Red Hat & AMD extend AI & virtualisation for hybrid cloud

Red Hat and AMD have announced a strategic collaboration aimed at expanding customer options for artificial intelligence (AI) and virtualisation across hybrid cloud environments.

The partnership integrates Red Hat's open-source expertise with AMD's high-performance computing hardware. The companies state that the collaboration will allow organisations to deploy optimised, efficient, and production-ready environments for AI-enabled workloads, while also providing tools for efficiently modernising traditional virtual machines (VMs).

A core component of the collaboration involves full enablement of AMD Instinct GPUs on Red Hat OpenShift AI. This means that customers deploying AI across hybrid cloud environments can take advantage of AMD's processing power without requiring extensive resources.

AMD Instinct MI300X GPUs, when paired with Red Hat Enterprise Linux AI, have demonstrated success in AI inferencing for both small and large language models (SLMs and LLMs). Testing conducted with Microsoft Azure ND MI300X v5 showed that these models could be deployed across multiple GPUs within a single VM. This approach may reduce the need to deploy across multiple VMs and potentially lower performance-related costs.

Red Hat and AMD are also collaborating in the upstream vLLM community, an effort aimed at delivering increased inference performance and improved support for multi-GPU applications on AMD hardware. This collaboration includes technical contributions such as upstreaming the AMD kernel library and optimisations to the Triton kernel and FP8 support, allowing for faster and more efficient execution of vLLM workloads on AMD Instinct MI300X GPUs.

Enhanced collective communication and workload optimisation for multi-GPU environments are also being targeted, which could offer scalability and improved energy efficiency for AI deployments dependent on distributed computing. Collaborative efforts between Red Hat, AMD, and other industry parties such as IBM are intended to accelerate development within the vLLM project and benefit users relying on AMD hardware for AI inference and training.

The companies indicate that AMD Instinct GPUs will provide out-of-the-box support for Red Hat AI Inference Server, a distribution of vLLM designed for enterprise-grade requirements. As the primary commercial contributor to the vLLM project, Red Hat emphasises its commitment to compatibility across various hardware platforms, including those from AMD.

The collaboration also extends to AMD EPYC CPUs, which have been validated for Red Hat OpenShift Virtualization. This capability supports the running of VM workloads in an environment that aims to unify VMs and containerised applications, whether on-premises, in public clouds, or across a hybrid cloud setup.

Red Hat OpenShift Virtualization's validation for AMD EPYC processors allows the leveraging of these CPUs' performance and power efficiency characteristics, with support for leading server platforms including Dell PowerEdge, HPE ProLiant, and Lenovo ThinkSystem. The aim is to achieve higher infrastructure consolidation ratios, which could result in a reduced total cost of ownership (TCO) relating to hardware, software licensing, and energy.

Ashesh Badani, Senior Vice President and Chief Product Officer at Red Hat, said: "Fully realizing the benefits of AI means that organizations must have the choice and flexibility to optimize their IT footprint for the rigors of scaling demand. Our extended collaboration with AMD expands the spectrum of options for organizations seeking to ready their IT environments for an ever-evolving future, from modernizing existing investments on a high-performing CPU architecture and virtualization platform to preparing for production AI with next-generation hardware accelerators and open source AI technologies."

Philip Guido, Executive Vice President and Chief Commercial Officer at AMD, said: "As enterprise customer workloads grow more diverse and demanding, they require solutions that can scale. By combining Red Hat's industry-leading open source platforms with world-class AMD Instinct GPUs and AMD EPYC CPUs, we're delivering the performance and efficiency customers demand to accelerate AI, virtualization and hybrid-cloud innovation."

The companies are aiming to provide customers with a comprehensive platform capable of supporting both traditional workloads and demanding AI applications across the hybrid cloud.