In January 2003, the MERI Department of Sheffield Hallam University approached OCF to help design a Parallel Supercomputer that was scalable not only in terms of performance, but also scalable in terms of budget and growth. To this end OCF architected a solution based around Intel Xeon Processors

The first phase of the system consisted of 2 Nodes connected with the Dolphin SCI High Speed Interconnect. However, the infrastructure allowed the addition of up to 22 Nodes, meaning that as the department received small amounts of funding, they could easily add additional nodes to the cluster with the minimum of fuss

Within the next 3 months the system grew to 18 processors connected with the Dolphin SCI Interconnect with a dedicated, highly resilient Master Node. This growth methodology continued over the following months with the adoption of AMD Opteron processors, and a migration from SCI to an Infiniband Interconnect

The most recent upgrade has added an additional 36 AMD Opteron Cores (based on Opteron 280 Processors) with the Pathscale Infinipath Interconnect and a Voltaire ISR9024 Switch. This upgrade has taken the departments Parallel resource to 96 Cores, with a heterogeneous mix of 32Bit Xeon, x86_64 Single Core Opteron and x86_64 Dual Core Opteron

Please Click Here for Terry Hudson, the MERI Senior Team Leader's, testimonial