Singapore-MIT Alliance (SMA) is following suit the trend pointing towards a growing demand for grid computing on corporate networks by successfully deploying a 15-node Linux-based supercomputer from Hewlett Packard (HP). The HydraIII cluster has 60 Intel Itanium-2 Processors with 60GB of memory, with the nodes interconnected by Myrinet at a data rate 2 Gbits/s, one way.

Using LINPACK Benchmark, the cluster achieves a performance of 167GFLOPS, which is only a few GFLOPS away from the 500th supercomputer currently measuring at 195GFLOPS, as listed on the http://www.top500.org/ website.

In a surge to power its research in life and physical sciences, the HP Itanium 2-based systems was chosen for its ability to provide faster processing speed with increased memory capabilities. It is also proven to be the ideal platform for high-performance technical compute clusters. “Itanium 2 should be 100% backwards compatible, allowing easy two-way scalability and migration from IA-32 to IA-64 and vice versa, complimenting our existing systems,” said Associate Professor Khoo Boo Cheong, Programme Chair of High Performance Computation for Engineered System (HPCES) at SMA.

HydraIII cluster supports more than 50 SMA researchers and post-graduate students involved in various projects, ranging from computational fluid dynamics to bioengineering, with the aim of providing users valuable exposure to next-generation technologies.

“The Itanium 2 processor is designed to support vast amounts of data and users and highly complex calculations, making it an important enabling technology for Singapore’s scientific and industrial researchers. The university develops its own software and Linux provides us that flexibility to write our software applications to suit our research needs, making it a cost effective solution for the university,” said Assoc Prof Khoo.

Latest Update!!

The HydraIII cluster has recently became the first large-scale Intel Itanium 2 system to be deployed with the open-source Rocks cluster toolkit, commissioned by The Linux Competency Centre at Singapore Computer Systems (SCS-LCC).

Leading its development, the San-Diego Supercomputer Center installed the NPACI Rocks and had application code running in less than a day. Running on the current Red Hat Linux operating system software, with an upgraded NPACI Rocks version 2.3.2, the cluster can now achieve LINPACK performance of 240GFLOPS.

Updated on 1 August 2003