High Performance Computer
The NIASRA HPC is a collection of powerful computers available for use by NIASRA staff and students for research purposes only.
You can view the current status of the cluster on the NIASRA HPC status page.
Computer Specifications
Head node
Name: hpc.niasra.uow.edu.au
Operating System: CentOS Linux
Storage: 20TB
This computer provides shared storage and access to the other computers.
Compute nodes
Name: niasra-node1
Operating System: Fedora Linux
CPU: 4x AMD Opteron 6376 (64 Threads)
RAM: 256GB
SSD Swap: 256GB
Applications: R (OpenBLAS), Python, MATLAB
Name: niasra-node2
Operating System: Fedora Linux
CPU: 4x AMD Opteron 6376 (64 Threads)
RAM: 256GB
SSD Swap: 256GB
Applications: R (OpenBLAS), Python, MATLAB
Name: niasra-node3
Operating System: Fedora Linux
CPU: 2x Intel Xeon E5-2683 v4 (64 Threads)
RAM: 256GB
SSD Swap: 256GB
Applications: R (OpenBLAS), Python, MATLAB
GPU: Nvidia Kepler Titan 6GB
Name: niasra-node4
Operating System: Fedora Linux
CPU: 2x Intel Xeon Silver 4216 (64 Threads)
RAM: 384GB
SSD Swap: 64GB
Applications: R (OpenBLAS), Python, MATLAB
GPU: Nvidia Telsa v100 32GB
Name: niasra-node5
Operating System: Fedora Linux
CPU: 2x Intel Xeon Gold 6348 (112 Threads)
RAM: 512GB
SSD Swap: 8GB
Applications: R (OpenBLAS), Python, MATLAB
GPU: Nvidia A100 80GB
Storage: 40TB
This computer provides shared storage to the other computers.
Name: niasra-node6
Operating System: Fedora Linux
CPU: 2x Intel Xeon Gold 6348 (112 Threads)
RAM: 512GB
SSD Swap: 8GB
Applications: R (OpenBLAS), Python, MATLAB
GPU: Nvidia A100 80GB
Storage: 40TB
How to access the HPC
RStudio Server
RStudio in your web browser.
- RStudio Server - niasra-node1
- RStudio Server - niasra-node2
- RStudio Server - niasra-node3
- RStudio Server - niasra-node4
- RStudio Server - niasra-node5
- RStudio Server - niasra-node6
Secure Shell (SSH)
Use a SSH client (such as PuTTY for Windows) to log in to hpc.niasra.uow.edu.au and then use SSH from there to connect to a computing node, for example:
ssh niasra-node1
You can then run the interactive R interpreter with this command:
R
Or run a script file (for example, myscript.r) in non-interactive mode with this command:
Rscript myscript.r
The Linux utilities screen or tmux can be used to keep your programs running after you close the SSH connection.
SSH key fingerprints:
RSA 2048 SHA256:cp8CcZejIDw+yxiUBlbLWZjNF6tKwYoKBw/VuwRI9nE
ECDSA 256 SHA256:YEJMF7iarqjOD/28HcMcWZlCoUhEoDyKqaB3eYyQIHw
ED25519 256 SHA256:VW3O/PV4SBQC4xAvWoon8u1ZeTxmtA/V8/KJkoERolU
Secure File Transfer (SFTP) or Secure Copy (SCP)
Link: Instructions for SFTP.
Remember to always keep a backup of important code and data on your own computer!
GPU
Node3 contains a Nvidia Titan 6GB GPU, Node4 contains a Nvidia Tesla V100 32GB GPU, Node5 and 6 contains a Nvidia A100 80GB GPU.
- MATLAB: Using GPUs in MATLAB
- Python: TensorFlow
- R: Keras
You might be able to use the GPU without any changes to your R code by automatically intercepting certain BLAS routines with NVBLAS. Log in to niasra-node3 or niasra-node4 by SSH and run R with this command:
LD_PRELOAD=/usr/local/cuda/lib64/libnvblas.so NVBLAS_CONFIG_FILE=/etc/nvblas.conf R
or
LD_PRELOAD=/usr/local/cuda/lib64/libnvblas.so NVBLAS_CONFIG_FILE=/etc/nvblas.conf Rscript myscript.r
Changing your password
Please do not use the passwd command! You can change your password by visiting this page: hpc.niasra.uow.edu.au/password.py