CERES is a Linux based computational cluster. It consists of a number of servers configured to work together
and has 1008 processor cores available for computation (1870 threads).
There is a command node (controls the cluster), a number of compute nodes (which do the actual work), and
storage nodes (which store your files). These are all linked by a private high speed network. RAM on the
compute nodes ranges from 512Gb to 1.5Tb.
Users can submit jobs to the cluster which are managed and balanced by a scheduler called
a grid engine or run "interactive sessions" on compute nodes.
A limited number of GPU based servers are also available for scientific / research use
(though Crypto-mining is strictly forbidden).
Why should I use it?
For the right sort of job, CERES allows you to do a large amount of processing
in a relatively short space of time (real life example, a user recently did
2001 days worth of processing in 30 days)
It gives you access to a powerful computational resource without having to
expend your own funds / resources on expensive computers
Will allow you to run analyses you may not otherwise be able to run without
outside help / resources
Students may well encounter CERES during their course as CERES is often used for teaching.
Running a large analysis on the cluster will allow you to continue using your
own computer without it being impacted or affected by the job (i.e. your computer
will not grind to a halt because all its processing power / memory is being
consumed by your analysis - the cluster handles all that).
Who can use CERES?
Any student studying at UoE or UoE member of staff. In exceptional circumstances, access can be
granted to external collaborators working with current staff.
What software does it have?
CERES has hundreds of open source packages installed (word cloud opposite shows a few)
Multiple versions of R
Most open source packages with source code available can be installed for you (if not already)
Some proprietry software such as MATLAB (including the Distributed Computing Server) and the NAG libraries
EEG software (EEGLab & Letswave)
A number of Reference Genomes are also installed (hg,mm,dm,rn) for those working with genomic data
The gcc & gfortran compilers are available for users wishing to compile their own software
Python 2 & 3
If you need software that is not installed, we will attempt to install it for you provided it is available & if needed, a valid license owned.
What knowledge do I need?
Because CERES is built on Linux, you will need to become familiar with
the Linux command line and how to use a grid engine. For users who have
never used a command line before, cluster support can help you with this
either in person or via this online reference guide plus an online tutorial
that covers everything from logging in to submitting multiple batch jobs.
What software do I need?
You will an SSH (secure shell) client to login to the cluster and some
form of SCP (secure copy) in order to transfer files to / from CERES.
If you want to use windowed applications you will also need an X manager.
Depending on your operating system you may already have some or all of these.
How do I use CERES?
Firstly you have to login using your SSH client, then there are two ways of running
jobs on CERES: either interactively which involves using a compute node as though it
were a normal (Linux) computer, or using batch mode where you submit jobs to the grid
engine which runs them for you and reports back to you when they are finished (fire
and forget). Files are transfered to/from the cluster using an SCP client (see above).
Online tutorial can be run as a supervised practical if enough users need it
Cluster support can/will write submission and wrapper scripts for first time users so
you have a working example for the future
Help writing wrapper & submission scripts for all users
Drop in (doors are always open if we are in). 4SW.5.20 (Stuart), 3SW3.12 (Chris)
In order to use CERES, you will need a user account. UoE users will not automatically get an
account on the cluster - you need to obtain one by requesting registration via the
account registration page.