Unterschiede

Hier werden die Unterschiede zwischen zwei Versionen angezeigt.

--- hpc:hpc_tubit:slurm_usage:batch_jobs [2019/07/24 14:02] – kraus
+++ hpc:hpc_tubit:slurm_usage:batch_jobs [2020/02/26 10:56] (aktuell) – kraus
@@ Zeile 1: / Zeile 1: @@
+====== Batch Jobs ======
+  * <html>
+<span style="color:black;font-size:110%;font-weight:bold"> Equeueing of scripts launched via: <span style="color:red;font-family:monospace;font-size:140%;white-space:pre">  sbatch job_script</span> </span>
+</html>
+  * <html>
+<span style="color:black;font-size:110%;font-weight:bold">Please DO NOT employ the parameter <span style="font-size:140%;color:red;font-family:monospace"> --ntasks-per-core= </span> unless you really know what you do. Erroneous usage of this flag </br>(especially in combination with the parameter <span style="font-size:140%;color:red;font-family:monospace">--cpus-per-task</span>) can lead to catastrophic errors in your jobs!</span>
+</html>
+<code bash slurm_jobscript_header.sh>
+#!/bin/bash --login
+###################################################################################################
+# WARNING: Please adapt all relevant parameters so that they fit the requirements of your job(s). #
+# Questions and Remarks welcome to Sebastian Kraus                                                #
+###################################################################################################
+# %x: job name; %j: job id; %N: node; %t: task id; %a: array id (and others)
+# #SBATCH -o %x.%j.%N.out  # for debugging purposes: redirect SLURM's stdout (please see man 1 sbatch for explanation of replacement symbols)
+# #SBATCH -e %x.%j.%N.err  # for debugging purposes: redirect SLURM's stderr (please see man 1 sbatch for explanation of replacement symbols)
+# #SBATCH -D $PWD               # if needed change to current working directory
+#SBATCH -J jobname              # job name
+#SBATCH -n 1                    # total number of tasks/cores for your job
+#SBATCH --hint=nomultithread     # IMPORTANT: hyper-threading is activated; switch off and attribute whole core to task
+# #SBATCH --ntasks-per-node=1   # number of tasks per node
+#SBATCH -N 1                    # number of nodes
+# #SBATCH -p smp                # partition the job gets scheduled: standard (default), smp, gpu (uncomment, if you want your job to run on hosts in SMP partition)
+#SBATCH --time=00:15:00         # job run (wall clock) time in HH:MM:SS
+#SBATCH --mem=4GB               # amount of resident main memory PER NODE(!)
+# #SBATCH --mem-per-cpu=1GB     # amount of resident main memory PER CORE(!) (set only, if needed)
+# #SBATCH --gres=gpu:tesla:1    # GPU resources (only with gpu partition!)
+# #SBATCH --mail-type=END       # if your want to receive notifications about job status (cf. man 1 sbatch)
+# #SBATCH --mail-user=username
+# and now your job definition ;-)
+module add [your_modules]
+[your_commands]
+</code>
+  * <html>
+<span style="color:black;font-size:120%;"> SLURM defines lots of environment variables starting with <span style="color:blue;font-size:90%;">SLURM_</span> - e.g. <span style="color:blue;font-size:90%;"> SLURM_JOB_NUM_NODES, SLURM_JOB_ID, SLURM_NTASKS_PER_NODE </span> - you are able to access during the job run (see also <span style="font-family:monospace;font-size:110%;"> man 1 sbatch</span>).</span>
+</html>
+  * <html>
+<span style="color:black;font-size:120%;"> Beware of the fact that <span style="color:blue;font-size:90%;"> SLURM_NTASKS_PER_NODE </span> is allowed to constitute an array of regular expressions and can only easily be parsed in case of a <span style="color:blue;font-size:100%;">fixed number of tasks per node!</span> </span>
+</html>
+\\
+<WRAP rightalign>
+date of revision: 07-24-2019 © kraus
+</WRAP>