Difference between revisions of "Rocky Kivlin"

From NIMBioS
 
(One intermediate revision by the same user not shown)
Line 1: Line 1:
= About =
= About =
Kivlin Lab has purchased compute resources on the [[Rocky User Guide|Rocky cluster]].  These resources are only available to accounts associated with the Kivlin Lab.  Please specify you are associated with the Kivlin Lab when [[Rocky Access Form|requesting an account]].
Kivlin Lab has purchased compute resources on the [[Rocky User Guide|Rocky cluster]].  These resources are only available to accounts associated with the Kivlin Lab.  When [[Rocky Access Form|requesting an account]], please specify your association with the lab to gain access to these resources.




Line 8: Line 8:
{| class="wikitable"
{| class="wikitable"
|-
|-
! Node !! Type !! Architecture !! vCPU !! Memory !! Exclusivity
! Node !! Type !! Architecture !! vCPU !! Memory  
|-
|-
| bull1 || compute || Xeon Gold 6430 (Sapphire Rapids) || 128 || 512G || 2/13/2024 - 2/13/2027
| bull1 || compute || Xeon Gold 6430 (Sapphire Rapids) || 128 || 512G  
|}
|}



Latest revision as of 14:53, 13 February 2024

About

Kivlin Lab has purchased compute resources on the Rocky cluster. These resources are only available to accounts associated with the Kivlin Lab. When requesting an account, please specify your association with the lab to gain access to these resources.


Kivlin Partition

Below are the nodes in the Kivlin Partition:

Node Type Architecture vCPU Memory
bull1 compute Xeon Gold 6430 (Sapphire Rapids) 128 512G


Usage

By default, accounts will use the compute_all partition which includes all shared pool resources. When submitting a job to Rocky, you will need to specify the kivlin partition as part of your batch file.

Example batch file:

#SBATCH --partition=kivlin,compute_all

#SBATCH --job-name=PYTHON_PRIME
#SBATCH --output=python_prime_%j.out
#SBATCH --mail-user=me@test.com
#SBATCH --mail-type=END

module load Python

python3 prime.py

In the above batch file, you can see that we specify the kivlin partition followed by the compute_all partition. With these parameters, the job scheduler will look for available resources on the kivlin partition first. If all those are in use it will then use resources from the shared pool.