Course access to the DCC
We are currently (2025 Spring, Summer, and Fall) offering dedicated resources on the DCC for course support. We are providing access via SLURM SSH or OnDemand (web) for Rstudio, Jupyter, of a linux desktop. Resources dedicated for courses include 50 Infiniband connected servers with 86 cores and 479GB RAM each. 10 of these servers also have have 2 P100 (16GB RAM) GPUs (20 GPUs total).
For most courses at Duke, computational support is provided through OIT Virtual Labs (Virtual machines and containers for courses). DCC course access may be useful for:
- 500 level or higher course with heavy computational needs
- ML / AI courses
- By request of the instructor
Requesting access for course
Instructors can email OITResearchSupport@duke.edu.
Please provide course ID, brief summary of need, TAs and any special requirements.
DCC conventions for course support
- Students will be added to the DCC coursess25 group access through the class roster automatically.
- Student jobs/work should be submitted through the
courses
andcourses-gpu
partitions - Instructors should place course materials in
/hpc/group/coursess25/<courseid>
(will need to use 'mkdir' to create first) - Students should store their files on the dcc at
/hpc/group/coursess25/<courseid>/<netid>
(will need to use 'mkdir' to create first, after instructors have created the course directory)