I don't think this is an issue with changing anything in LAM/MPI. This
is a queue system issue. What you need is a queue system that supports
dynamic processor allocation. I do not know of one. I have asked the
people at platform computing (the makers of LSF) about this and they
have no plans to incorporate dynamic process functionality into their
system.
Dave.
Prakash Velayutham wrote:
> Hi,
>
> Thanks for the response. In this case, the entire set of nodes that are
> in the LAM universe will not be assigned any other jobs by the scheduler
> (even if they are actually not running any MPI jobs), correct? I want
> the scheduler-MPI interaction to be more dynamic than that. This way I
> could ask for more nodes in between iterations (or several iterations)
> and if the scheduler has gotten some nodes from other completed jobs
> (and there are no other queued jobs (based on priorities and other
> scheduling policies), it could assign them to me and I could run the MPI
> job in more nodes than I started with. After after some iterations, I
> could give back some nodes when the MPI job I am running does not need
> so many.
>
> Can anyone explain to me what can let me do this? If this is currently
> not possible, how much effort would it be for me to modify the LAM
> sources to do this? That could be best determined by one of the
> developers, I guess. The point that I am wondering about is that if this
> cannot be supported, then what is the point in MPI-2 dynamic process
> management in the context of a resource manager? Wondering ...!!!
>
> Regards,
> Prakash
>
>
>>>>Josh Hursey <jjhursey_at_[hidden]> 05/12/05 5:34 PM >>>
>
>
> The TM interface is used in lamboot to create the LAM universe. Once
> the LAM universe has been created, then you can use the MPI-2 dynamic
> process functionality just as if you were outside of the scheduler
> environment. The limitation is that you have to work inside the LAM
> universe that you created during lamboot when creating processes.
>
> However, the TM interface doesn't support the addion/subtraction of
> resources to the LAM universe (such as asking the scheduler for more
> nodes).
>
> Does that address your question?
>
> Cheers,
> Josh
>
> On May 12, 2005, at 1:18 PM, Prakash Velayutham wrote:
>
>
>>Prakash Velayutham wrote:
>>
>>
>>>Hi,
>>>
>>>Is LAM/MPI's MPI-2 features like Dynamic Process Management
>>>integrated with any Scheduler at all? I would like to work on this if
>
>
>>>it does not already exist. Is there a developer's mailing list
>>>seperately available?
>>>
>>>Regards,
>>>Prakash
>>
>>Sorry to be posting again. I am interested to know whether the current
>
>
>>TM interface of OpenPBS/PBSPro/Torque can handle dynamic process
>>creation and management feature of MPI-2. If not, is there a plan to
>>integrate this feature with some scheduler?
>>_______________________________________________
>>This list is archived at http://www.lam-mpi.org/MailArchives/lam/
>>
>
> ----
> Josh Hursey
> jjhursey_at_[hidden]
> http://www.lam-mpi.org/
>
>
> _______________________________________________
> This list is archived at http://www.lam-mpi.org/MailArchives/lam/
>
--
Dr. David Cronk, Ph.D. phone: (865) 974-3735
Research Leader fax: (865) 974-8296
Innovative Computing Lab http://www.cs.utk.edu/~cronk
University of Tennessee, Knoxville
|
