Skip to Main Content

The University of Tennessee

Newton header banner

Frequently Used Tools:

Home » Documentation » Memorandum of Understanding

Newton Program – Memorandum Of Understanding

The Newton HPC program is an Office of Research shared computing facility to support computationally intensive research. OIT provides hosting and technical management for the program. Computing hardware is funded through a buy-in agreement between groups who wish to participate in the program.

To join the program, a research group provides funds to OIT to purchase computer processors that will best fulfill the researcher's needs while remaining compatible with existing infrastructure. Once the hardware is installed in the facility, the group’s priority to use resources will be set to reflect the added investment in the program. Contributing users, members, are free to use all resources in the facility at a priority level that is proportional to their monetary investment relative to the investments of the other members. The policies of the facility are designed to ensure that each member is able to make use of their share of the resources of the facility. Members are free to assign their shares to specific sub-groups, projects, and individual users.

Each group in the program holds a share of the full computational facility, defined as the ratio of their active monetary contributions to the total active contributions to the facility. Shares are active for a period of three years starting from the group's initial hardware installation. This three-year period is representative of the typical support period of computing hardware.

The facility policies are designed to ensure equitable use of resources and to ensure a reasonable turnaround time for resource requests. All resource requests are submitted though a central resource scheduler that implements the share-based policies according to the Fairshare priority algorithm. This algorithm compares a member's resource usage over a given time period to the share of resources to which a member is entitled. The priority is then set relatively higher or lower based on this comparison with the objective of having the used resources equal to the allocated share of resources. The scheduler favors resource requests in order from highest to lowest priority. In addition, throughput limits (based on group priority share) are placed at the user and group levels in order to guarantee system-level turnaround time goals.

Members will also have a limited opportunity to make advanced reservations of specific resources in order to meet deadlines, throughput targets, or turnaround goals.