Node Purchase Estimates
Overview
Below are the price estimates to be used for planning possible node investments within Beartooth. This type of purchase is only available to UWyo faculty PIs. See 2.2 Cost of resources and services | Compute for explanation of our investment model.
NOTE:
Price estimates are valid for UW Internal users, only.
These are estimates and ARCC will go out to quote/bid for new purchases. Price is subject to market conditions and is usually within a +/- $2,000 margin per node relative to the numbers listed here.
UWyo faculty PIs will pay actual cost for equipment, ARCC does not include any additional markup or fees.
Date of Most Recent Estimate
2/21/24
Included with each estimate
2 Socket CPU
AMD EPYC 9454 48C 205W 2.75GHz Processor, equating to 96 CPU cores per individual node
All estimates include 5 years warranty/support from OEM. Nodes run beyond this are ‘best effort.’
Operating system
Red Hat Enterprise Linux
High speed networking card
Mellanox Connect-X latest gen
Power & network cabling
Rack space
network ports
Large purchases may require purchasing additions to our core network
IPMI or equivalent administration software
All other basic components required to make the system work with ARCC clusters
GPU Node Estimates
768GB Memory 8x Nvidia A30 24GB GPU Node | Price is subject to market conditions, ~$48,000 6/1/2024 |
768GB Memory 8x Nvidia L40 48GB GPU Node | Price is subject to market conditions, ~$65,000 6/1/2024 |
1.5TB Memory 8x Nvidia H100 80GB GPU Node | Price is subject to market conditions ~$250,000 6/1/2024 |
NOTE: These are estimates and ARCC will go out to quote/bid for new purchases. Price is subject to market conditions.
Non-GPU node estimates
1TB Memory Compute Node | Price is subject to market conditions, ~$20,000 6/1/24 |
NOTE: These are estimates and ARCC will go out to quote/bid for new purchases. Price is subject to market conditions. Please contact arcc-help@uwyo.edu to obtain up-to-date prices.
Specialty Nodes
Nodes with a non-standard config must be assessed by ARCC staff for viability within our clusters. Reach out to us at arcc-help@uwyo.edu for questions.