You are here

Numerical Computing

Printer-friendly versionPrinter-friendly version

DICE is currently based on Scientific Linux. This is a distribution put together by CERN and Fermilab providing a strong platform for numerical computing. Scientific Linux is also the standard Linux platform across the university so anything developed on DICE should be usable on IS and other hardware.

Users at the school have access to:

Hardware

School Facilities

For more information about accessing any of these services please submit a support request

Compute servers
The school maintains a small number of Commodity compute servers for running large computational or large memory jobs. Please use the appropriate server.
Institute servers
We maintain a number of specialist (large memory, database, GPGPU) servers on behalf of the various research institutes and are always happy to give advice on specifying and purchasing new equipment.
Beowulf cluster/Hadoop
The school maintains a small scale cluster for cluster research and to run non commodity cluster software on (currently Hadoop).
GPGPU Computing
We maintain a server dedicated to GPGPU comuting with 4 NVidia GTX 980 cards.
General Parallel File System (GPFS)
GPFS is a parallel filesystem which is available to compute-nodes and servers in our server room. our GPFS filesystem has been replaced in part by a Gluster filesystem.

University Facilities

Registration/Enquiry links are provided, or feel free to submit a Support request.

ECDF/Eddie
ECDF maintains the University's large scale Compute facility (eddie). This is the school's primary resource for High Performance or High Throughput computing. Eddie also includes some provisioning for GPGPU, infiniband and SMP.
EPCC
Supercomputer facilities are provided to the University by EPCC including HECToR (Genomic Analysis, CFD(EBL), Heart Modelling, CP2K and VASP)

Software

Scientific Linux is rich in numercial software and libraries. Language support includes C, GNU FORTRAN, Python, Perl, Java and Erlang. More modern languages include Go and several JavaScript runtimes. It also includes several scientific/numerical languages such as R, MATLAB and Maple.

Commonly-requested libraries include GNU Scientific Library, NumPy, Scipy, ATLAS and CUDA (see GPGPU computing for more details on our CUDA support).

This is far from an exhaustive list. The DICE software page explains how to search the available software and have new software installed.

Last reviewed: 
14/09/2017

System Status

Home dirs (AFS)
Network
Mail
Other services
Scheduled downtime

Choose a topic