Ngarrgu Tindebeek

The latest Swinburne supercomputer, installed in 2023, Ngarrgu Tindebeek offers expanded computational capacity to Swinburne staff and students, Victorian researchers and the national astronomy community.

What's in a name?

The name Ngarrgu Tindebeek was provided by Wurundjeri elders through the assistance of the Moondani Toombadool Centre at Swinburne. It translates as “Knowledge of the Void” in the local Woiwurrung language and represents the goal of harnessing the power of a supercomputer to enable researchers to explain the unknown, to push the boundaries of knowledge. Understanding black holes – how they come together within galaxies, collide and create gravitational waves – is a key use case and a prime example.

Technical Specifications

Ngarrgu Tindebeek is the latest addition to the supercomputer environment at Swinburne. Funding for the hardware was provided by the Victorian Higher Education State Investment Fund (VHESIF) with operational support to be provided by Swinburne IT, the Centre for Astrophysics & Supercomputing (CAS), and funding from the National Collaborative Research Infrastructure Strategy (NCRIS) through Astronomy Australia Limited (AAL).

Installed in 2022/23, Ngarrgu Tindebeek comprises over 11,000 AMD EPYC 7543 compute cores and 88 NVIDIA A100 graphics processing units (GPUs) connected by NDR Infiniband to an upgraded data storage facility.

Astronomers and researchers in general, whether they work in health, land-care, space technology or engineering, to name a few, naturally feel a close connection to the universe and what can be learnt from the world around us. We are aware that life is about more than just the present, given that we often work with data that reaches us after travelling for millions of years across cosmic time. Through the partnership between CAS, the Moondani Toombadool Centre and the Indigenous community we are excited to learn from a culture that has the deepest history of connecting with the land and skies, looking forward to developing a partnership of shared knowledge and discovery.

Standard Compute Nodes (150)
Dell PowerEdge R6525 Server
2 x AMD EPYC 7543 2.8 GHz 32-core processors
256 GB DDR5 RAM
2 TB NVMe local SSD
High Memory Nodes (10)
Dell PowerEdge R6525 Server
2 x AMD EPYC 7543 2.8 GHz 32-core processors
1024 GB DDR5 RAM
2 TB NVMe local SSD
Data Accelerator Nodes (22)
Dell PowerEdge XE8545 Server
2 x AMD EPYC 7543 2.8 GHz 32-core processors
4 x NVIDIA A100 80 GB GPUs
512 or 1024 GB DDR5 RAM
2 TB NVMe local SSD
Fabric Interconnect
200 Gigabit per second HDR Infiniband with NDR switching
High performance Omni-Path architecture
Lustre ZFS Target File System
13 Petabytes
Improved data integrity with transaction-based, copy-on-write operations, snapshots, and persistent data and metadata checksums on every write.
Slurm Workload Manager
Highly scalable cluster management and job scheduling
Open source