Mercurial > hg > cc > cirrus_home

annotate bin/runme.sh @ 202:491b6d11dd14

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
works now with somewhat redundant --array=a-b and runme.sh -m a b
author Henry S. Thompson <ht@inf.ed.ac.uk>
date Mon, 20 Apr 2026 18:22:38 +0100
parents 93dc3b9f70a6
children 2aa9db279d56
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
167
3213a8bb2ed1 new style batch jobs, see cirrus_work repo for _xxx.sh
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
1 #!/usr/bin/bash
181
55169996c3bf add support for multiple calls to srun with a counter
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 178
diff changeset
2 # Invoke this as e.g. sbatch --ntasks=10 -c 20 runme.sh [-m first last] [args for work/bin/_runme.sh]
167
3213a8bb2ed1 new style batch jobs, see cirrus_work repo for _xxx.sh
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
3 # It will run on the login node
181
55169996c3bf add support for multiple calls to srun with a counter
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 178
diff changeset
4 # -m i j will launch _runme.sh multiple times with a -a in {first..last}
167
3213a8bb2ed1 new style batch jobs, see cirrus_work repo for _xxx.sh
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
5
3213a8bb2ed1 new style batch jobs, see cirrus_work repo for _xxx.sh
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
6 #SBATCH --time=01:00:00
3213a8bb2ed1 new style batch jobs, see cirrus_work repo for _xxx.sh
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
7 #SBATCH --partition=standard
3213a8bb2ed1 new style batch jobs, see cirrus_work repo for _xxx.sh
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
8 #SBATCH --qos=standard
196
436c976ed27f try to get -m working properly
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 181
diff changeset
9 #SBATCH --account=ec276-hst
167
3213a8bb2ed1 new style batch jobs, see cirrus_work repo for _xxx.sh
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
10 #SBATCH --job-name runme
3213a8bb2ed1 new style batch jobs, see cirrus_work repo for _xxx.sh
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
11
3213a8bb2ed1 new style batch jobs, see cirrus_work repo for _xxx.sh
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
12 export W=/work/dc007/dc007
3213a8bb2ed1 new style batch jobs, see cirrus_work repo for _xxx.sh
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
13
181
55169996c3bf add support for multiple calls to srun with a counter
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 178
diff changeset
14 if [ "$1" = "-m" ]
55169996c3bf add support for multiple calls to srun with a counter
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 178
diff changeset
15 then
202
491b6d11dd14 works now with somewhat redundant --array=a-b and runme.sh -m a b
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 201
diff changeset
16 # Can't get this to work: timed out after an hour, and didn't run all 5 jobs
491b6d11dd14 works now with somewhat redundant --array=a-b and runme.sh -m a b
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 201
diff changeset
17 # See slurm_s20-29x5_array.out and comments about that in notes.txt
491b6d11dd14 works now with somewhat redundant --array=a-b and runme.sh -m a b
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 201
diff changeset
18 #echo "-m not supported" 1>&2
491b6d11dd14 works now with somewhat redundant --array=a-b and runme.sh -m a b
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 201
diff changeset
19 #exit 1
201
93dc3b9f70a6 maybe got -m working using --array
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 199
diff changeset
20 a=$SLURM_ARRAY_TASK_ID
93dc3b9f70a6 maybe got -m working using --array
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 199
diff changeset
21 echo "Array id $a $SLURM_ARRAY_TASK_COUNT" > mtest$a.out
93dc3b9f70a6 maybe got -m working using --array
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 199
diff changeset
22 ss=$2-$3
181
55169996c3bf add support for multiple calls to srun with a counter
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 178
diff changeset
23 shift; shift; shift;
201
93dc3b9f70a6 maybe got -m working using --array
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 199
diff changeset
24 echo $(date) Launching $SLURM_JOB_NUM_NODES nodes in $(pwd) for runme -a $ss "$@" from $(hostname) to do $SLURM_NTASKS tasks, $SLURM_CPUS_PER_TASK cpus each >> mtest$a.out
93dc3b9f70a6 maybe got -m working using --array
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 199
diff changeset
25 echo $(date) Launching $a >> mtest$a.out
202
491b6d11dd14 works now with somewhat redundant --array=a-b and runme.sh -m a b
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 201
diff changeset
26 srun -c $SLURM_CPUS_PER_TASK $W/shared/bin/sing $W/hst/bin/_runme.sh -a $a "$@" >> mtest$a.out
201
93dc3b9f70a6 maybe got -m working using --array
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 199
diff changeset
27 echo $(date) Finished $a $? >> mtest$a.out
202
491b6d11dd14 works now with somewhat redundant --array=a-b and runme.sh -m a b
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 201
diff changeset
28 exit 0
201
93dc3b9f70a6 maybe got -m working using --array
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 199
diff changeset
29 if [ "$exclude" ]
93dc3b9f70a6 maybe got -m working using --array
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 199
diff changeset
30 then
93dc3b9f70a6 maybe got -m working using --array
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 199
diff changeset
31 exclude+=",$SLURMD_NODENAME"
93dc3b9f70a6 maybe got -m working using --array
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 199
diff changeset
32 else
93dc3b9f70a6 maybe got -m working using --array
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 199
diff changeset
33 export exclude="--exclude=$SLURMD_NODENAME"
93dc3b9f70a6 maybe got -m working using --array
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 199
diff changeset
34 fi
93dc3b9f70a6 maybe got -m working using --array
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 199
diff changeset
35 echo $exclude >> mtest$a.out
181
55169996c3bf add support for multiple calls to srun with a counter
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 178
diff changeset
36 else
202
491b6d11dd14 works now with somewhat redundant --array=a-b and runme.sh -m a b
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 201
diff changeset
37 echo $(date) Launching $SLURM_JOB_NUM_NODES nodes in $(pwd) for runme "$@" from $(hostname) to do $SLURM_NTASKS tasks, $SLURM_CPUS_PER_TASK cpus each
167
3213a8bb2ed1 new style batch jobs, see cirrus_work repo for _xxx.sh
Henry S. Thompson <ht@inf.ed.ac.uk>
parents:
diff changeset
38
181
55169996c3bf add support for multiple calls to srun with a counter
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 178
diff changeset
39 srun --unbuffered -c $SLURM_CPUS_PER_TASK $W/shared/bin/sing $W/hst/bin/_runme.sh "$@"
55169996c3bf add support for multiple calls to srun with a counter
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 178
diff changeset
40 echo $(date) Finished
55169996c3bf add support for multiple calls to srun with a counter
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 178
diff changeset
41 fi
201
93dc3b9f70a6 maybe got -m working using --array
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 199
diff changeset
42
93dc3b9f70a6 maybe got -m working using --array
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 199
diff changeset
43
93dc3b9f70a6 maybe got -m working using --array
Henry S. Thompson <ht@inf.ed.ac.uk>
parents: 199
diff changeset
44