Software platform emulates AI and HPC datacenter racks for development
Cnuas: A Software-Defined AI/HPC Rack-scale Emulation Platform and Hyperscale Data Center Facility Twin
Distributed, Parallel, and Cluster ComputingHardware ArchitectureNetworking and Internet Architecture
Summary
Modern AI and high-performance computing (HPC) systems use lots of specialized hardware packed tightly in racks, which are expensive and rare to access for software development. The authors created Cnuas, a software platform that mimics these racks so developers can experiment and learn without needing the real hardware. It offers tools to visualize how all the parts connect and work together, helping users understand their software’s needs across devices. While it doesn’t match the speed of real systems, it supports exploring device interfaces and rack management in a realistic environment.
What this means in practice
- •For data center engineers: Assess software and device interactions in a virtual rack-scale environment matching open hardware standards before deploying on physical AI/HPC hardware.
- •For cloud infrastructure developers: Develop and test rack management and network software using the platform's web interface and management firmware without real hardware.
Authors
Weqaar Janjua, Eoin OConnell, Mihai Penica
Abstract
Modern AI and HPC systems integrate accelerators, high-speed networks, and management controllers at rack scale. Developing software for this infrastructure typically requires access to scarce, costly hardware, while software abstractions can obscure how workloads depend on resources across servers and accelerators. This paper presents Cnuas, an open-source, experimental rack-scale emulation platform whose baseline architecture follows the Open Compute Project (OCP) Open Rack v3 specifications. Through functional emulation, it supports experimentation, learning and software development within academic and industrial research and development, rather than matching the throughput or latency of physical hardware. Its web-based user interface visualizes racks, devices and their interconnections to help developers build a system-level mental model of the infrastructure supporting their workloads. At its core, CnuasNIC and CnuasSwitch implement a guest-visible remote direct memory access (RDMA) adapter and a host-resident hybrid software switch supporting both RoCEv2 and native InfiniBand. The platform also provides a dedicated AI/ML accelerator (GPU) peer fabric and OpenBMC-based rack management with executable power supply and battery backup firmware over RS-485. These components support the study of device, driver, and firmware interfaces on commodity hosts. The accelerator software stack remains an early research prototype, and facility modeling with OpenUSD is an exploratory extension. The paper presents the architecture, interfaces, and bounded prototype results as a basis for community collaboration across the core platform and its extensions.