Hi folks!
I am trying to launch *MPI master branch* with srun (simple send/recv
program, see attach) and using *openib*, but unfortunately I get a *segfault
*.
Below is my workflow.
1) I configured ompi/master with following line:
./autogen.sh && ./configure --prefix=$PWD/install --with-openib
What revision level are you at? I just checked and it worked fine for me
On Sep 20, 2013, at 2:33 AM, Victor Kocheganov
wrote:
> Hi folks!
>
> I am trying to launch MPI master branch with srun (simple send/recv program,
> see attach) and using openib, but unfortunately I get a segfault.
>
>
I have HEAD on git revision:
commit 4c282fe5bc8a4143a8c6ac5c0f8d4af591277f6f
Author: Ralph Castain
List-Post: devel@lists.open-mpi.org
Date: Sun Sep 15 15:33:51 2013 +
May be there is a difference in PMI? I have PMI-1 on machine.
On Fri, Sep 20, 2013 at 5:37 PM, Ralph Castain wrote:
>
That looks out-of-date to me - there have been commits all week long. Best
guess is that the git mirror is behind
On Sep 20, 2013, at 6:48 AM, Victor Kocheganov
wrote:
> I have HEAD on git revision:
> commit 4c282fe5bc8a4143a8c6ac5c0f8d4af591277f6f
> Author: Ralph Castain
> Date: Sun Sep 15
Just to close my end of this loop: as of trunk r29213, it all works for me.
Thanks!
On Sep 18, 2013, at 12:52 PM, Ralph Castain wrote:
> Thanks George - much appreciated
>
> On Sep 18, 2013, at 9:49 AM, George Bosilca wrote:
>
>> The test case was broken. I just pushed a fix.
>>
>> George
Short Version:
The OMPI GitHub mirror of the OMPI SVN history currently contains some bad
history, specifically related to the version tags. Prior to roughly September
16th, this repository also contained several other bits of bad
information/history. The git history was *rewound* some time a
"bzero" should be avoided for maximum portability. Just use "memset" instead.
Even older versions of GCC know how to spot the 0 constant and substitute the
right compiler intrinsic(s), assuming they are available for the target
platform.
http://pubs.opengroup.org/onlinepubs/009695399/function
Mellanox --
Please remove all uses of bzero. memset is the portable way to go here.
On Sep 20, 2013, at 12:30 PM, "David Goodell (dgoodell)"
wrote:
> "bzero" should be avoided for maximum portability. Just use "memset"
> instead. Even older versions of GCC know how to spot the 0 constant
On Sep 19, 2013, at 3:07 PM, "Hjelm, Nathan T" wrote:
> I have implemented simple tests for cartesian, graph, and dist graph
> topologies for the Open MPI/IBM test suite and all tests pass. I will push
> those tests to MTT tomorrow.
Consider also grabbing the dist graph and neighborhood collec
Hello,
In Debian, we are in the process of switching the default MPI
implementation from version 1.4 to 1.6.
Every architectures are fine beside ia64. Any program based on OpenMPI
1.6.5 is freezing.
With a basic test case:
MPI_Init(NULL, NULL);
MPI_Finalize();
mpirun -c 4 foo
=> freeze
The
On 20/09/2013 21:49, Sylvestre Ledru wrote:
> Hello,
>
> In Debian, we are in the process of switching the default MPI
> implementation from version 1.4 to 1.6.
>
> Every architectures are fine beside ia64. Any program based on OpenMPI
> 1.6.5 is freezing.
>
> With a basic test case:
> MPI_Init(N
Nathan: can you add these tests (or morphed versions of these tests) into the
ompi-tests/ibm tree? That way, we'll all be running them against MTT.
On Sep 20, 2013, at 12:40 PM, "David Goodell (dgoodell)"
wrote:
> On Sep 19, 2013, at 3:07 PM, "Hjelm, Nathan T" wrote:
>
>> I have implemente
On Fri, Sep 20, 2013 at 08:18:28PM +, Jeff Squyres (jsquyres) wrote:
> Nathan: can you add these tests (or morphed versions of these tests) into the
> ompi-tests/ibm tree? That way, we'll all be running them against MTT.
Looks like that has already been done with at least one of those tests.
Nothing leaps to mind - not seeing that anywhere else. If that's all you saw,
then it doesn't look like it even got to the point of trying to launch
something. You might set the following:
mpirun -mca plm_base_verbose 5 -mca ras_base_verbose 5 -mca rmaps_base_verbose
5 -mca ess_base_verbose 5 f
On 20/09/2013 23:36, Ralph Castain wrote:
> Nothing leaps to mind - not seeing that anywhere else. If that's all you saw,
> then it doesn't look like it even got to the point of trying to launch
> something. You might set the following:
>
> mpirun -mca plm_base_verbose 5 -mca ras_base_verbose 5 -
That's it?? Wow, that was useless. Can you attach to mpirun with gdb and tell
me where it is sitting?
On Sep 20, 2013, at 2:41 PM, Sylvestre Ledru wrote:
> On 20/09/2013 23:36, Ralph Castain wrote:
>> Nothing leaps to mind - not seeing that anywhere else. If that's all you
>> saw, then it does
Hi,
I found a very unexpected behavior with r29217:
% cat ~/.openmpi/mca-params.conf
#pml_base_verbose=0
pml_base_verbose=0
% mpicc -o helloworld helloworld.c
Then if i update the mca-params.conf to have two identical entries, i have
segfaults:
% cat ~/.openmpi/mca-params.conf
pml_base_ver
On 20/09/2013 23:46, Ralph Castain wrote:
> That's it?? Wow, that was useless.
Isn't it ? ;) It is why I asked for your help...
> Can you attach to mpirun with gdb and tell me where it is sitting?
>
It is about as useful at the previous command:
http://paste.debian.net/43882/
Built with:
$ mpicc
Occurs to me - I bet you didn't configure this with --enable-debug, did you? If
not, please reconfigure it and rerun so we can see the debug output
On Sep 20, 2013, at 2:54 PM, Sylvestre Ledru wrote:
> On 20/09/2013 23:46, Ralph Castain wrote:
>> That's it?? Wow, that was useless.
> Isn't it ?
I will look into this.
-Nathan
On Fri, Sep 20, 2013 at 05:49:41PM -0400, Vallee, Geoffroy R. wrote:
> Hi,
>
> I found a very unexpected behavior with r29217:
>
> % cat ~/.openmpi/mca-params.conf
> #pml_base_verbose=0
> pml_base_verbose=0
>
> % mpicc -o helloworld helloworld.c
>
> Then if i up
IU moved the nightly build cron jobs to a new machine today, and they failed.
I'm manually running the build cron jobs on the old build machine (eddie) right
now.
I've alerted IU to what I think the error was in the move; hopefully they'll be
able to fix it over the weekend.
--
Jeff Squyres
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
On 21/09/13 05:49, Sylvestre Ledru wrote:
> Does it ring a bell to anyone ?
Possibly, if you run the binary without mpirun does it do the same?
If so, could you try and run it with strace -f and see if you see
repeating SEGV's?
cheers!
Chris
- --
22 matches
Mail list logo
