More data, more computing resources, more results
We offer core services to all users for data storage and for accessing distributed computing resources.
Our users are allocated a system account (currently about 350 active accounts) to access data storage space and distributed computing resources.
Our IT management team supports over 140 individual computers (workstations and rack-mounted servers) on our local network.
We provide a grid-computing cluster, composed of more than 550 CPU cores in varying memory configurations (from 48 up to 512 GB of memory per server) yielding more than 1,100 virtual cores (logical processors). In terms of GPU computing, we currently provide a dual-GPU adapter with 24 GB of memory yielding 4,992 NVIDIA CUDA cores, 3 Tesla C2070 adapters with 6 GB of memory each yielding 1,344 NVIDIA CUDA cores. Adding to that, the GPU computing ability of individual workstations provides more than 20,000 CUDA cores.
Overall, around 1 million computing hours are scheduled per year on average on the different grid compute queues amongst the different BIC labs.
We have also greatly upgraded and expanded our storage capacity, now exceeding 750 TB and building up to 1.5 PB of networked space (in situ and in the cloud).
A new local storage space warehouse is hosted in the BIC server room. It consists in a Dell/EMC/Isilon cluster composed at the moment of 5 Isilon storage nodes for a effective usable capacity of 500TB. It can be seamlessly scaled up in size and performance by simply adding more storage nodes up to 15PB in storage capacity.
Access to the BIC Isilon Cluster storage space by the compute cluster nodes is through a dedicated 10Gigabit/s Fibre optic network. As for the workstations spread over all the different BIC labs, access to the cluster storage space is achieved through a 1Gigabit network provided by McGill IT services (100Mb to the workstation). McGill IT service now offers 1GbE to the desktop if so desired.
A new 40-core server for Virtual Machines provisioning is also now available, in order to provide researchers with compute and analytic tools for specific tasks and workflows that cannot be easily met with the existing core infrastructure.
In addition, we configure and support high-end BIC Neuroimaging workstations for everyone. Service includes expert consulting for workstation or data server specification, purchase, installation and maintenance (Linux). All workstation installations comprise a series of specialized software applications for neuroimaging research (MINC, SPM, FSL, AFNI, Brainstorm, etc.). Installation of additional software packages can be accommodated upon request.
FAQ, troubleshooting and helpful information can be found here.