For reference and record-keeping, here is the list of decommissioned systems that Pawsey operated in the past.

Magnus

Activity period: 2013 - 2022

Magnus was a Cray XC40 supercomputer, a massively parallel architecture consisting of 1,488 individual nodes that were connected via the Dragonfly network topology by a high-speed network, the HPC-optimised Aries interconnect. Each compute blade had a single Aries ASIC, which provides about 72Gbits/sec for each of the 4 nodes on the blade. All compute nodes in Magnus had the same architecture. Each included two Intel Xeon E5-2690 v3 (Haswell) 12-core CPUs for a total of 24 cores per node, providing a total of 35,712 cores for the entire system. Each node had 64 GB of DDR4 memory shared between 24 cores. Each core had 32KB instruction and data caches, and one 256KB L2 cache; 12 cores (per socket, or NUMA region) share one 30 MB L3 cache. In total, the system had 93 terabytes of memory.

Galaxy

Activity period: 2013 - 2022

Galaxy was a Cray XC30 supercomputer. It was only available for radio-astronomy-focused operations. In particular, it was used to support ASKAP and MWA, which are two of the Square Kilometre Array precursor projects currently underway in the northwest of Western Australia. For ASKAP, Galaxy acted as a real-time computer, allowing direct processing of data delivered to the Pawsey Centre from the Murchison Radio Observatory.

Zeus

Activity period: 2014 - 2022

Zeus was an HPE Linux cluster, containing different types of CPU nodes to support different types of computational workflows:

• For high-throughput workflows requiring large numbers of jobs that each use a modest number of cores.
  • An 80-node partition, workq, was available. Each node had two Intel Xeon E5-2680 v4 2.4 GHz (Broadwell) 14-core CPUs and 128 GB of RAM.
• For long-running workflows requiring several days to complete:
  • An 8-node partition, longq was also available for computationally intensive jobs with wall times up to 4 days. Each node had 28 cores with 128 GB of RAM.
• For large memory workflows requiring more than 128 GB of RAM:
  • A 6-node partition, higmemq, was available for memory intensive computational jobs with 1TB of RAM each. Each node had 16 cores with 1TB of RAM.
• For debugging and development work:
  • An 8-node partition, debugq, was available for code development and prototyping work. Each node had 28 cores with 128 GB of RAM.
• For data transfer jobs:
  • An 8-node partition, copyq, was available for copying and transferring data. Each node contained two Intel Xeon E5-2650 (Sandy Bridge) 8-core CPUs and 64 GB of RAM.

Topaz

Topaz was a commodity 33-node Linux cluster equipped with NVIDIA GPUs. Topaz is dedicated to projects whose workflows can take advantage of the hardware accelerators provided. It was acquired to complement the previous supercomputer Magnus, and now Phase 1 of Setonix, which only possesses CPU computing capabilities. It had two main compute Slurm partitions which provide access to different hardware:

• gpuq : provides access to nodes with two NVIDIA V100 GPUs over PCIe. Nodes in these partitions are configured as shared resources, that is, multiple users can run jobs on the same node.
• nvlinkq: provides access to nodes with four NVIDIA P100 GPUs, interconnected through the NVLINK technology and communicating with the CPU through PCIe. Nodes in these partitions are configured as exclusive resources, meaning that all GPUs (and other resources such as RAM and CPU cores) on a given node are allocated to a single user at a time, and service units are a function of the total GPU-hours per node.

Key characteristics:

• An NVIDIA GPU cluster for accelerated workflows.
• 22 compute nodes with 2 Volta GPUs each.
• 11 compute nodes, each with 4 P100 interconnected with the NVLINK technology.