[mpich-discuss] array_of_argv in the MPI_Comm_spawn_multiple

[Federico Golfrè Andreasi]

I add a printf and also checked using a debugger, but the program stops in
the MPI_Comm_spawn_multiple.

Do you think is a bug or there is a limitation of the argv dimension?
If so can you advise me a possible solution to exchange the argv between the
two program?

thank you very much!

Federico




2009/1/13 Rajeev Thakur <thakur at mcs.anl.gov>

>  Don't know why this may happen. Just do a printf before the
> Comm_spawn_multiple to see if the hang occurs even before that, i.e.,
> outside of MPI.
>
> Rajeev
>
>  ------------------------------
> *From:* mpich-discuss-bounces at mcs.anl.gov [mailto:
> mpich-discuss-bounces at mcs.anl.gov] *On Behalf Of *Federico Golfrè Andreasi
> *Sent:* Monday, January 12, 2009 4:19 AM
>
> *To:* mpich-discuss at mcs.anl.gov
> *Subject:* Re: [mpich-discuss] array_of_argv in the
> MPI_Comm_spawn_multiple
>
> Here I send you a little test program:
>
> - the manager.cpp file contains the manager part that initializes a Petsc
> Session, prints some informations about the command line options and then
> executes the "worker program" with MPI_Comm_spawn_multiple. The firts
> command line option of the master must the executable name of the worker.
> - the worker.cpp file prints informations about the command line options
> and exit.
>
> I use MPICH-2 and Petsc 2.3.3 and the gcc 1.4.2 compiler.
>
>
> To execute the program I type:
>
> mpiexec -n 2 ./manager ./worker -arg1 1 -arg2 2 -arg3 3 -arg4 4 -arg5 5
> -arg6 6 -arg7 7 -arg8 8 -arg9 9 -arg10 10 -arg11 11 -arg12 12 -arg13 13
> -arg14 14 -arg15 15 -arg16 16 -arg17 17 -arg18 18 -arg19 19 -arg20 20 -arg21
> 21 -arg22 22 -arg23 23 -arg24 24 -arg25 25 -arg26 26 -arg27 27 -arg28 28
> -arg29 29 -arg30 30 -arg31 31 -arg32 32 -arg33 33 -arg34 34 -arg35 35 -arg36
> 36
>
> Stopping at -arg35 35 the program executes.
>
> Thank you very much!
> Federico
>
>
>
>
>
> 2009/1/9 Rajeev Thakur <thakur at mcs.anl.gov>
>
>>  Can you send us a small test program that demonstrates the error?
>>
>> Rajeev
>>
>>  ------------------------------
>>  *From:* mpich-discuss-bounces at mcs.anl.gov [mailto:
>> mpich-discuss-bounces at mcs.anl.gov] *On Behalf Of *Federico Golfrè
>> Andreasi
>> *Sent:* Thursday, January 08, 2009 3:34 AM
>> *To:* mpich-discuss at mcs.anl.gov
>> *Subject:* Re: [mpich-discuss] array_of_argv in the
>> MPI_Comm_spawn_multiple
>>
>>   I noticed that problem when the argv has dimension more than about 50
>> elements and 450 characters.
>>
>> Thank you,
>> Federico
>>
>>
>>
>> 2009/1/7 Rajeev Thakur <thakur at mcs.anl.gov>
>>
>>>  How big is the argv you are passing?
>>>
>>> Rajeev
>>>
>>>  ------------------------------
>>> *From:* mpich-discuss-bounces at mcs.anl.gov [mailto:
>>> mpich-discuss-bounces at mcs.anl.gov] *On Behalf Of *Federico Golfrè
>>> Andreasi
>>> *Sent:* Wednesday, January 07, 2009 4:17 AM
>>> *To:* mpich-discuss at mcs.anl.gov
>>> *Subject:* [mpich-discuss] array_of_argv in the MPI_Comm_spawn_multiple
>>>
>>>   Hi,
>>>
>>>
>>>         I've a problem in using the MPI_Comm_spawn_multiple.
>>> When I use that function with a large number of arguments in the
>>> array_of_argv parameter
>>> the program execution freezes in the point.
>>>
>>> There is a maximum length of the argv or a maximum number of elements of
>>> argv that can be used?
>>>
>>> Thank you!
>>> Federico
>>>
>>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/mpich-discuss/attachments/20090114/85ccf024/attachment.htm>