Red Hat and AMD to Boost Cloud AI and Virtualization

Big news for businesses navigating the complex world of AI and virtual machines. Red Hat and AMD are deepening their strategic partnership to give customers more choices and power across the hybrid cloud. This collaboration aims to make deploying AI models more efficient and modernizing traditional virtual machines (VMs) more cost-effective.

Why This Matters Now

As AI demands soar, organizations are struggling to keep up with the intensive computational needs. Most data centers are built for traditional IT, leaving little room for AI workloads. That’s where Red Hat and AMD come in, combining Red Hat’s open-source expertise with AMD’s high-performance computing architectures to tackle these challenges head-on.

Driving More Efficient Generative AI

Red Hat AI is joining forces with AMD’s x86 processors and GPU architectures to create optimized, production-ready environments for AI workloads.

• AMD Instinct GPUs on Red Hat OpenShift AI: This powerful combination gives customers the high-performance processing needed for AI deployments across the hybrid cloud, without requiring extreme resources.
• Scaling Language Models: Red Hat and AMD successfully demonstrated AI inferencing for both small and large language models (SLMs and LLMs) using AMD Instinct MI300X GPUs with Red Hat Enterprise Linux AI on Microsoft Azure. This setup allowed deployment across multiple GPUs on a single VM, significantly reducing the need for multiple VMs and lowering performance costs.

Enhancing AI Performance with vLLM

Red Hat and AMD are actively collaborating in the upstream vLLM community to accelerate AI inference performance and tuning capabilities. This collaboration aims to deliver:

• Improved Performance on AMD GPUs: By optimizing components like the Triton kernel and FP8, and upstreaming the AMD kernel library, they’re boosting inference performance for both dense and quantized models, leading to faster execution on AMD Instinct MI300X accelerators.
• Enhanced Multi-GPU Support: Improvements in collective communication and multi-GPU workload optimization will lead to more scalable and energy-efficient AI deployments, especially for distributed computing tasks.
• Expanded vLLM Ecosystem Engagement: This cross-industry collaboration (including IBM) will accelerate upstream development, benefiting vLLM users who rely on AMD hardware for AI inference and training.

Building on this, AMD Instinct GPUs will support Red Hat AI Inference Server (Red Hat’s enterprise-grade distribution of vLLM) out-of-the-box. As a top commercial contributor to vLLM, Red Hat is committed to ensuring compatibility when deploying vLLM on your hardware of choice, including AMD Instinct GPUs, for outstanding optimization and performance.

It’s also worth noting that AMD EPYC CPUs are ideal for hosting GPU-enabled systems, which can further improve performance and ROI for even the most demanding AI workloads.

Transforming the Modern Data Center

Optimizing existing data center footprints allows organizations to reinvest resources into AI innovation. Red Hat OpenShift Virtualization, a feature of Red Hat OpenShift, provides a streamlined way to migrate and manage VM workloads with cloud-native simplicity.

• Validated for AMD EPYC Processors: Red Hat OpenShift Virtualization leverages the excellent performance and power efficiency of AMD EPYC processors across the hybrid cloud, while providing a clear path to a cloud-native future.
• Optimizing Application Deployment: This solution on AMD EPYC CPUs helps enterprises optimize application deployment on leading servers like Dell PowerEdge, HPE ProLiant, and Lenovo ThinkSystem products.
• Lowering Total Cost of Ownership (TCO): When refreshing legacy data centers, Red Hat OpenShift Virtualization unifies VMs and containerized applications, whether on-premise, in public clouds, or across the hybrid cloud. This can lead to significantly lower TCO in terms of hardware, software licensing, and energy, freeing up IT teams to manage current workloads and apply resources to future AI initiatives.

This collaboration between Red Hat and AMD is set to provide more robust, efficient, and flexible solutions for businesses looking to harness the full potential of AI and optimize their virtualization strategies.