‘shosts’ shows each Compute node’s information (processors, RAM memory etc.) on Andromeda.
To view the free compute nodes’ information on Andromeda, type:
[johnchris@a002 ~]$ shosts -f
- A/I/O/T means Allocated/Idle/Other/Total
- Memory is in GB
- The default Partition has a ‘*’
NODELIST CPUS(A/I/O/T) CPU_LOAD FREE_MEM TOTAL_MEM GRES_USED STATE
c001 43/5/0/48 46.31 30GB 187GB N/A mixed
c002 42/6/0/48 56.12 6GB 187GB N/A mixed
c003 36/12/0/48 21.05 0GB 187GB N/A mixed
c004 36/12/0/48 20.36 0GB 187GB N/A mixed
c005 44/4/0/48 34.13 27GB 187GB N/A allocated
c006 44/4/0/48 16.74 8GB 187GB N/A allocated
c007 42/6/0/48 58.23 38GB 187GB N/A mixed
c008 44/4/0/48 52.06 112GB 187GB N/A allocated
c009 37/11/0/48 10.17 1GB 187GB N/A mixed
c010 44/4/0/48 17.02 7GB 187GB N/A allocated
c011 30/18/0/48 20.90 135GB 187GB N/A mixed
c012 36/12/0/48 11.19 50GB 187GB N/A mixed
c013 44/4/0/48 60.52 95GB 187GB N/A allocated
c014 36/12/0/48 11.30 23GB 187GB N/A mixed
c015 40/8/0/48 134.59 0GB 187GB N/A mixed
...
cb001 92/4/0/96 69.40 225GB 2014GB N/A allocated
cb002 92/4/0/96 86.03 738GB 2014GB N/A allocated
cb003 92/4/0/96 129.96 274GB 2014GB N/A allocated
cb004 92/4/0/96 100.06 1124GB 2014GB N/A allocated
cb005 92/4/0/96 130.30 979GB 2014GB N/A allocated
cb006 92/4/0/96 100.20 253GB 2014GB N/A allocated
cb007 92/4/0/96 130.03 784GB 2014GB N/A allocated
cb008 92/4/0/96 100.13 757GB 2014GB N/A allocated
cb009 92/4/0/96 100.40 666GB 2014GB N/A allocated
cb010 89/7/0/96 117.64 522GB 2014GB N/A mixed
cb011 92/4/0/96 70.27 208GB 2014GB N/A allocated
cb012 92/4/0/96 70.37 1112GB 2014GB N/A allocated
cb013 90/6/0/96 68.37 806GB 2014GB N/A mixed
cb014 92/4/0/96 48.46 685GB 2014GB N/A allocated
cb015 92/4/0/96 98.68 279GB 2014GB N/A allocated
...
g016 0/64/0/64 8.10 149GB 250GB gpu:a100:0(IDX:N/A) idle
g017 0/64/0/64 8.17 204GB 250GB gpu:a100:0(IDX:N/A) idle
g018 0/64/0/64 8.04 200GB 250GB gpu:a100:0(IDX:N/A) idle
g019 0/64/0/64 8.02 235GB 250GB gpu:a100:0(IDX:N/A) idle
gb001 0/64/0/64 8.07 1014GB 2014GB gpu:h200:0(IDX:N/A) idle
PARTITION NODELIST NODES(A/I/O/T) CPUS(A/I/O/T)
interactive c[001-092],g[001-009] 91/7/3/101 3591/1225/144/4960
short* c[001-226],cb[001-020],g[001-019],gb001 235/19/12/266 12534/2906/720/16160
medium c[001-221],cb[001-020],g[001-019],gb001 230/19/12/261 12250/2870/720/15840
long c[001-221],cb[001-020],g[001-019],gb001 230/19/12/261 12250/2870/720/15840
[johnchris@a002 ~]$
We recommend that users use the -f option to check compute node information before submitting jobs to the cluster. This option lists free memory and available processors for each compute node on Andromeda, providing a useful reference for users.
| NODELIST | CPU nodes: c001 to c226, cb001 to cb020 GPU node: g002 to g019, gb001 |
|---|---|
| PARTITION | Name of partitions: interactive, short, medium, long |
| STATE | Possible states: allocated, completing, down, drained, draining, fail, failing, future, idle |
| GRES_USED | Generic resources used: N/A for all CPU nodes; for GPU nodes, it gives GPU type (v100/a100/h200), number of allocated GPUs, and indices of allocated GPUs after the initial “gpu:” character string. |
| TOTAL_M | Total RAM memory for the Compute node: 187GB, 250GB, or 2014GB |
| FREE_MEM | Free RAM memory available on the compute node |
| CPU_LOAD | Total CPU usage (%) |
| CPUS(A/I/O/T) | A: Allocated; I: Idle; O: Other; T: Total. For example: 92/4/0/96 |