menu

The system is managed by the scheduler.

Please refer to the parallel server user manual for basic submitting jobs.

Making a job script

To execute a compiled executable file as a batch job, you need a job script to submit to the job scheduler.

Job Script

Please refer to following link.


After logon to pcc, sample script files can be found in the path below. Please copy it to your own directory before using

pcc: /Samples

Job submission

1. Submit the job script to the scheduler as below.

% qsub -q <Queue> <Script>

 

2. Job submission (with mail notification)
Notification of the starting / ending the job can be requested as below,

% qsub -q <Queue> -M <mail address> -m be <Script>

   -M <contact email>
   -m be    # b -> begin(Job starting) , e -> end(Job ending)

 

3. Job deletion
Only the owner of the job can delete his already submitted jobs

% qdel <Job id>

 

 

Queue classes

For Single Node: (Default Chunk Size:16CPU+64GB)

Queue Size Core MemoryWalltime Number of Simultaneous Job Number of Simultaneous Job(per User)
TINY ~1/8node ~16core ~64GB30min 90job 16job
DEFAULT
~1/2node ~64core ~256GB1week 300job 15job
SINGLE ~1node ~128Core ~512GB1week 150job 10job
LONG ~1node ~128Core ~512GB3week 30job 2job

For Multi Nodes:

Queue Core Walltime Number of Simultaneous Job Number of Simultaneous Job(per User)
SMALL ~640Core 1week 50job 7job
LARGE ~1536Core 1week 20job 3job
XLARGE ~4096Core 5day 7job 1job
LONG-L ~640Core 2week 10job 1job
  • MPI is required for multi-node computation.
    (The job will start without MPI, but the computation will be performed in the single node)

For GPU Nodes:

Queue Size GPUs GPU Walltime
Number of Simultaneous Job
Number of Simultaneous Job(per User)
GPU-1 0.5node 1 NVIDIA A40 7day 30job 4job
GPU-S 1node 1-2 NVIDIA A40 5day 16job 2job
GPU-L 1-4node 1-8 NVIDIA A40 3day 2job 1job
GPU-1A 0.5node 1 NVIDIA A100 7day 20job 2job
GPU-LA1-4node1-8NVIDIA A1003day2job1job