Please explain the meaning of "#PBS -l nodes=2:ppn=1" and its use in Launching a Distributed-Memory Job?

[AjayBadita]

I followed the same example as mentioned in the link https://devcloud.intel.com/datacenter/learn/advanced-queue/distributed-memory-architecture. But after mentioning the nodes as 2 (#PBS -l nodes=2:ppn=1) in the distrJob2 the output says

Launching the parallel job from mother superior s001-n053...

Hello world from host s001-n053 (rank 1)!

Hello world from host s001-n101 (rank 2)!

Hello world from host s001-n053 (rank 0)!

Hello world from host s001-n101 (rank 3)!

Why is it showing four? Also please explain me the meaning of rank?

[Ansif_M_Intel · Accepted solution]

Hi,

PPN :Process per node

Nodes:Number of nodes requested for the job.

Rank : Serial number given to each process in the MPI job

Here you have given PPN as 1 but devcloud by default reverts it back to 2. You can find the same in your job output logs. In short, your job is launched on 2 nodes with ppn 2 each.i.e. It will run a total of 4 jobs, 2 in each node.

This explain the logs obtained.

Hope this helped.

[Ansif_M_Intel]

Hi,

Thanks for the response.

We would be glad to assist you .Please create a new thread if you face any issues.

After case closure, you will receive a survey email. We appreciate it if you can complete this survey regarding the support you received.

[AjayBadita]

@AnsifM_Intel​  Thanks for the help!