srun Option

Description

LSF-HPC Equivalent

--uid=user

Root attempts to submit or run a job as

You cannot use this option. LSF-HPC uses it to create

 

normal user.

allocation.

 

 

 

-t

Establish a time limit to terminate the job

bsub -W runlimit

--time=minutes

after specified number of minutes.

 

 

 

 

 

 

--gid=group

Root attempts to submit or run a job as

You cannot use this option. LSF-HPC uses this option

 

group.

to create allocation.

 

 

 

-A

Allocate resource and spawn a shell.

You cannot use this option. LSF-HPC uses this option

--allocate

 

to create allocation.

 

 

 

 

 

--no-shell

Immediately exit after allocating resources.

You cannot use this option. LSF-HPC uses this option

 

 

to create allocation.

 

 

 

-I

Allocate immediately or fail.

You cannot use this option. LSF-HPC uses this option

--immediate

 

to create allocation.

 

 

-s

Share nodes with other running jobs.

--share

SHARED=FORCE shares all nodes in

 

partition.

 

SHARED=YES shares nodes if and only if

 

–shareis specified.

 

SHARED=NO means do not share the node.

You cannot use this option. LSF-HPC uses this option to create allocation.

-O

Overcommit resources.

Use when launching parallel tasks.

--overcommit

 

 

 

 

 

-b

Submit in “batch mode”.

Meaningless under LSF-HPC.

--batch

 

 

 

 

 

-r

Run a job step relative to node n of the

Use as an argument to srun when launching parallel

--relative=n

current allocation. It is about placing tasks

tasks.

within allocation.

 

 

 

 

 

 

-D

Specify the working directory of the job.

Job starts in job submission directory by default.

--chdir=path

 

 

 

 

 

-k

Do not automatically terminate a job if one

When creating allocation, LSF-HPC does not use the

--no-kill

of the nodes it has been allocated fails.

–koption. SLURM always terminates a job if one of

 

allocated nodes fails

 

 

 

 

 

-T

The srun command uses nthreads to

Use as an argument to srun when launching parallel

--threads=nthreads

initiate and control the parallel job.

tasks.

 

 

-l

Prepend task number to lines of

--label

stdout/stderr.

 

Use as an argument to srun when launching parallel tasks.

-u

Do not line buffer stdout from remote tasks.

Use as an argument to srun when launching parallel

--unbuffered

 

tasks.

 

 

 

 

 

-m

Distribution method for remote processes.

Use as an argument to srun when launching parallel

--distribution=blockcyl

 

tasks.

 

 

 

 

 

--mpi=mpi_type

Identify the type of MPI to be used.

Use as an argument to srun when launching parallel

 

 

tasks.

 

 

 

--jobid=id

Initiate a job step under an already allocated

Do not use. SLURM_JOBID is set in job environment

 

job id.

to associate job with SLURM allocation.

 

 

 

-v

Verbose operation.

Use as an argument to srun when launching parallel

 

 

tasks.

 

 

 

-d

A debug level for slurmd.

Use as an argument to srun when launching parallel

--slurm-debug=level

 

tasks.

 

 

 

 

 

88 Using LSF