For Faculty Members
Contributing to the Cluster
TTIC faculty members are invited to contribute hardware to the Beehive cluster. This section explains the benefits and options available when contributing resources.
Pooled Resource Model
The Beehive cluster operates on a resource-sharing principle that provides users with access to more computing power than they would have individually. By contributing hardware, you help create a more robust computing environment while gaining priority access to the shared GPU pool.
Contribution Options
When contributing hardware to Beehive, you'll choose between two priority access models:
In both cases your group receives a dedicated Slurm partition named after it (for example greg-gpu or willett-gpu) that only members of your group's Unix group can submit to with -p <partition>. The current group partitions are listed on the Jobs page.
Option 1: Guaranteed Access to Your Hardware
With this option:
- Your research group is guaranteed access to your contributed machines within an 8-hour window
- Once you request access, any non-group jobs will complete (up to their 8-hour time limit)
- You maintain control over your hardware; depending on your arrangement, your group's partition on it has either an 8-hour or an unlimited time limit
- Your hardware is available to other users only when you're not using it
- You can reserve your machines (minimum one-week periods) if you need exclusive access
Option 2: Higher Priority Across All Cluster Resources
With this option:
- Your group receives a higher partition priority on the shared GPU pool, via a dedicated group partition covering the same nodes as
gpu - That priority is one weight in the fair-share formula, not an absolute override: it tilts scheduling toward your group, but a job from a user with much lower recent usage can still run first
- You can still reserve your machines for up to four weeks per year (minimum one-week periods)
Private (unshared) hardware
Hardware can also be hosted and managed in the cluster without joining the shared pool: it gets its own <name>-own-gpu partition with no time limit, restricted to the owning group, and is never scheduled for other users.
For more information about contributing hardware to Beehive, please contact the TTIC system administrators at admin@ttic.edu.