Get Services >  High Performance Computing >  Parallel Computing >  Facilities

Eligibility to Use

We provide both computational and visualisation resources in SVU.  Our supercomputing and visualisation resources are available to all academic staff and student (undergraduate and postgraduate) of NUS.   Visit our HPC Portal for more details.

Our Hardware

Our hardware resources are used by many researchers and research students all over the campus to run computationally intensive and time consuming application programs.  Some hardware are also dedicated to provide remote, collaborative visualisation across the campus network as well as the Internet.

Parallel Computing x86 HPC Linux Clusters
  The Linux Cluster is made up of the following sets of servers:
  • Atlas cluster:  16 Intel servers with 2 Pentium Xeon 3.06GHz processors and 2GB memory on each node. The nodes are connected together via high-speed, low-latency Myrinet interconnects. The combined theoretical performance of the 32-CPU cluster is 195.84 GFlops.  
  • Atlas2 cluster:  22 Intel servers with 2 Pentium Xeon EM64T 3.6GHz processors and 4GB memory on each node. The nodes are connected together via high-speed, low-latency Infiniband interconnects. The combined theoretical performance of the 44 CPU cluster is 316.8 GFlops.
  • Atlas3 cluster:  65 Intel servers with 2 Xeon dual-core 3.0GHz processors and 8GB memory on each node. The nodes are connected together via high-speed, low-latency Infiniband interconnects. Five servers in the cluster are equipped with quad-core processors. The combined theoretical performance of the 65-node cluster is 997.12 GFlops.
  • Atlas4 cluster:  80 Intel servers with 2 Xeon quad-core 2.66GHz processors and 16GB memory on each node, plus 8 Intel servers with 4 Xeon quad-core 2.4GHz processors and 64GB memory on each node. The nodes are connected together via high-speed, low-latency Infiniband interconnects. One of the 4-way quad-core servers is equipped with 128GB memory. The combined theoretical performance of the 88-node cluster is 4019.2 GFlops.

The Linux Cluster is located at the machine room in Computer Centre and School of Computing. They are accessible through its head nodes, which is connected to the campus-wide network.

Parallel Computing HP rx4640 Itanium2 Servers
  The HP Integrity rx4640 server is a 4-way system made up of four 1.5GHz Itanium2 processors with 8GB of RAM (for cougar1, cougar2, cougar3 and cougar4) or four 1.3GHz Itanium2 processors with 4GB of RAM (for cougar5).

This group of servers gives a total (theoretical) performance of 116.8 GFlops of computing power.

The HP rx4640 Itanium2 servers are accessible through the campus-wide network.
Visualisation Remote and Collaborative Visualization Service
 

A virtual visualisation laboratory (MobileVisLab) that allows you to run your favourite modelling software such as Fluent, Abaqus and GaussView, and create high impact and insightful animation using tools such as EnSight and Analyze right from your laptop. The remote and collaborative visualization service is supported by 17 high end graphics servers in a machine room. Each server has two 3.6 GHz Intel Xeon CPUs, 4 GB memory, and Nvidia 3450/4500 graphics cards.

MobileVisLab allows you to access the same full range of services available at the physical visualisation laboratory if you connect to the campus network. It can support more users working remotely, being able to access MobileVisLab services anywhere and anytime increases convenience to users. It also supports collaborative visualization, where two or more collaborators at different locations can view and work on the same graphical model. Besides being used for one-to-one collaborative discussion, it can also be used for group teaching.

 

top