.0]$ ls /usr/include/slurm/pmi2.h
> /usr/include/slurm/pmi2.h
> [bernadm@tin openmpi-4.0.0]$ ls /usr/lib64/libpmi2*
> /usr/lib64/libpmi2.so  /usr/lib64/libpmi2.so.0  /usr/lib64/libpmi2.so.0.0.0
> 
> but when I try to compile, it fails with error:
> 
> make[2]: Entering directory 
> `/home_tin/bernadm/configuration/110_compile_mpi/OpenMPI/openmpi-4.0.0/opal/mca/pmix/s1'
>   CC   mca_pmix_s1_la-pmix_s1.lo
> pmix_s1.c:29:17: fatal error: pmi.h: No such file or directory
>  #include 
>  ^
> compilation terminated.
> make[2]: *** [mca_pmix_s1_la-pmix_s1.lo] Error 1
> make[2]: Leaving directory 
> `/home_tin/bernadm/configuration/110_compile_mpi/OpenMPI/openmpi-4.0.0/opal/mca/pmix/s1'
> make[1]: *** [all-recursive] Error 1
> make[1]: Leaving directory 
> `/home_tin/bernadm/configuration/110_compile_mpi/OpenMPI/openmpi-4.0.0/opal'
> make: *** [all-recursive] Error 1
> [bernadm@tin openmpi-4.0.0]$ less Makefile
> [bernadm@tin openmpi-4.0.0]$ pwd
> /home_tin/bernadm/configuration/110_compile_mpi/OpenMPI/openmpi-4.0.0
> [bernadm@tin openmpi-4.0.0]$ head config.log 
> This file contains any messages produced by compilers while
> running configure, to aid debugging if configure makes a mistake.
> 
> It was created by Open MPI configure 4.0.0, which was
> generated by GNU Autoconf 2.69.  Invocation command line was
> 
> 
> ## - ##
> ## Platform. ##
> 
> poking around config.log and Makefile, I don't see any evidence of anything 
> like "-I/usr/local/slurm", so maybe it's not surprising that compilation 
> fails, but am I supposed to be doing something different/additional to get 
> this to work?
> 
>   thanks,
>   Noam
> 
> ___
> users mailing list
> users@lists.open-mpi.org
> https://lists.open-mpi.org/mailman/listinfo/users


-- 
Jeff Squyres
jsquy...@cisco.com

___
users mailing list
users@lists.open-mpi.org
https://lists.open-mpi.org/mailman/listinfo/users

[OMPI users] OpenMPI 4 and pmi2 support

[Noam Bernstein via users]

Hi - I'm trying to compile openmpi 4.0.0 with srun support, so I'm trying to 
tell openmpi's configure where to find the relevant files by doing
  $ ./configure --with-verbs --with-ofi --with-pmi=/usr/include/slurm 
--with-pmi-libdir=/usr/lib64 --prefix=/share/apps/mpi/openmpi/4.0.0/ib/gnu
verbs and ofi are there from past versions (recycling the commands from openmpi 
3.1.X), although I think it's actually using ucx for our IB interfaces.

The directories I'm specifying are correct:
[bernadm@tin openmpi-4.0.0]$ ls /usr/include/slurm/pmi2.h
/usr/include/slurm/pmi2.h
[bernadm@tin openmpi-4.0.0]$ ls /usr/lib64/libpmi2*
/usr/lib64/libpmi2.so  /usr/lib64/libpmi2.so.0  /usr/lib64/libpmi2.so.0.0.0

but when I try to compile, it fails with error:

make[2]: Entering directory 
`/home_tin/bernadm/configuration/110_compile_mpi/OpenMPI/openmpi-4.0.0/opal/mca/pmix/s1'
  CC   mca_pmix_s1_la-pmix_s1.lo
pmix_s1.c:29:17: fatal error: pmi.h: No such file or directory
 #include 
 ^
compilation terminated.
make[2]: *** [mca_pmix_s1_la-pmix_s1.lo] Error 1
make[2]: Leaving directory 
`/home_tin/bernadm/configuration/110_compile_mpi/OpenMPI/openmpi-4.0.0/opal/mca/pmix/s1'
make[1]: *** [all-recursive] Error 1
make[1]: Leaving directory 
`/home_tin/bernadm/configuration/110_compile_mpi/OpenMPI/openmpi-4.0.0/opal'
make: *** [all-recursive] Error 1
[bernadm@tin openmpi-4.0.0]$ less Makefile
[bernadm@tin openmpi-4.0.0]$ pwd
/home_tin/bernadm/configuration/110_compile_mpi/OpenMPI/openmpi-4.0.0
[bernadm@tin openmpi-4.0.0]$ head config.log 
This file contains any messages produced by compilers while
running configure, to aid debugging if configure makes a mistake.

It was created by Open MPI configure 4.0.0, which was
generated by GNU Autoconf 2.69.  Invocation command line was


## - ##
## Platform. ##

poking around config.log and Makefile, I don't see any evidence of anything 
like "-I/usr/local/slurm", so maybe it's not surprising that compilation fails, 
but am I supposed to be doing something different/additional to get this to 
work?

thanks,
Noam

___
users mailing list
users@lists.open-mpi.org
https://lists.open-mpi.org/mailman/listinfo/users

Re: [OMPI users] _init function being called for every linked OpenMPI library

[Jeff Squyres (jsquyres) via users]

Yes, it's the DLL init function.  It's not in our source code; it's put there 
automatically by the compiler/linker.


> On Mar 22, 2019, at 2:12 PM, Simone Atzeni  wrote:
> 
> Hi,
>  
> I was debugging a program compiled with `mpicxx` and noticed that when the 
> program is being launched the `_init` function gets called multiple time, 
> once for each library linked.
> For example, in my case I see that it gets called from all these libraries:
>  
> _init () from libopen-pal.so.40
> _init () from libopen-rte.so.40
> _init () from libmpi.so.40
> _init () from libmpi_cxx.so.40
>  
> My example program uses OpenMP, so because of the call to `_init` for each 
> library the runtime gets initialized 4 times.
>  
> Is this `_init` function a constructor for each one of this libraries? What 
> does it actually do? I couldn’t find it in the OpenMPI source.
>  
> Thank you!
> Simone
> This email message is for the sole use of the intended recipient(s) and may 
> contain confidential information.  Any unauthorized review, use, disclosure 
> or distribution is prohibited.  If you are not the intended recipient, please 
> contact the sender by reply email and destroy all copies of the original 
> message.
> ___
> users mailing list
> users@lists.open-mpi.org
> https://lists.open-mpi.org/mailman/listinfo/users


-- 
Jeff Squyres
jsquy...@cisco.com

___
users mailing list
users@lists.open-mpi.org
https://lists.open-mpi.org/mailman/listinfo/users

[OMPI users] _init function being called for every linked OpenMPI library

[Simone Atzeni]

Hi,

I was debugging a program compiled with `mpicxx` and noticed that when the 
program is being launched the `_init` function gets called multiple time, once 
for each library linked.
For example, in my case I see that it gets called from all these libraries:

_init () from libopen-pal.so.40
_init () from libopen-rte.so.40
_init () from libmpi.so.40
_init () from libmpi_cxx.so.40

My example program uses OpenMP, so because of the call to `_init` for each 
library the runtime gets initialized 4 times.

Is this `_init` function a constructor for each one of this libraries? What 
does it actually do? I couldn't find it in the OpenMPI source.

Thank you!
Simone

---
This email message is for the sole use of the intended recipient(s) and may 
contain
confidential information.  Any unauthorized review, use, disclosure or 
distribution
is prohibited.  If you are not the intended recipient, please contact the 
sender by
reply email and destroy all copies of the original message.
---
___
users mailing list
users@lists.open-mpi.org
https://lists.open-mpi.org/mailman/listinfo/users

Re: [OMPI users] Are there any issues (performance or otherwise) building apps with different compiler from the one used to build openmpi?

[Aurelien Bouteiller]

You can set the environment variable OMPI_CC to change what compiler mpicc uses 
as the application compiler, irrespective of what compiler Open MPI was built 
with. 

That should work for C applications just fine.

For Fortran programs the story is more complicated, as the MPI Fortran module 
would be tied to the compiler version.


Best,
Aurelien

> On Mar 20, 2019, at 04:26, Daniel Letai  wrote:
> 
> Hello,
> 
> Assuming I have installed openmpi built with distro stock gcc(4.4.7 on rhel 
> 6.5), but an app requires a different gcc version (8.2 manually built on dev 
> machine).
> 
> Would there be any issues, or performance penalty, if building the app using 
> the more recent gcc with flags from wrapper compiler's --showme as 
> perhttps://www.open-mpi.org/faq/?category=mpi-apps#cant-use-wrappers 
>  ?
> Openmpi is built with both pmix and ucx enabled, all built with stock 
> gcc(4.4.7).
> 
> Since the constraint is the app, if the answer is yes I would have to build 
> openmpi using non-distro gcc which is a bit of a hassle.
> 
> Thanks in advance
> --Dani_L.
> ___
> users mailing list
> users@lists.open-mpi.org 
> https://lists.open-mpi.org/mailman/listinfo/users 
> 
___
users mailing list
users@lists.open-mpi.org
https://lists.open-mpi.org/mailman/listinfo/users

Re: [OMPI users] Error initializing an UCX / OpenFabrics device. #6300

[Jeff Squyres (jsquyres) via users]

Greetings Charlie.

Yes, it looks like you replied to a closed issue on Github -- would you mind 
opening a new issue about it?

You can certainly refer to the old issue for context.  But replying to closed 
issues is a bit dangerous: if we miss the initial email from GitHub (and all of 
us have overflowing INBOXes), then we won't see the issue on any GitHub reports 
because we typically only look at open issues / PRs.



> On Mar 22, 2019, at 7:14 AM, Charles A Taylor  wrote:
> 
> Anyone else running into the issue below with OpenMPI 4.0.0?
> 
>https://github.com/open-mpi/ompi/issues/6300  (Error initializing an UCX / 
> OpenFabrics device)
> 
> I’m hitting it and don’t really see why.  I posted to the bug but maybe I 
> need to just open a new issue.
> 
> Charlie Taylor
> UF Research Computing
> ___
> users mailing list
> users@lists.open-mpi.org
> https://lists.open-mpi.org/mailman/listinfo/users


-- 
Jeff Squyres
jsquy...@cisco.com

___
users mailing list
users@lists.open-mpi.org
https://lists.open-mpi.org/mailman/listinfo/users

[OMPI users] Error initializing an UCX / OpenFabrics device. #6300

[Charles A Taylor]

Anyone else running into the issue below with OpenMPI 4.0.0?

   https://github.com/open-mpi/ompi/issues/6300 
  (Error initializing an UCX / 
OpenFabrics device)

I’m hitting it and don’t really see why.  I posted to the bug but maybe I need 
to just open a new issue.

Charlie Taylor
UF Research Computing___
users mailing list
users@lists.open-mpi.org
https://lists.open-mpi.org/mailman/listinfo/users