skip to main content

Reference Architecture: Multi-Node HPC Workload Scaling with Cornelis CN5000 Omni-Path on Lenovo ThinkSystem Servers

Reference Architecture

Home
Top
Published
7 Aug 2026
Form Number
LP2486
PDF size
22 pages, 779 KB

Abstract

High-performance computing (HPC) workloads such as computational fluid dynamics (CFD), computer-aided engineering (CAE), molecular dynamics, weather modeling, and materials science increasingly rely on scalable, low-latency interconnects to efficiently utilize growing CPU core counts across multiple compute nodes. As workloads scale, application performance becomes increasingly dependent on the ability of the network fabric to deliver high bandwidth, low latency, and predictable communication performance.

This Reference Architecture (RA) demonstrates how Lenovo ThinkSystem servers, Lenovo Neptune liquid cooling, and Cornelis CN5000 Omni-Path networking can be combined into a scalable HPC platform. The reference design features a Lenovo N1380 Neptune Direct Water Cooled Chassis populated with 8 compute trays totalling 16 nodes of Lenovo ThinkSystem SC750 V4 servers powered by Intel Xeon 6 Processors connected with high speed internode communication handled by Cornelis CN5000 Omni-Path SuperNICs and Switching.

Table of Contents

Introduction
Hardware Overview
Toolchain Modules and Performance Tuning
Application Performance Review
Conclusions
References

To view the document, click the Download PDF button.

Related product families

Product families related to this document are the following: