ROAR User Guide   »   Roar Collab System Specs
Feedback [ + ]

Roar Collab System Specs

Roar Collab (RC) is the flagship computing cluster for Penn State researchers. Designed with collaboration in mind, the RC environment allows for more frequent software updates and hardware upgrades to keep pace with researchers’ changing needs. RC utilizes the Red Hat Enterprise Linux (RHEL) 8 operating system to provide users with access to compute resources, file storage, and software. RC is a heterogeneous computing cluster comprised of different types of compute nodes, each of which can be categorized as a Basic, Standard, High-Memory, GPU, or Interactive node.

Node Type
24 cores, 126 GB
64 cores, 255 GB
Connected via Ethernet
Configured to offer about 4 GB of memory per core
Best used for single-node tasks
24 cores, 258 GB
48 cores, 380 GB
48 cores, 512 GB
Connected both via Infiniband and Ethernet
Infiniband connections provide higher bandwidth inter-node communication
Configured to offer about 10 GB of memory per core
Good for single-node tasks and also multi-node tasks
48 cores, 1 TB
56 cores, 1 TB
Connected via Ethernet
Configured to offer about 25 GB of memory per core
Best for memory-intensive tasks
28 cores, 256 GB
28 cores, 512 GB
48 cores, 380 GB
Feature GPUs that can be accessed either individually or collectively
Both A100 and P100 GPUs are available
36 cores, 500 GB Feature GPUs that are specifically configured for GPU-accelerated graphics
Best for running graphical software that requires GPU-accelerated graphics