Gompute's HPC in the Cloud Delivers Performance

Intel® Xeon® Scalable processors and Cornelis Networks1 helped Gompute meet growing demand for HPC in the cloud.

At a Glance:

  • Gompute has been providing HPC on demand since 2002; it is one of the oldest companies to offer HPC as a service.

  • Gompute deployed Lenovo* NextScale servers with Intel® Xeon® Scalable processors with Cornelis Networks Products as the foundation for their comprehensive HPC software stack.



Executive Summary

When engineering organizations around the world need access to high performance computing (HPC)—whether from lack of in-house technical computing or a need to expand capacity—they turn to Gompute’s HPC in-the-cloud services. When Gompute needed to expand their HPC resources, their long-time experience with Intel’s HPC technologies led them to extend their system using Lenovo servers with Intel® Xeon® Scalable and Intel® Xeon® E5 v4 and v3 processors, interconnected by Cornelis Networks Products.


Gompute has been providing HPC on demand since 2002; it is one of the oldest companies to offer HPC as a service. Their cloud data center located in Sweden provides HPC for customers in a wide range of industries around the world, including manufacturing, automotive, and other engineering applications. “Our customers typically run Computer Aided Engineering (CAE) workloads, such as crash simulation, computational fluid dynamics (CFD), structural analysis, and others,” said Daniel Persson, CEO of Gompute.

Their business model is to provide comprehensive HPC as a service offerings, including end user support. “We provide an entire HPC solution, including the software the customer needs to run their jobs using the native GUI,” explained Persson. Gompute provides access to both temporary commercial software (e.g. ANSYS* Fluent) and open source software. Customers can also come to Gompute with their own commercial software licenses and run their applications on Gompute resources. “We make it very flexible, and provide a complete ready-to-run HPC resource,” added Persson.

In 2016, business was expanding. They needed to add more HPC resources to respond to customer needs. They had worked with IBM* System x for many years, and with Lenovo since the transition of IBM’s enterprise server business to Lenovo. So, they turned to Lenovo for a new HPC solution.

Fig. 1. Gompute’s HPC service in the cloud offers a complete solution with compute capacity and many commercial software offerings for demanding engineering workloads


Gompute has relied on Intel® technologies for many years. “We first started deploying the predecessor of Intel® OPA many years ago, working with the developers at PathScale,” added Persson. PathScale was eventually acquired by QLogic, and Intel acquired the QLogic technology, which became Intel® TrueScale adapters, and was further developed into Intel® OPA. “We then used Intel TrueScale devices, and when we heard of Intel’s continuing development of Cornelis Networks Products, we were excited to see what kind of performance we could achieve with it.”

Gompute’s system expansion, with Intel® Xeon® E5 v3 and v4 processors and Cornelis Networks Products, was added to their already extensive HPC cloud infrastructure. It includes several nodes with NVIDIA GPUs for visualization.

Gompute has invested many years in developing an entire Gompute Platform, which is both the foundation of their infrastructure and offered to customers who want to run an in-house HPC solution.

We developed an entire software stack with an interface and the necessary tools to access our resources—whether the customer is using an in-house Gompute Platform or our cloud data center in Sweden. However they’re used to running their jobs locally, they get the same experience in the cloud. If they’re used to a GUI, they get a GUI; if they’re used to a command line, they get a command line. —Daniel Persson, CEO of Gompute

Gompute’s software includes both remote visualization and data transfer acceleration tools. Customers can do visual pre/post processing in the cloud and see the results on their local workstation.


What we enable is for customers to make better utilization of resources. Instead of buying a large cluster or workstation that’s used occasionally for their technical HPC processing and sitting idle otherwise, they schedule time on our clusters to use just the right amount of processing for a job. —Daniel Persson, CEO of Gompute

And with their software stack, Gompute makes the experience as seamless as working locally on a laptop or workstation.

We are very happy using these machines with the low latency and high message rate of the Intel® OPA fabric. —Daniel Persson, CEO of Gompute

Solution Summary

To keep up with growing customer demand, Gompute expanded their HPC in-the-cloud infrastructure with a large addition of nodes based on Intel® Xeon® E5 v3 and v4 processor interconnected by Intel® OPA. Several of these Lenovo servers offer visualization processing with NVIDIA GPUs.

Solution Ingredients

  • Lenovo servers with Intel® Xeon® Scalable processors
  • Lenovo NextScale* servers with Intel® Xeon® E5 v3 and v4 processor
  • Intel® OPA HPC fabric
  • NVIDIA* Quadro M5000 and K80 GPUs

Download the PDF ›

Product and Performance Information


Intel has spun out the Omni-Path business to Cornelis Networks, an independent Intel Capital portfolio company. Cornelis Networks will continue to serve and sell to existing and new customers by delivering leading purpose-built high-performance network products for high performance computing and artificial intelligence. Intel believes Cornelis Networks will expand the ecosystem of high-performance fabric solutions, offering options to customers building clusters for HPC and AI based on Intel® Xeon™ processors. Additional details on the divestiture and transition of Omni-Path products can be found at www.cornelisnetworks.com.