The Register® — Biting the hand that feeds IT

Feeds

section icon HPC

Southampton Uni getting Nehalem supercomputer

  • alert
  • print
  • tweet

8000-core IBM iDataplex jobby

Free whitepaper – Enabling Datacenter and Cloud Service Management for Mid-Tier Enterprises

Southampton Uni is getting an IBM Nehalem-based supercomputer.

It's a iDataplex-based system, the first one in the UK public sector, built by UK firm OCF. The university says its 8,000 cores will enable it to run at around 74 trillion calculations (74 teraflops) a second, placing the system amongst the top 100 supercomputers world-wide, based on the June 2009 Top 500 list.

The system will feature 1,000 nodes, thought to be half-depth, 2-socket dx360 M2 ones, 2,000 quad-core Nehalem processors, and 100TB of IBM DS4700 storage. The software includes Cluster Resources' Adaptive HPC Suite, to provide a mixed Linux and Windows workload.

Medical researchers will be one of the main groups of users. Geneticist Professor Andrew Collins talked of: “mapping the disease genes implicated in breast cancer, IBD and glaucoma. With the volume of genome data increasing hugely each year, its analysis requires the most highly-sophisticated facilities.”

The university's Complex Systems Simulation Doctoral Training Centre will use the supercomputer for medical and chemical systems, climate, pharmaceuticals, bioscience, nanoscience, transport, environmental, engineering and computing research. The University Technology Centre for Computational Engineering will also use it for aircraft and aero engine design work.

IBM, which seems to be having some success selling iDataPlexes, will receive £1.8m from the iDataPlex sale, with the contract being worth £3m to OCF. ®

Free whitepaper – Enabling Datacenter and Cloud Service Management for Mid-Tier Enterprises

Spotlight

SC12 Massively parallel 3-D reservoir sims just 1 of the apps in student cluster compo
SC12 If ye would have peace, prepare for cluster WAR
HPC blog You cannae break the laws of physics - and 7nm is the limit
Adapteva's Epiphany-IV chip
Feature Adapteva's parallel dash for community cash
Sysadmin blog I'm not ashamed to admit I was drooling over those racks
SoftLayer's data center
HPC blog US gov travel caps: Agency bods may have to skip supercomputing shows
Webcast Can’t fix it unless you can quantify it
GTC 2012 Hyper-Q and Dynamic Parallelism make GPUs sweat
HPC blog Why I geek out for GTC
HPC blog But what’s Cray going to do with the Intel cash?