gres.conf(5) Slurm Configuration File gres.conf(5) (NAME) gres.conf - Slurm (GRES) (DESCRIPTION) gres.conf (ASCII) (Generic RESources GRES) . GRES slurm.conf gres.conf . gres.conf (Controller) . slurm.conf . GRES slurm.conf ( <> <> <> GRES ) gres.conf slurm.conf . slurm.conf GRES gres.conf . gres.conf NodeName . Name . . <<#>> () . Slurm SIGHUP <> . : Slurm gres/[mps|shard] select/cons_tres . MPS : https://slurm.schedmd.com/gres.html#MPS_Management Sharding : https://slurm.schedmd.com/gres.html#Sharding GRES : https://slurm.schedmd.com/gres.html : AutoDetect GRES. : full slurmd (NVML NVIDIA RSMI oneAPI NRT) GPU . : (slurmd -Z) gres.conf AutoDetect Name=gpu . nrt AWS Trainium/Inferentia. nvidia NVIDIA. MIG NVlink . Slurm 24.11 . nvml NVIDIA. (NVML) . off GPU. . oneapi (Intel). Intel Graphics Compute Runtime oneAPI Level Zero OpenCL (oneapi) . rsmi AMD. (ROCm SMI) . AutoDetect gres.conf . AutoDetect NodeName . AutoDetect AutoDetect . AutoDetect . . AutoDetect AutoDetect GRES <> : NodeName=tux3 AutoDetect=off Name=gpu File=/dev/nvidia[0-3]. AutoDetect . AutoDetect . File Cores Links AutoDetect (Sanity check) . (Invalid) (Drain) . Count / . File ( ) . <> <> <> <> <

> : <>. Cores * . * . Slurm CPU (Thread) GRES * . GRES * . * SlurmdParameters=numa_node_as_socket NUMA SlurmdParameters=l3cache_as_socket (l3cache) . <<->> ( <<0,1,2,3>> <<0-3>>). --gres-flags=enforce-binding . . --gres-flags=enforce-binding . --gres-flags=disable-binding Slurm . --gres-flags=disable-binding GRES ( GPU ). Slurm cores . Cores slurmctld . : Slurm . . . (Core L#) <> . File . Slurm ( File=/dev/nvidia[0-3]). ( ). (cgroups) task/cgroup Slurm cgroup . Cgroups Slurm : https://slurm.schedmd.com/cgroups.html File Count ( ). MPS/Sharding . GRES GPU File Count GPU . MPS . Sharding GPU . (MIG) MultipleFiles . File MultipleFiles ( ) . : File GRES gpu . : File Slurm . Slurm . : File ( GPU ) (Drain) . GRES . : File Count ( ). Flags GRES . : CountOnly GRES GRES GRES . . : GRES GRES . env_uuid UUID . : CUDA_VISIBLE_DEVICES ROCR_VISIBLE_DEVICES. AutoDetect UUID . <> : <> explicit GRES ( --exclusive Exclusive=NODE ) . : GRES GRES . one_sharing GRES . GRES (mps) GRES (gpu) GRES GRES . MPS . : GRES GRES . all_sharing GRES . all_sharing GRES . one_sharing GRES (gpu) GRES (mps) . : GRES GRES . one_sharing GRES . nvidia_gpu_env CUDA_VISIBLE_DEVICES () . amd_gpu_env ROCR_VISIBLE_DEVICES () . intel_gpu_env ZE_AFFINITY_MASK () . opencl_env GPU_DEVICE_ORDINAL () . no_gpu_env GPU . . nvidia_gpu_env amd_gpu_env intel_gpu_env opencl_env . AutoDetect AutoDetect=nvml AutoDetect=nvidia nvidia_gpu_env AutoDetect=rsmi amd_gpu_env AutoDetect=oneapi intel_gpu_env . AutoDetect . GRES . GRES . AMD ROCm ROCR_VISIBLE_DEVICES CUDA_VISIBLE_DEVICES . Flags=amd_gpu_env AMD ROCR_VISIBLE_DEVICES . Links (Coscheduling) . . - . GRES ( File ). AutoDetect . GPU NVLink . GPU (Minor number) ( X /dev/nvidiaX) / . GPU PCI (PCI bus ID) . : https://slurm.schedmd.com/gres.html#GPU_Management MultipleFiles . (MIG) . ( MultipleFiles=/dev/nvidia[0-3]). MultipleFiles GPU (Drain) . GRES . GPU MIG File . MultipleFiles File ( ) . Name . . GresTypes slurm.conf . . : gpu (GPU) mps (CUDA MPS) nic (NIC) shard GPU (Shards) NodeName NodeName gres.conf () . NodeName Slurm (hostlist) . Type . GPU . Type scontrol reconfig slurmd slurmctld . slurmctld slurmd . : (autodetect) Type gres.conf (underscore) . (EXAMPLES) ################################################################## # Slurm's Generic Resource (GRES) configuration file # Define GPU devices with MPS support, with AutoDetect sanity checking ################################################################## AutoDetect=nvml Name=gpu Type=gtx560 File=/dev/nvidia0 COREs=0,1 Name=gpu Type=tesla File=/dev/nvidia1 COREs=2,3 Name=mps Count=100 File=/dev/nvidia0 COREs=0,1 Name=mps Count=100 File=/dev/nvidia1 COREs=2,3 ################################################################## # Slurm's Generic Resource (GRES) configuration file # Overwrite system defaults and explicitly configure three GPUs ################################################################## Name=gpu Type=tesla File=/dev/nvidia[0-1] COREs=0,1 # Name=gpu Type=tesla File=/dev/nvidia[2-3] COREs=2,3 # NOTE: nvidia2 device is out of service Name=gpu Type=tesla File=/dev/nvidia3 COREs=2,3 ################################################################## # Slurm's Generic Resource (GRES) configuration file # Use a single gres.conf file for all compute nodes - positive method ################################################################## ## Explicitly specify devices on nodes tux0-tux15 # NodeName=tux[0-15] Name=gpu File=/dev/nvidia[0-3] # NOTE: tux3 nvidia1 device is out of service NodeName=tux[0-2] Name=gpu File=/dev/nvidia[0-3] NodeName=tux3 Name=gpu File=/dev/nvidia[0,2-3] NodeName=tux[4-15] Name=gpu File=/dev/nvidia[0-3] ################################################################## # Slurm's Generic Resource (GRES) configuration file # Use NVML to gather GPU configuration information # for all nodes except one ################################################################## AutoDetect=nvml NodeName=tux3 AutoDetect=off Name=gpu File=/dev/nvidia[0-3] ################################################################## # Slurm's Generic Resource (GRES) configuration file # Specify some nodes with NVML, some with RSMI, and some with no AutoDetect ################################################################## NodeName=tux[0-7] AutoDetect=nvml NodeName=tux[8-11] AutoDetect=rsmi NodeName=tux[12-15] Name=gpu File=/dev/nvidia[0-3] ################################################################## # Slurm's Generic Resource (GRES) configuration file # Define 'bandwidth' GRES to use as a way to limit the # resource use on these nodes for workflow purposes ################################################################## NodeName=tux[0-7] Name=bandwidth Type=lustre Count=4G Flags=CountOnly (COPYING) (C) 2010 . (.: ). (C) 2010-2022 SchedMD LLC. Slurm . : Slurm (GNU GPL) ( ) / . Slurm . . (SEE ALSO) slurm.conf(5) Slurm 26.05 Slurm Configuration File gres.conf(5)