Open MPI logo

Open MPI User's Mailing List Archives

  |   Home   |   Support   |   FAQ   |   all Open MPI User's mailing list

Subject: Re: [OMPI users] some mpi processes "disappear" on a cluster of servers
From: George Bosilca (bosilca_at_[hidden])
Date: 2012-09-05 20:40:19


Andrea,

As suggested by the previous answers I guess the size of your problem is too large for the memory available on the nodes. I can runs ZeusMP without any issues up to 64 processes, both over Ethernet and Infiniband. I tried the 1.6 and the current trunk, and both perform as expected.

What is the content of your zmp_inp file?

  george.

On Sep 1, 2012, at 16:01 , Andrea Negri <negri.andre_at_[hidden]> wrote:

> I have tried to run with a single process (i.e. the entire grid is
> contained by one process) and the the command free -m on the compute
> node returns
>
> total used free shared buffers cached
> Mem: 3913 1540 2372 0 49 1234
> -/+ buffers/cache: 257 3656
> Swap: 1983 0 1983
>
>
> while top returns
> top - 16:01:09 up 4 days, 5:56, 1 user, load average: 0.53, 0.16, 0.10
> Tasks: 63 total, 3 running, 60 sleeping, 0 stopped, 0 zombie
> Cpu(s): 49.4% us, 0.7% sy, 0.0% ni, 49.9% id, 0.0% wa, 0.0% hi, 0.0% si
> Mem: 4007720k total, 1577968k used, 2429752k free, 50664k buffers
> Swap: 2031608k total, 0k used, 2031608k free, 1263844k cached
> _______________________________________________
> users mailing list
> users_at_[hidden]
> http://www.open-mpi.org/mailman/listinfo.cgi/users