Computational Neuroscience Cluster
The computational neuroscience group manages a compute cluster with 500TB storage capacity and high-performance compute nodes, many of which are GPU enabled. The cluster is supported by two CFI awards (Diedrichsen / Khan) and BrainsCAN’s computational and human cores.
When will the cluster be available?
The computational neuroscience cluster is currently being installed and should be operational before the end of 2019.
How will the cluster work?
Each research group will have one more multiple disk shares (i.e. storage volumes), which can be accessed by mounting them to local machines, or through other file transfer protocols. Users can also login to the system to access a virtual Linux desktop computer, which can automatically see the storage volumes. Standard users can access 4-core machines, high-performance users (extra charge, see below) can access virtual machines with up to 16 cores, 32GB of RAM, and GPUs.
How do I get access to the cluster?
The PI of the group will need to submit an Application form for the group and provide a speed code for the access charge.
What does access to the cluster cost?
Fees are charged on an annual basis. The annual fees are currently set to:
- $50/TB for storage
- Standard user accounts are free
- $1000 for the first high-performance user in the group
- $500 for each following high-performance user in the group
These prices are based on a 30% cost recovery model over 5 years – the fees will be used for upgrades and extensions of the compute system. We reserve the right to adjust the fee structure on a yearly basis depending on usage.