[mpich-discuss] mpiexec on 2 windows computer

Jayesh Krishna jayesh at mcs.anl.gov
Wed Jun 27 12:31:03 CDT 2012


 Make sure that you compile the code on a machine that has the *correct* version (same as the nodes where you run your job) of MPICH2 installed.
 I would recommend resolving the MinGW/VS_2010 issues on your machine and recompiling your code.

(PS: Mismatch of env can lead to a lot of errors and it would be difficult to debug it over emails.)
Regards,
Jayesh

----- Original Message -----
From: "Sinta Kartika Maharani" <sintakm114080010 at gmail.com>
To: "Jayesh Krishna" <jayesh at mcs.anl.gov>
Sent: Wednesday, June 27, 2012 11:28:45 AM
Subject: Re: [mpich-discuss] mpiexec on 2 windows computer

I guess because the first time I try to run cpi.exe in path
C:\Program Files\MPICH2\examples> mpiexec -n 2 cpi
Hehe :)
I'm sorry before.

Honestly, I didn't recompile it, because the code was compiling at
MinGW first (compiling the code in my friend's computer). While, I
have Visual Studio 2010. I try compile using VS2010, but error. And I
have some problem when trying to install MinGW in my computer, it
can't installed well.

Thanks

2012/6/27 Jayesh Krishna <jayesh at mcs.anl.gov>:
> Hi,
>  So cpi seems to be working correctly (Any guesses why you were getting the "smpd version mismatch" without full paths to mpiexec and cpi?).
>  Did you recompile your code before running it with the latest version (If not, please recompile your code and let us know if it works for you)?
>
> Regards,
> Jayesh
>
> ----- Original Message -----
> From: "Sinta Kartika Maharani" <sintakm114080010 at gmail.com>
> To: "Jayesh Krishna" <jayesh at mcs.anl.gov>
> Sent: Wednesday, June 27, 2012 11:09:04 AM
> Subject: Re: [mpich-discuss] mpiexec on 2 windows computer
>
> This :
> C:\Program Files\MPICH2\bin> mpiexec -n 2 C:\Program Files\MPICH2\examples\cpi
> Enter the number of intervals: (0 quits)
> (I try choose 1)
> pi is approximately 3.000000000000002, error is 0.584073464102071
> wall clock time = 0.000058
> Enter the number of intervals: (0 quits)
>
> FYI, I can't go to path C:\Program Files\MPICH2\bin\mpiexec .
>
> Thanks.
>
> 2012/6/27 Jayesh Krishna <jayesh at mcs.anl.gov>:
>>  Can you run
>>
>> "C:\Program Files\MPICH2\bin\mpiexec" -n 2 "C:\Program Files\MPICH2\examples\cpi"
>>
>>  on 10.11.121.21 and provide us with the output?
>>
>> Regards,
>> Jayesh
>>
>> ----- Original Message -----
>> From: "Sinta Kartika Maharani" <sintakm114080010 at gmail.com>
>> To: "Jayesh Krishna" <jayesh at mcs.anl.gov>
>> Sent: Wednesday, June 27, 2012 10:43:58 AM
>> Subject: Re: [mpich-discuss] mpiexec on 2 windows computer
>>
>> Hi,
>> I've installed just a version of MPICH2, 1.4.1p1 on all of my node.
>> The status of smpd (in 10.11.121.21) : smpd running in 10.11.121.21. I
>> can't check in other node because the computer was in lab.
>> When the version of MPICH2 is the newest, smpd is running well, what
>> else the problem?
>>
>> Thanks,
>> Sinta
>>
>> 2012/6/27 Jayesh Krishna <jayesh at mcs.anl.gov>:
>>> Hi,
>>>  It looks like you have multiple versions of MPICH2 installed in your system. We do not recommend users to install multiple versions of MPICH2 (You need to uninstall the previous version and install the new version).
>>>  Try specifying the complete path to mpiexec when you run your job (C:\Program Files\MPICH2\bin\mpiexec -n 2 "C:\Program Files\MPICH2\examples\cpi").
>>>  Also check the status of smpd, the process manager, by using the full path (c:\Program Files\MPICH2\bin\smpd -status).
>>>
>>> Regards,
>>> Jayesh
>>>
>>> ----- Original Message -----
>>> From: "Sinta Kartika Maharani" <sintakm114080010 at gmail.com>
>>> To: "Jayesh Krishna" <jayesh at mcs.anl.gov>
>>> Sent: Wednesday, June 27, 2012 10:03:09 AM
>>> Subject: Re: [mpich-discuss] mpiexec on 2 windows computer
>>>
>>> I'm try in node with IP 10.11.121.21 with command "mpiexec -n 2 cpi"
>>> appear warning : "Unable to connect to sinta-PC, smpd version
>>> mismatch. When try it out with command "mpiexec -hosts 2 10.11.121.21
>>> 1 10.11.121.21 1 cpi" appear same warning.
>>> What does it mean with "smpd version mismatch" ?
>>>
>>> Thanks,
>>> Sinta
>>>
>>> 2012/6/27 Jayesh Krishna <jayesh at mcs.anl.gov>:
>>>> Hi,
>>>>  Before running your code (konvensional.exe) can you try out the commands with cpi.exe (c:\Program Files\MPICH2\examples\cpi.exe)?
>>>>
>>>> Regards,
>>>> Jayesh
>>>>
>>>> ----- Original Message -----
>>>> From: "Sinta Kartika Maharani" <sintakm114080010 at gmail.com>
>>>> To: "Jayesh Krishna" <jayesh at mcs.anl.gov>
>>>> Sent: Wednesday, June 27, 2012 9:41:20 AM
>>>> Subject: Re: [mpich-discuss] mpiexec on 2 windows computer
>>>>
>>>> I'm executing the first (failing) command in the node with ip
>>>> 10.11.121.25. MPI jobs run well in this node ( I've try it with
>>>> command "mpiexec -hosts 2 10.11.121.25 1 10.11.121.25 1 konvensional
>>>> random 2" and "mpiexec -hosts 2 10.11.121.21 1 10.11.121.21 1
>>>> konvensional random 2"  or simply command "mpiexec -n 2 konvensional
>>>> random 2" )
>>>> In other words, the node with IP 10.11.121.25 can instruct node with
>>>> IP 10.11.121.21 to executes the MPI jobs, right? And it can also
>>>> instruct itself to executes the MPI jobs.
>>>> I've turned off the Windows firewall in both node.
>>>>
>>>> Sinta.
>>>>
>>>> 2012/6/27 Jayesh Krishna <jayesh at mcs.anl.gov>:
>>>>> Hi,
>>>>>  Which host are you executing the first (failing) command? Do MPI jobs run on that node?
>>>>>  Do you have any firewalls (including Windows firewall) running on the machines? If so please turn off the firewalls on the machines.
>>>>>
>>>>> Regards,
>>>>> Jayesh
>>>>>
>>>>> ----- Original Message -----
>>>>> From: "Sinta Kartika Maharani" <sintakm114080010 at gmail.com>
>>>>> To: mpich-discuss at mcs.anl.gov
>>>>> Sent: Wednesday, June 27, 2012 12:49:05 AM
>>>>> Subject: [mpich-discuss] mpiexec on 2 windows computer
>>>>>
>>>>> Hi, I want to execute a program on 2 computers. but when i try to do
>>>>> it, it can't run. The program was multiplying random matrix with order
>>>>> 2.
>>>>>
>>>>> C:\> mpiexec -hosts 2 10.11.121.25 1 10.11.121.21 1 konvensional random 2
>>>>> Fatal error in MPI_Init: Other error, error stack :
>>>>> MPIR_Init_thread(392).................................:
>>>>> MPID_Init(139)............................................: Channel
>>>>> initialization failed
>>>>> MPIDI_CH3_Init(38).....................................:
>>>>> MPID_nem_init(196)....................................:
>>>>> MPIDI_CH3I_Seg_commit(366).....................:
>>>>> MPIU_SHMW_Hnd_deserialize(324).............:
>>>>> MPIU_SHMW_Seg_open(863)......................:
>>>>> MPIU_SHMW_Seg_create_attach_templ(763): unable to allocate shared
>>>>> memory - OpenFileMapping The sytem cannot find the file specified.
>>>>>
>>>>> job aborted:
>>>>> rank: node: exit: code[:error message]
>>>>> 0: 10.11.121.25: 123
>>>>> 1: 10.11.121.21: 1: process 1 exited without calling finalize
>>>>>
>>>>> but when i use command
>>>>> C:\> mpiexec -hosts 2 10.11.121.25 1 10.11.121.25 1 konvensional random 2
>>>>> or
>>>>> C:\> mpiexec -hosts 2 10.11.121.21 1 10.11.121.21 1 konvensional random 2
>>>>>
>>>>> it can execute well.
>>>>>
>>>>> How to resolve the problem?
>>>>> Thanks
>>>>>
>>>>> Sinta
>>>>> _______________________________________________
>>>>> mpich-discuss mailing list     mpich-discuss at mcs.anl.gov
>>>>> To manage subscription options or unsubscribe:
>>>>> https://lists.mcs.anl.gov/mailman/listinfo/mpich-discuss


More information about the mpich-discuss mailing list