2 lsf-hpc equivalents of slurm srun options – HP XC System 3.x Software User Manual

Page 87

Advertising
background image

Table 9-2

LSF-HPC Equivalents of SLURM srun Options

LSF-HPC Equivalent

Description

srun

Option

bsub -n num

Number of processes (tasks) to run.

-n

--ntasks=ntasks

HP XC does not provide this option because the
meaning of this option can be covered by bsub -n
and mincpus=n.

Specifies the number of cores per task. Min
processors per node = MAX(ncpus, mincpus)

-c

--processors-per-task=ntasks

-ext "SLURM[nodes=minmax]"

Specifies the minimum and, optionally,
maximum number of nodes allocated to job.

-N

--nodes=min[-max]

where minmax is min[-max]

The job allocation will contain at least the
minimum number of nodes.

-ext "SLURM[mincpus=n]"

Specifies the minimum number of cores per
node. Min processors per node = MAX(-c
ncpus, --mincpus=n). Default value is 1.

--mincpus=n

-ext "SLURM[mem=MB]"

Specifies a minimum amount of real memory
of each node. By default, job does not
require -ext.

--mem=MB

-ext "SLURM[tmp=MB]"

Specifies a minimum amount of temporary
disk space of each node. By default, job
does not require -ext.

--tmp=MB

-ext "SLURM[constraint=list]"

Specifies a list of constraints. The list may
include multiple features separated by the &

-C

--constraint=list

character (meaning ANDed) or the |
(meaning ORed). By default, job does not
require -ext.

-ext "SLURM[nodelist=node1,...nodeN]"

Requests a specific list of nodes. The job will
at least contain these nodes. The list may be

-w

--nodelist=node1,...nodeN

specified as a comma-separated list of
nodes, or a range of nodes. By default, job
does not require -ext.

-ext "SLURM[exclude=node1,...nodeN]"

Requests that a specific list of hosts be
excluded in the resource allocated to this
job. By default, job does not require.

-x

--exclude=node1,...nodeN

You cannot use this option. The lsf partition

Requests resources from partition
partition

.

-p

--partition=partition

is the only one provided.

-ext "SLURM[contiguous=yes]"

Requests a contiguous range of nodes. By
default, job does not require contiguous
nodes.

--contiguous

bsub -o output_file

Specifies the mode for stdout redirection.

-o

--output=file_name

You can use this option when launching parallel tasks.

Specify the mode for stdout redirection.

-o

--output=none,tasked

bsub -i input_file

Specify how stdin is to be redirected.

-i

--input=file_name

You can use when launching parallel tasks.

Specify how stdin is to be redirected.

-i

--input=none,tasked

bsub -e error_file

Specify how stderr is to be redirected.

-e

--error=file_name

You can use when launching parallel tasks.

Specify how stderr is to be redirected.

-e

--error=none,tasked

You cannot use this option. When creating allocation.
SLURM sets LSF-HPC job id automatically.

Specify a name for the job.

-J

--job-name=job_name

Using LSF-HPC

87

Advertising