HPC Getting Started

From Storrs HPC Wiki
Revision as of 11:58, 12 February 2016 by Eds08006 (talk | contribs)
Jump to: navigation, search

Overview of cluster nodes

There are 3 classes of nodes available on our HORNET HPC cluster, each named after the Intel CPU architecture, listed in the table below. In terms of CPU cycles, the Westmere and Sandy Bridge nodes are nearly equivalent, since the number of cores per node multiplied with the peak CPU frequency is the same.

Description of CPU compute nodes
Name CPU name Nodes Cores per Node Cores Total RAM (GB) CPU Frequency (GHz) Host Names
Westmere Xeon X5650 60 12 720 48 2.67 cn01 - cn64
Sandy Bridge Xeon E5-2650 40 16 640 64 2.00 cn65 - cn104
Ivy Bridge Xeon E5-2680 32 20 640 128 2.80 cn105 - cn136
Haswell Xeon E5-2690 112 24 2688 128 2.60 cn137 - cn248

The GPU cluster is installed in the Westmere node class. How to submit jobs to the specific class is described here.

Description of GPU compute nodes
GPU name Nodes Cards per Node Cores per Card Cores Total RAM on Card (MB) GPU Frequency (GHz) Host Names
nVidia Tesla M2050 4 8 448 14336 2687 1.15 cn17, cn18, cn35, cn36

Connecting to the cluster

If you don't have an account, learn how to get one here.

To access the cluster resources and send commands, SSH is used. SSH stands for secure shell. It is the industry standard for remote access and command execution.

SSH client software

No installation is necessary for Mac or GNU/Linux, as they come with ssh installed.

Windows users, can use PuTTY.

SSH access

On Mac and GNU/Linux, from the a terminal simply run:

ssh <Your Net ID>@hornet-login3.engr.uconn.edu

Windows users can login using PuTTY.

This gives you access to the login node, cn65, and you should see a terminal prompt like:

[<Your Net ID>@cn65 ~]$ 

Off-campus Access

For security reasons, we limit SSH connections to on-campus addresses from the "UCONN-SECURE" wireless network or the wired network.

Thus, there are 3 ways to connect to HORNET off campus:

  1. VPN: GNU/Linux users can simply install OpenConnect version 7 or later and connect to the VPN with: openconnect --juniper sslvpn.uconn.edu
  2. UConn Skybox: To log into Windows and then access the Hornet cluster via PuTTY in the virtual PC.
  3. Engineering SSH Alternatively, if you have a School of Engineering account, you can first SSH to icarus.engr.uconn.edu, then to the cluster.

Transferring files in and out of the cluster

Before running your code on the cluster, you need to upload your code and data to the cluster first. You can also download the results from the cluster to your local machine. Check our file transfer guide for details.

Submitting Jobs

All job submission, management and scheduling is done using SLURM. To learn more about job submission and management, please read our SLURM Guide.

Always run jobs via SLURM. If you do not, your process will be killed.

Please read our usage policy for more details.

How to Use a Specific Software Application

See the software guides.

Troubleshooting

For any errors, please read FAQ first.