Purchasing Compute Nodes

The West Virginia University HPC systems allow users to purchase dedicated compute node capacity for their groups. For questions and pricing for purchasing nodes, please contact helpdesk@hpc.wvu.edu .

Thorny Flat

Thorny Flat is close to its end of life, and purchasing nodes on this cluster is no longer possible.

Dolly Sods

GPU Capable Compute Nodes (Up to 8 GPUs per Node) will go into the GPU cluster Dolly Sods. Specifications are changing, the compute node could be similar in memory configurations to those in Thorny Flat.

The cost per GPU card is listed below. Consumer cards and GPUs intended for workstations are not an option for use in data center systems, and server-grade GPUs from the Nvidia Ampere Series listed below are available in limited configurations due to the supply chain. Contact us if you’re interested in pricing out a GPU node.

  • NVIDIA® A10 GPU Computing Accelerator - 24GB GDDR6 - PCIe 4.0 x16 - Passive Cooler $4000 approx

  • NVIDIA® A30 GPU Computing Accelerator - 24GB HBM2 - PCIe 4.0 x16 - Passive Cooler $5000 approx

  • NVIDIA® A40 GPU Computing Accelerator - 48GB GDDR6 - PCIe 4.0 x16 - Passive Cooling $5300 approx

  • NVIDIA® A100 GPU Computing Accelerator - 80GB HBM2 - PCIe 4.0 x16 - Passive Cooling $14000 approx

How long will my compute node be available?

All HPC compute nodes are purchased with a five-year warranty. Compute nodes will be allowed to run for up to 7 years within the following parameters:

  • The final two years of the compute node life are outside warranty and are on a best-effort basis.

  • If software/OS can no longer support compute equipment before the end of its 7-year life, the HPC team may, in consultation with the HPC Policy Committee, determine that the life of compute equipment is shorter than seven years. Should this occur, the HPC team will strive to provide at least six months of notice to the HPC community before equipment is decommissioned.

  • If compute nodes fail outside of warranty, they will not be repaired. The HPC team will attempt to keep investor partitions at the purchased capacity, to the extent possible, based on the following process and guidelines.

  • Investor compute nodes that fail out of warranty will be replaced with compute nodes from the UI partition within the same generation of hardware.

  • Compute nodes will be replaced with the same or higher-specified hardware when possible. However, this will not always be possible. In cases where this is not possible, the HPC team will contact the investor with available options.

  • Compute nodes will be transferred from the UI partition to investor partitions in the order in which failures occur.

  • UI partition computes node availability is finite and is unlikely to sustain all investor partitions at full capacity for a 7-year life. As such, investors should not assume that their partition will remain at full capacity for the two-year life outside of warranty.

Renting Compute Nodes

HPC node rentals are not available at the moment.