Research IT Services
Published: Oct 16, 2013 11:48:00 AM
Published: Jul 24, 2013 4:47:00 PM
Research Computing Platform Services
Research Computing Platform Services offer a range of advanced platforms for computationally intensive research. These services are currently available to UCL researchers, their partners and collaborators.
RITS' research computing platforms are free for UCL researchers to use on a fair share basis, or as a dedicated resource. They can be used to solve different types of problem:
- The Legion and Iridis distributed computing clusters are designed to allow users to run a wide variety of computational jobs, both serial and parallel. The Legion service also supports programs requiring very large amounts of memory, accelerators and co-processors.
- The Emerald General-Purpose Graphics Processing Unit (GPGPU) cluster is a unique facility within the UK and among the largest GPGPU systems in Europe, with a sustained capability of 114 teraflops (trillion floating-point operations per second).
Iridis and Emerald are shared regional facilities owned by the Centre for Innovation, of which UCL is a founding member.
How the service works
A computer cluster is made up of a collection of processing nodes connected
together by a fast network, effectively operating as a single
system. Legion has over 5,000 processor cores and Iridis nearly 6,000 cores available for UCL use, giving users the ability to run jobs
they cannot perform on desktop or other local systems, or get through workloads that would be impractical on
local resources, whilst being available at a fraction of the cost of a bespoke
a heterogeneous cluster, consisting of different types of nodes which are suited to different types of computational task. Many of the nodes are particularly suited to running ‘parallel’ jobs where a significant
the processing task can be split up and delegated to independent
units running simultaneously. However, other kinds of computational task need to be run on systems that share a very large amount of memory between processors; Legion has three large memory nodes that
support jobs requiring up to 1TB of RAM. All of the nodes have access to the same pre-installed and centrally maintained libraries and software, increasing the flexibility of the system.
Iridis is a large homeogeneous distributed cluster which is reserved for highly parallelised jobs that require more than 36 cores to run.
Emerald has 372 high performance NVIDIA Tesla Graphics Processor Units (GPUs), whose highly parallel architecture can be more efficient than general-purpose CPUs when using specialised software libraries and program. These specially tailored implementations use the hundreds of small computational cores within each graphics chip to perform more than an order of magnitude faster than traditional CPUs. Uses include ray tracing, molecular dynamics, protein folding calculations, machine learning, scientific image processing, statistics, computational finance, large-scale modelling, and evolutionary computation.
Find out more about Emerald
Who should use Research Computing Platform Services?
Researchers from a wide variety of disciplines make use of our Research Computing Platform Services and we are keen to encourage and assist new users from across the faculties. Some examples of research projects currently making use of Legion, Iridis and Emerald are:
- modelling the epidemiology of infectious disease
- modelling wireless telecom networks
- molecular dynamics simulations
- determining the shape of galaxies from astronomical survey images
- modelling evolutionary development of species over time and geography
- combining sets of images and 3D meshes to create textured visualisations of archaeological landscapes from real-world data
- running electronic structure calculations on nanoparticles, using active spaces too large to fit into the memory of most machines
Using Research Computing Platform Services requires some specialist knowledge and familiarity with programming (scripting), batch computing systems and in some cases cloud computing APIs. To help you with this, the Research Computing team also provides:
- regular introductory training courses
- access to locally-hosted specialist programming courses
- extensive user guides (on the main ISD website)
- an email-based user support facility
- specialist advice and consultancy
For further information see Help for UCL researchers
Applications to use Research Computing Platform Services can be made for any research projects carried out at, or led by UCL.
The members of the Research Computing team are:
- Dr Owain Kenway – Research Computing Analyst
- Brian Alston – Research Computing Analyst
- Dr Ian Kirker – Research Computing Analyst
Talk to us
We’d love to hear from you, especially if:
- you would like to discuss any aspect of the services offered by our Research Computing team
- have research computing requirements you’d like to explore with us
Please email email@example.com
(using this email address will ensure that your query is directed to the most appropriate person).
For queries relating to external collaboration or partnership please see our Partner with us page.