Hello, I am a new bie to SLURM. First of all, thank you for the support you are rendering to the people.
I would like to share few configuration about my cluster configuration. 1. The hyperthreading is disabled at SLURM level in my cluster. Each of the computing node has 4 cores. Hence, no more than 4 tasks can be submitted in one computing node (they are have 4 cores). # COMPUTE NODES NodeName=node[1-6] Procs=4 RealMemory=15360 Sockets=1 CoresPerSocket=4 ThreadsPerCore=1 State=UNKNOWN PartitionName=batch Nodes=node[1-6] Default=YES MaxTime=INFINITE State=UP 2. The /proc/cpuinfo in computing nodes has the following parameters in it CPU 0 1 2 3 4 5 6 7 apicid 16 18 20 22 17 19 21 23 core id 0 1 2 3 0 1 2 3 >From these values, what I interpreted is that both the logical CPUs 0 and 4 >belong to core 0, and so on. 3. The CPU allocation during parallel tests in one computing node (each has 4 cores) is as follows Tasks Thread CPUs allocated Corresponding core id 1 1 1 1 2 1 1,4 1,0 3 1 0,1,4 0,1,0 [ core 0 is overloaded?? I Should it be something like 0,1,2 here ? ] 4 1 0,1,4,5 0,1,0,1 [ core 0 and 1 is overloaded?? Should it be 0,1,3,4?] I use the following bash file to submit the job through slurm, #!/bin/sh #SBATCH --exclusive mpirun -x OMP_NUM_THREADS=1 -np <TASKS> `MYAPPLICATION` 'PARAMETERS` I see a performance drop when going beyond 2 tasks and I attribute this drop to the above reason, that is, SLURM overloading same core with more than one task. If this statement is correct, how to come out of this situation? If not, can someone tell me where things are going wrong. When I directly run with 'mpirun' command on the computing node, the tasks are spanned across individual cores unlike the above case. With many thanks and regards Balachandar The information in this e-mail is confidential. The contents may not be disclosed or used by anyone other than the addressee. Access to this e-mail by anyone else is unauthorised. If you are not the intended recipient, please notify Airbus immediately and delete this e-mail. Airbus cannot accept any responsibility for the accuracy or completeness of this e-mail as it has been sent over public networks. If you have any concerns over the content of this message or its Accuracy or Integrity, please contact Airbus immediately. All outgoing e-mails from Airbus are checked using regularly updated virus scanning software but you should take whatever measures you deem to be appropriate to ensure that this message and any attachments are virus free.
