CIQ chips out Rocky Linux for AI

CIQ is the founding support and services partner of Rocky Linux.

The company has now come forward with the preview launch of the somewhat laboriously named Rocky Linux from CIQ for AI (RLC-AI).

The clue is in the name here, this is an operating system engineered and optimised for artificial intelligence workloads. 

Where general-purpose operating systems require more extensive manual setup, configuration, validation and tuning, CIQ says RLC-AI delivers kernel-level and user-space optimisations designed specifically for AI model enablement, tuning and inference.

It does all this while maintaining the stability and security expected from an enterprise-grade Linux distribution.

The RLC-AI kernel starts with the stable kernel.org long-term (LT) release and enhances it specifically for AI applications while maintaining compatibility. Key features include AI-optimised build configurations, support for “confidential computing” and support for Nvidia, Mellanox, AMD and other hardware acceleration components.

“Linux has always served as the critical surface area between software and hardware advancements,” said Gregory Kurtzer, CEO and founder of CIQ. “And with the rate of innovation in AI technologies, it is critical that the operating system helps to maintain compatibility. Not only this, but the operating system must also play a more active role in how we optimise these heavy workloads and how quickly we can take advantage of new advancements. RLC-AI enhances the Linux kernel and the user space to help organisations of all sizes adopt AI, improve utilisation of important compute resources, and maintain competitive advantage.”

RLC-AI includes both proprietary and out-of-tree modules alongside standard drivers, eliminating post-installation setup and ensuring users can get the most out of new hardware. 

For organisations deploying thousands of instances daily, this saves significant operational costs and time by removing per-instance driver installation overhead, preventing compatibility issues, and avoiding the need for custom images.