Buy Compute Nodes

The University of Iowa HPC systems allow users to purchase dedicated compute node capacity for their groups. At present the following options are available for purchase. For questions or to purchase nodes please contact hpc-sysadmins@iowa.uiowa.edu. For information on data storage options please visit https://hpc.uiowa.edu/resources/storage.

Neon Cluster - No further purchases available. Neon nodes are out of warranty and have an end of life date of January 2021. Neon nodes will not be repaired out of warranty.

Argon Phase 1

Argon Phase 1 Compute Nodes - Available Now - (56 compute slots, 2 x Xeon E5-2680v4 (28 Cores at 2.4GHz), 1TB SSD, 1Gbps Ethernet, 100Gbps Omnipath, Warranty through May 2022)

  • Standard Compute Node (128GB, No Accelerator Support) - $5399 - Qty 1 Available
  • Mid-Memory Compute Node (256GB, No Accelerator Support) - $5930 - Qty 5 Available
  • Standard Compute Node (128GB, Accelerator Support) - $5781 - None Available
  • Mid-Memory Compute Node (256GB, Accelerator Support) - $6219 - Qty 1 Available
  • High Memory Compute Node (512GB, Accelerator Support) - $6804 - Qty 1 Available

Argon Phase 1 Accelerator Cards - Must be ordered, typical lead time is 2-4 weeks

  • Nvidia Titan V - $3179 (1 per accelerator capable node)
  • Nvidia K80m – $4529 (Up to 2 per accelerator capable node)
  • Nvidia P100 - $5890 (Up to 2 per accelerator capable node)
  • Nvidia P40 - $7079 (Up to 2 per accelerator capable node)
  • *NOT Supported* Nvidia 2080ti, GV100, 1080 cards are *not* supported in Lenovo/Argon Phase 1 systems. 

Argon Phase 2

Consumer GPU friendly compute nodes are now available for purchase in the Argon system. While we are currently taking orders there is some uncertainty around product supply. Our current target for initial deployment is late July/early August 2018. Memory levels have been adjusted due to the new "Skylake" CPU architecture in these systems. We are including GPU (Silver) and CPU (Gold) intensive versions of these systems as options. Expected warranty for these nodes is through September 2023.  OmniPath is *not* included in this pricing.

Silver Compute Nodes - Designed for GPU intensive computing with modest CPU use.

  • Low Memory: 20 Cores (Xeon Silver 4114@2.2ghz DDR4-2400 96GB, 1TB SSD (supports 4GPUs, none included) - $6600
  • Standard Memory: 20 cores (Xeon Silver 4114@2.2ghz DDR-2400 192GB, 1TB SSD (support 4GPUs, none included) - $7850
  • Mid-Memory: 20 cores (Xeon Silver 4114@2.2ghz DDR-2400 384GB, 1TB SSD (supports 4GPUs, none included)- $10,700

Gold Compute Nodes - Designed for GPU intensive computing with significant CPU use.

  • Standard Memory: 32 cores (Xeon Gold 6130@2.1ghz DDR-2666 192GB, 1TB SSD (supports 4GPUs, none included) $12,000
  • Mid-Memory: 32 cores (Xeon Gold 6130@2.1ghz DDR-2666 384GB, 1TB SSD (supports 4GPUs, none included) $14,000

GPUs - As indicated above these configurations do not include GPU cards. We are suggesting a "Bring Your Own GPU" model with the following cards supported:

  • Nvidia Titan V - $3179/card
  • Nvidia Titan 1080 Ti - ~$1000/card
    • Supported 1080 cards not directly manufactured by Nvidia that are supported:
      • PNY: VCGGTX1080T11PB-CG
      • MSI GTX 1080 Ti AERO 11G OC
      • If you have a 1080 Ti family card you're looking at, ask us and we can request information from our vendor on supportability in the server chassis.  The general guidance at this time is that cards with a single fan on them will work and cards with 3 fans will not.  A single fan card is referred to as "reference design" by Nvidia.
  • *NOT SUPPORTED* Nvidia Founders Edition/Nvidia direct 2080ti. 

OmniPath - The above configuration does not include OmniPath high speed 100Gbps networking.  It can, however be added for $800 per node.

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.

  1. The final two years of the compute node life are outside warranty and are on a best effort basis.
  2. If software/OS can no longer support compute equipment prior to 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 7 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.
  3. In the event that compute nodes fail outside of warranty they will not be repaired. The HPC team will attempt to keep investor queues at the purchased capacity, to the extent possible, based on the following process and guidelines.
    1. Investor compute nodes that fail out of warranty will be replaced with compute nodes from the UI queue within the same generation of hardware.
    2. When possible compute nodes will be replaced with same or higher specification hardware. This will not be possible in all cases. In cases where this is not possible the investor will be contacted by the HPC team with available options.
    3. Transfer of compute nodes from the UI queue to investor queues will occur in the order in which failures occur.
    4. UI queue compute node availability is finite and is unlikely to be able to sustain all investor queues at full capacity for a 7 year life. As such investors should not assume that their queue will remain at full capacity for the duration of the two year life outside of warranty.