Facilities
Supercomputing Resources
As a 24 hour per day by 7 day a week facility, the HPCCC selected
and configured its supercomputing equipment to provide 100% availability
100% of the time.
The HPCCC operates a flagship
NEC SX-6,
chosen for its very high sustainable performance on weather and
climate applications, as well as its demonstrated high reliability.
The SX-6 is a commercialised version of the
Japanese Earth Simulator,
one of the most powerful supercomputers in the world today.
The HPCCC NEC SX-6 is a 28 node system with 224 parallel vector processors providing 1.8 TFLOPS (trillions of floating point operations per second) peak performance, and an aggregate 1.8 TB (trillion bytes) of main memory.
The SX-6 is integrated with other HPCCC, Bureau, and CSIRO systems
through multiple gigabit Ethernets and dual NEC TX7/i9510
servers. A 22 TB disk resource is configured to the TX7 servers and hosts the NEC SX-GFS (global file system). The twenty-eight nodes of the SX-6 see the SX-GFS file system as a single coherent system, a feature that greatly simplifies operations and which provides a resilient and robust file service.
|
Photo Credit: Benjamin Healley, Museum Victoria. |
Past Supercomputer Systems
Prior to the establishment of the HPCCC both the Bureau and CSIRO operated
combinations of large scientific mainframes and supercomputers, beginning
in the 1960's. From the middle 1980's both organisations procured and operated
similar vector, and then parallel vector, supercomputing systems.
The formation of the HPCCC introduced common equipment for both organisations.
The main supercomputing systems from 1997 through 2Q2004 included:
| Year | System | Performance |
| 1997 | NEC SX-4/16 | 32 GFLOPS |
| 1997 | NEC SX-4/32 | 64 GFLOPS |
| 1999 | NEC SX-4/32 and SX-5/16 | 192 GFLOPS |
| 2000 | NEC SX-5/32 | 256 GFLOPS |
| 2003 | NEC SX-6/144 | 1.2 TFLOPS |
| 
