NCI Announces Receipt of Fastest Australian Supercomputer – To Go Live in November

The National Computational Infrastructure (NCI) of Australia has got a new supercomputer, which it claims to deliver performance that is up to 10 times faster than the supercomputer that is being replaced. The computer, named in the Ngunnawal language as Gadi, which means ‘to search for’, native to the Canberra region is expected to be set up in the Australian National University campus.

Supplied to the organization by Fujitsu, the new supercomputer is expected to become operational in November 2019. According to the manufacturer the newly developed system has been built for the NCI, with the combined contribution of multiple vendors, including Fujitsu.

The National Computational Infrastructure works as a formalized partnership between the Commonwealth Scientific and Industrial Research Organization (CSIRO), the Australian National University (ANU), Geoscience Australia, and the Australian Bureau of Meteorology (BOM). Further, the body is supported by a number of universities that are bolstered by the Australian Research Council.

Project under NCRIS Government Funding Worth AUD 70 Million

The Gadi supercomputer is expected to replace the Raijin, which was also provided to the NCI by Fujitsu. The upgrade from Raijin to Gadi comes after the government of Australia provided federal funding worth AUD 70 million on the basis of the National Collaborative Research Infrastructure Strategy (NCRIS) program.

The older Raijin system which the NCI claimed to be the most powerful super computer in the entire southern hemisphere could provide a capacity of 1.67 petaflops. It was further upgraded by IBM with Power System servers. However, after the Gadi becomes operational, the Raijin will be decommissioned by the NCI.

In comparison the Gadi is equipped with Fujitsu Primergy CX2570 M5 servers, which will function with 3200 nodes. In addition the computer will also work with the aid of Xeon Platinum processors by Intel, V100 GPUs by Nvidia, and Intel’s Optane DC persistent memory. These components will enable deep training, learning, and inferencing features.

Further, according to a statement by Fujitsu, the new system will make use of direct liquid cooling technologies, with warm water, developed by Lenovo Neptune and Fujitsu, that will enable high density computational operations.