You can set a partition QoS which specifies a minimum.  We have such a qos on our large-gpu partition; we don’t want people scheduling small stuff to it, so we have this qos:

 

$ sacctmgr show qos large-gpu  --json | jq '.QOS[] | { name: .name, min_limits: .limits.min }'

{

  "name": "large-gpu",

  "min_limits": {

    "priority_threshold": {

      "set": false,

      "infinite": true,

      "number": 0

    },

    "tres": {

      "per": {

        "job": [

          {

            "type": "cpu",

            "name": "",

            "id": 1,

            "count": 32

          },

          {

            "type": "mem",

            "name": "",

            "id": 2,

            "count": 262144

          },

          {

            "type": "gres",

            "name": "gpu",

            "id": 1002,

            "count": 3

          }

        ]

      }

    }

  }

}

 

i.e. the user has to request 32 cores and at least 3 GPUs.  If I try to allocate less I get an error:

 

$ salloc -p large-gpu --gres=gpu:1 -c 32 --mem 256G

salloc: error: QOSMinGRES

salloc: error: Job submit/allocate failed: Job violates accounting/QOS policy (job submit limit, user's size and/or time limits)