I am pleased to announce that the CPU and GPU components of the Balena HPC service is available for general use.
I apologise that this announcement has come rather suddenly, but this project has not gone as smoothly as hoped and has been severely delayed during the implementation and acceptance testing stages, which has made forward planning rather difficult. To date a significant amount of work has taken place to make the cluster fit for production use and it must be said that there is still more to do before this project is considered fully complete. However, we have now reached a stage where we are happy to open up the cluster and allow general users on to submit workloads.
As mentioned, this project is not fully completed which means that the full functionality of the cluster is not available for general use, but the cluster is in a state where it is offering the same, if not more functionality to that of Aquila, Belana's predecessor.
Lastly, once again I would like to thank you all for your patience over the last several months.
- HPC Team
Expected functionality of Balena HPC
It should be noted that the system is not fully accepted and that not all parts of the service are fully functional and generally available. However, the below list of capabilities/functions of the cluster are available.
- 168 standard nodes providing 2,688 cores
- 88 nodes with 4GB of memory per core (64GB per compute node)
- 80 nodes with 8GBs of memory per core (128GB per compute node)
- 2 High-memory nodes each with 512GB (32GBs per core)
- 6 nodes each with one NVIDIA K20x Kepler GPU card
- 4 nodes each with four NVIDIA K20x Kepler GPU cards
- 1 node with two AMD S10000 GPU card
- 2 Interactive test and development (ITD) nodes each with a NVIDIA K20x card and 128GBs memory
- Shared home area storage based on NFS (5GB quota per user area)
- 0.22PBs of non-archivable storage based on a BeeGFS/FhGFS parallel filesystem for high I/O
The below list of capabilities/functions are still being addressed and will be introduced at a later date.
- Intel Xeon Phi systems
- High-end visualisation service
- Remote storage mapping
- Web based metrics / accounting information
- Energy reporting
Respond