The content below is taken from the original (Edinburgh University to flog its supercomputer for £0.0369 per core hour), to continue reading please visit the site. Remember to respect the Author & Copyright.
The University of Edinburgh’s supercomputer, Cirrus, is now being rented to businesses for their mega-performance computing needs.
Cirrus is housed at the University’s Advanced Computing Facility at Easter Bush, which also hosts the UK’s national supercomputing service, ARCHER, although it doesn’t really compare to ARCHER’s 118,080 processing cores.
Time on Cirrus is charged at £0.0369 per core hour (exclusive of VAT), although to celebrate Cirrus’ launch the University is offering 1,000 core hours of free use to the first 20 companies that apply to use it.
Customers will be able to use the machine (worth £1,000,000) to tackle whatever challenges are being thrown up by their research or design efforts. The service is fully supported and customers will also have access to Edinburgh Parallel Computing Centre’s (EPCC) consulting expertise in high performance computing and data analytics.
Cirrus is advertised as a “mid-range, industry standard Linux cluster” on which customers can run their own codes or access standard commercial software tools.
In detail, it is an SGI ICE XA cluster with 56 compute nodes which utilises superfast Infiniband interconnect. Each of its compute nodes contains 36 cores, providing 2016 cores in total, and each also hs 256GB RAM. Hyperthreading is also enabled on each node providing a total of 72 threads per node.
There are three login nodes, with identical hardware to the compute nodes, which are provided for general use. Local Lustre storage is provided by a single Lustre filesystem, with 200TB of disk space, while users will also have access to EPCC’s data storage and archiving services.
George Graham, Commercial Manager of EPCC, said: “This newly installed computing power – in tandem with EPCC’s in-house expertise – means we are well placed to help businesses meet many of the computational challenges associated with developing new products and services.” ®
Optimizing the hybrid cloud