ARC offers classroom use of high-performance computing (HPC) cluster resources on the Great Lakes High-Performance Computing Cluster.
Support is $60.91, per student, per semester. Contact ARC for multi-semester courses to receive the funding up front. The $60.91 account is based on the class roster provided by the faculty, and not the number of Great Lakes accounts created. The ARC Help Desk will help you determine your needs.
These examples may help illustrate the allocation model:
- Example 1: If a course has 30 students for the semester, the course will be allocated $1,827.30.
- Example 2: If a course is offered more than once a year with different rosters, a new account will be created for each term, but not each section. E.g., If there are 10 students in a fall semester course, and 10 different students in the winter course, the course each term will be allocated $609.10.
- Example 3: If there is a class with five students, and the course runs in both the fall and winter with the same roster, the course will be allocated $609.10 and last both semesters.
Funding will be given as a lump sum to the class account to allocate by individual or team, at the instructor’s discretion and defaults with a per student limit to avoid single-student run away spending.
To ensure resources are available to classes for short-running jobs when the cluster is under heavy utilization, we have reservations set on the standard and gpu partitions that set aside nodes for use. The reservations do not need to be requested by the jobs:
- standard partition: jobs 2 hours and under
- gpu partition: jobs 1 hour and under
To encourage fairness for common use cases for coursework, user limits are set to encourage the turnover of resources to other course and cluster users. A per-user spending limit is also set to avoid the runaway use of the course budget by a single user. This differs from normal cluster accounts where limits are only at the entire slurm account level and not at the user by default. Jobs that request more than their limits will wait with an appropriate reason message in squeue output. Queued jobs may start as other course or user jobs finish or may never run if they exceed the current limits.
Instructors can request default limits be changed at any time as well as single-user limit changes. Most limits can be seen in the ARC Portal. Total course limits can also be modified from defaults as requested.
|Limit Type||Default Per User||Default Per Course||squeue Reason Job is not Running|
|Spending limit||$60.91||$60.91 * # of students||
|GPUs in use||1||5||
|Running Jobs||2 running (5000 queued)||N/A|
|Maximum Duration/Walltime||8 hours||N/A||
|RAM Limit||None||7GBytes * CPU Limit||
Limits should be set at the user level. Course instructors can determine student consumption, i.e., it does not need to be equal across students and can be used to support individual, group, or major project work. There can be restrictions on runtime, GPU, cores, and memory to reduce spending by students. The fewer resources a user can request the less a job will cost. If class computation requirements exceed its limits, another account can be created with a shortcode to supplement the “no cost” academic account. If you have any questions, please email email@example.com. Learn about the examples of the limits that can be set.
Exclusions: Courses that are not “classic courses” with instruction, syllabi, assignments, and grading are excluded currently. This course, for example, would not be eligible: graduate thesis courses such as BIOSTAT995 Dissertation Research For Doctorate In Philosophy. In addition courses that reflect faculty lead research will be excluded.
To request a course account, please contact firstname.lastname@example.org with the following information:
- Students to be put on the account
- List of individuals to administer the account
- Any limits to be placed on the either the users or the account as a whole
- The unit abbreviation and course and section numbers for the course you are leading (i.e., eecs498 section 400)
Ask a question
Email email@example.com with any questions about the applicability of HPC resources for your course.