Massive GPU Clusters from Nor-Tech

Astounding Power in a People-Friendly Design

Our Massive GPU Clusters Feature:

  • NVIDIA GPU Compute: The foundation is our high-performance servers integrated with NVIDIA Tensor Cores for Artificial Intelligence applications and CUDA cores for HPC applications.
  • Fast, All-Flash Storage: Our expertly engineered network storage servers provide rapid access to data. The administration dashboard simplifies cluster management.
  • HPC Networking: Our world-renowned engineering team will custom-design the best Ethernet or InfiniBand topology. Network designs leverage lightning-fast GPU Direct RDMA.
  • Nor-Tech’s Legendary Support: These massive GPU clusters come racked, stacked, labeled, and cabled; backed by support from the same engineers that built the system.
  • Easy to Deploy: They can be shipped fully assembled, ready to roll onto your data center floor and put into service the same day.
  • Thorough Validation: Compute, storage, and networking architectures are validated by Nor-Tech’s HPC engineers
  • Up to 40 H100 GPUs per rack including storage and network interconnect

 

A Single Point of Accountability and Support

Working with one vendor for the entire GPU solution eliminates the hassle of contacting multiple vendors and never really getting the answer you were looking for. Your massive GPU cluster’s compute, network, and storage are all provided by Nor-Tech. This has the added benefit of streamlining the procurement process.

A Custom Cluster of Nor-Tech GPU servers

Nor-Tech has a strong industry reputation for HPC customization expertise. Whether you’re looking to set up a traditional HPC cluster or a distributed cluster, we’ll engineer the best technology for your precise present and future needs.

All-Flash Network Storage

The Massive GPU reference design combines a high-speed storage fabric with local NVMe flash caches to dramatically speed up data transfer rates and eliminate bottlenecks. Nor-Tech has close working relationships with all storage appliance OEMs.

Incredibly Fast HPC Networking

Nor-Tech’s massive GPU clusters feature 100% port-to-port bandwidth spine & leaf topology. GPU compute nodes communicate via the latest InfiniBand fabric; with each node having multiple InfiniBand HCAs.

World Class Support

Nor-Tech is famous for industry-leading support. Every massive GPU cluster ships with the following:

  • An ironclad 3-year warranty.
  • No Wait Time Support: WE DO NOT PUT CLIENTS ON HOLD.
  • U.S. Based Support: all inquiries are routed to our Minnesota headquarters.
  • Flexibility: we provide both remote and onsite help.
  • Expertise: our certified support staff averages 10+ years of experience.
  • Familiarity: the same engineers that built your massive GPU cluster are available for support.
  • Patience: our engineers take the time to thoroughly listen before diagnosing the issue. They also explain solutions in terms that those without a deep technology background can understand.
  • Detailed, Customized Documentation: our HPC cluster clients receive manuals and Quick Start Guides that are customized down to the graphics.
  • Whether you have a hardware, software, or Linux system administration question, we’ll be there to help whenever you need us.

A Fraction of the Cost of Cloud

If you’re a heavy GPU cloud compute consumer, you can expect a TCO of anywhere from one half to one fifth of what you’re paying now. Download a NASA study summary on

 

The Advantages of On-Prem v. Cloud

 

Nor-Tech’s Massive GPU Cluster Experience

  • Ongoing builds for Nobel Physics Prize-Winning Project
  • Major storage product vendor
  • Multiple major research institutions
  • ISV for CAE
  • And many more

 

Engineered for You

Take advantage of Nor-Tech’s massive GPU cluster engineering expertise and experience.


  • Request Info







      resellerend user

    Start typing and press Enter to search