Bull to provide memory system to supercomputer

29 Sep 200544 Views

More details have emerged about the technology that will be underpinning a new supercomputer at the €3.3m Irish Centre for High-End Computing (ICHEC).

Last month, siliconrepublic.com reported that NUI Galway, on behalf of the ICHEC, had awarded a tender for a supercomputer to IBM, Bull Information Systems and HEAnet.

Bull has now given more details about its component of the supercomputer – an Itanium-based supercomputer system for the new national research facility.

Bull will supply a Non-Uniform Memory Access-based shared memory system running 32 Itanium II processors at 1.5Ghz and 256GB of RAM along with ancillary storage and a further four-processor test system. This will facilitate the running of code to support research efforts in fields such as computational chemistry, biomechanics and climate modelling.

As the operating system, the system will run Bull Advanced Server, which is an HPC-specific Linux version developed specifically to enhance the HPC working environment over and above standard Linux operating system releases. Future developments to support growth are expected to bring the number of processors to 128 and at the same time quadrupling the direct RAM to 1 terabyte.

Support Silicon Republic

In addition to Bull’s contribution, IBM is providing an integrated storage solution and computer cluster solution, while HEAnet will provide secure floor space, power and air-conditioning for the centre’s machines.

The ICHEC is intended to promote Ireland’s emerging research and development status. The centre is a collaboration of eight different third-level institutions around Ireland and the centre is unique insofar as it is a distributed project operating remotely between Dublin, Cork and Galway. It is funded by Science Foundation Ireland with a grant of €2.6m, a further €0.7m has been given by the Higher Education Authority-funded CosmoGrid project.

The new supercomputer, the single biggest Intel Itanium-based system in Ireland, will not only allow high-performance computing to be applied in many more areas of research but will also give existing users the scope to raise their levels of research, providing increased granularity and capability to run job sizes/routines, which were not feasible on existing systems.

By Brian Skelly