KYRIC

Description

Five large memory compute nodes dedicated for XSEDE allocation. Each of these nodes have 40 cores (Broadwell class and lntel(R) Xeon(R) CPU E7-4820 v4 @ 2.00GHz with 4 sockets, 10 cores/socket), 3TB RAM, and 6TB SSD storage drives. The 5 dedicated XSEDE nodes will have exclusive access to approximately 300 TB of network attached disk storage. All these compute nodes are interconnected through a 100 Gigabit Ethernet (l00GbE) backbone and the cluster login and data transfer nodes will be connected through a 100Gb uplink to lnternet2 for external connections.

Resource ID
704
Global Resource ID
kyric.uky.access-ci.org
Resource Type
Compute
Latest Status
production
Latest Status Begin
Latest Status End
Project Affiliation
ACCESS
Organization Name
University of Kentucky
RP Description

KyRIC is a large-memory compute cluster operated by the University of Kentucky's Center for Computational Sciences, consisting of five nodes, each with 40 Intel Xeon cores, 3 TB of RAM, and a 5 TB local SSD array. It is particularly well suited for memory-bound work that has to hold an entire dataset in RAM on a single node, and is often used for genome and pangenome assembly, large text-corpus processing, and in-memory graph analytics. It includes a great deal of bioinformatics and data-analytics software, delivered largely through Singularity containers.

MFA Required
Off
External Storage

The five ACCESS compute nodes have exclusive use of approximately 300 TB of network-attached disk storage, reachable from the compute nodes during job execution.

File Transfer Text

Use Globus for anything large or long-running, and scp or rsync for small ad-hoc copies. All three go through the data transfer node, kxc-dtn.ccs.uky.edu, rather than the login nodes. scp and rsync need your SSH key registered first.

If you have not used Globus before, see [How to Log In and Transfer Files with Globus]

 

File Transfer Methods
Transfer Method
Globus
Transfer Method
scp
Data Transfer Node / Globus Collection
kxc-dtn.ccs.uky.edu
Transfer Method
rsync
Data Transfer Node / Globus Collection
kxc-dtn.ccs.uky.edu
Storage Text

KyRIC gives you home, project, and scratch space on shared storage, plus a local SSD scratch area on each compute node. Nothing on this system is backed up at any tier - keep your own copies of anything you cannot regenerate.

Jobs Information

You can run jobs at different sizes and durations on KyRIC. The following lists the different queues that you can submit to, describing how many nodes you get, how long you can run, the type of resources you get, and the average wait time.

KyRIC runs Slurm. Jobs are single-node only - multi-node and MPI jobs are not supported. Submit from a login node with sbatch, and monitor with squeue, scancel, and sinfo. Your job script needs --partition, --time, --account, and --ntasks set explicitly.

KyRIC allocations are made in core-hours. The recommended method for estimating your resource needs for an allocation request is to perform benchmark runs.

KyRIC core-hour calculations should assume that all jobs will run in the normal partition and that they are charged for all 40 cores for the wall-clock time the node is held; if the job exits early, the node is released and charging stops.

Example job scripts and the full job documentation are in the [KyRIC Running Jobs].

*core-hours are cores × wall-clock hours

Storage Filesystems
Directory
Home
File System Path
$HOME
Quota size
10
Purge Policy
Never purged
Backup Policy
Not backed up
Notes
Scripts and configuration only, not for datasets
Directory
Project
File System Path
$PROJECT
Quota size
500
Purge Policy
Never purged
Backup Policy
Not backed up
Notes
Shared group data
Directory
Scratch
File System Path
$SCRATCH
Quota size
10000
Purge Policy
Purged after 30 days
Backup Policy
Not backed up
Notes
Main working area - write job output here rather than to $HOME
Directory
Node-local
Quota size
5000
Purge Policy
Purged at job end
Backup Policy
Not backed up
Notes
Fastest I/O available on the node, and shared by every job on it. Stage data in at job start and copy results out before the job ends
Queue Specifications
Queue Name
normal
Purpose
All production batch work on KyRIC; single-node jobs that need a full large-memory node
CPU Type
Intel Xeon E7-4820 v4
GPU Type
None
CPU Count
40
Node RAM
3000