Attachment #24325 for ticket #13753

View | Details | Raw Unified | Return to ticket 13753
Collapse All | Expand All

Lines 986-991 Multiple options can be requested in a comma separated list, for example: Link Here

(-)a/doc/man/man1/salloc.1 (-2 / +4 lines)
986	"\-\-gpus=volta:3,kepler:1".	986	"\-\-gpus=volta:3,kepler:1".
987	See also the \fB\-\-gpus\-per\-node\fR, \fB\-\-gpus\-per\-socket\fR and	987	See also the \fB\-\-gpus\-per\-node\fR, \fB\-\-gpus\-per\-socket\fR and
988	\fB\-\-gpus\-per\-task\fR options.	988	\fB\-\-gpus\-per\-task\fR options.
		989	.br
		990	\fBNOTE:\fR The allocation will contain at least one GPU per node.
989	.IP	991	.IP
990		992
991	.TP	993	.TP
Lines 1675-1682 Note that the environment Link Here
1675	variable \fBSLURM_JOB_NUM_NODES\fR will be set to the count of nodes actually	1677	variable \fBSLURM_JOB_NUM_NODES\fR will be set to the count of nodes actually
1676	allocated to the job. See the \fBENVIRONMENT VARIABLES \fR section	1678	allocated to the job. See the \fBENVIRONMENT VARIABLES \fR section
1677	for more information. If \fB\-N\fR is not specified, the default	1679	for more information. If \fB\-N\fR is not specified, the default
1678	behavior is to allocate enough nodes to satisfy the requirements of	1680	behavior is to allocate enough nodes to satisfy requested resources as expressed by
1679	the \fB\-n\fR and \fB\-c\fR options.	1681	per job specification options \fB\-n\fR, \fB\-c\fR and \fB--gpus\fR.
1680	The job will be allocated as many nodes as possible within the range specified	1682	The job will be allocated as many nodes as possible within the range specified
1681	and without delaying the initiation of the job.	1683	and without delaying the initiation of the job.
1682	The node count specification may include a numeric value followed by a suffix	1684	The node count specification may include a numeric value followed by a suffix

Lines 1105-1110 Multiple options can be requested in a comma separated list, for example: Link Here

(-)a/doc/man/man1/sbatch.1 (-2 / +4 lines)
1105	"\-\-gpus=volta:3,kepler:1".	1105	"\-\-gpus=volta:3,kepler:1".
1106	See also the \fB\-\-gpus\-per\-node\fR, \fB\-\-gpus\-per\-socket\fR and	1106	See also the \fB\-\-gpus\-per\-node\fR, \fB\-\-gpus\-per\-socket\fR and
1107	\fB\-\-gpus\-per\-task\fR options.	1107	\fB\-\-gpus\-per\-task\fR options.
		1108	.br
		1109	\fBNOTE:\fR The allocation will contain at least one GPU per node.
1108	.IP	1110	.IP
1109		1111
1110	.TP	1112	.TP
Lines 1784-1791 Note that the environment Link Here
1784	variable \fBSLURM_JOB_NUM_NODES\fR will be set to the count of nodes actually	1786	variable \fBSLURM_JOB_NUM_NODES\fR will be set to the count of nodes actually
1785	allocated to the job. See the \fBENVIRONMENT VARIABLES \fR section	1787	allocated to the job. See the \fBENVIRONMENT VARIABLES \fR section
1786	for more information. If \fB\-N\fR is not specified, the default	1788	for more information. If \fB\-N\fR is not specified, the default
1787	behavior is to allocate enough nodes to satisfy the requirements of	1789	behavior is to allocate enough nodes to satisfy requested resources as expressed by
1788	the \fB\-n\fR and \fB\-c\fR options.	1790	per job specification options \fB\-n\fR, \fB\-c\fR and \fB--gpus\fR.
1789	The job will be allocated as many nodes as possible within the range specified	1791	The job will be allocated as many nodes as possible within the range specified
1790	and without delaying the initiation of the job.	1792	and without delaying the initiation of the job.
1791	The node count specification may include a numeric value followed by a suffix	1793	The node count specification may include a numeric value followed by a suffix

Lines 1359-1364 Multiple options can be requested in a comma separated list, for example: Link Here

(-)a/doc/man/man1/srun.1 (-3 / +4 lines)
1359	"\-\-gpus=volta:3,kepler:1".	1359	"\-\-gpus=volta:3,kepler:1".
1360	See also the \fB\-\-gpus\-per\-node\fR, \fB\-\-gpus\-per\-socket\fR and	1360	See also the \fB\-\-gpus\-per\-node\fR, \fB\-\-gpus\-per\-socket\fR and
1361	\fB\-\-gpus\-per\-task\fR options.	1361	\fB\-\-gpus\-per\-task\fR options.
		1362	.br
		1363	\fBNOTE:\fR The allocation will contain at least one GPU per node.
1362	.IP	1364	.IP
1363		1365
1364	.TP	1366	.TP
Lines 2199-2206 variable \fBSLURM_JOB_NUM_NODES\fR (and \fBSLURM_NNODES\fR for backwards compati Link Here
2199	will be set to the count of nodes actually	2201	will be set to the count of nodes actually
2200	allocated to the job. See the \fBENVIRONMENT VARIABLES\fR section	2202	allocated to the job. See the \fBENVIRONMENT VARIABLES\fR section
2201	for more information. If \fB\-N\fR is not specified, the default	2203	for more information. If \fB\-N\fR is not specified, the default
2202	behavior is to allocate enough nodes to satisfy the requirements of	2204	behavior is to allocate enough nodes to satisfy requested resources as expressed by
2203	the \fB\-n\fR and \fB\-c\fR options.	2205	per job specification options \fB\-n\fR, \fB\-c\fR and \fB--gpus\fR.
2204	The job will be allocated as many nodes as possible within the range specified	2206	The job will be allocated as many nodes as possible within the range specified
2205	and without delaying the initiation of the job.	2207	and without delaying the initiation of the job.
2206	If the number of tasks is given and a number of requested nodes is also given,	2208	If the number of tasks is given and a number of requested nodes is also given,
2207	-

Return to ticket 13753