Red Hat and AMD have announced a significant strategic collaboration aimed at enhancing processing power and performance for artificial intelligence workloads across hybrid cloud environments.
This partnership leverages AMD’s advanced GPU architectures with Red Hat’s industry-leading open source platforms to deliver optimized, cost-efficient AI solutions for enterprises navigating the rapidly evolving artificial intelligence landscape.
The collaboration addresses the growing demand for AI capabilities as organizations face escalating workload requirements.
With traditional data centers primarily dedicated to conventional IT systems, there is limited capacity to support intensive AI workloads.
This strategic alliance combines Red Hat’s expertise in open source solutions with AMD’s comprehensive portfolio of high-performance computing architectures to meet these emerging challenges.
AMD Instinct GPUs Power AI on Red Hat OpenShift AI
AMD Instinct GPUs are now fully enabled on Red Hat OpenShift AI, providing customers with high-performing processing power necessary for AI deployments without extreme resource requirements.
The integration includes support for AMD Instinct MI300X GPUs with Red Hat Enterprise Linux AI, which has been successfully tested on Microsoft Azure ND MI300X v5 instances for AI inferencing across both small language models (SLMs) and large language models (LLMs).
The AMD GPU Operator simplifies deployment and management of AMD Instinct GPU accelerators within Kubernetes clusters, enabling seamless configuration of GPU-accelerated workloads, including machine learning and generative AI applications.
Organizations can now deploy the AMD GPU Operator using OpenShift’s Operator Lifecycle Manager (OLM), which requires prerequisites including Node Feature Discovery (NFD) Operator and Kernel Module Management (KMM) Operator.
Implementation involves creating a DeviceConfig custom resource to trigger GPU driver installation:
Red Hat and AMD are collaborating in the upstream vLLM community to foster more efficient AI inference capabilities.
Through this upstream activation, the partnership aims to deliver improved performance on AMD GPUs by optimizing various components including the Triton kernel and FP8 support, enabling faster execution of vLLM on AMD Instinct MI300X accelerators.
The collaboration focuses on enhanced multi-GPU support through improved collective communication and optimized multi-GPU workloads, opening doors to more scalable and energy-efficient AI deployments.
This is particularly beneficial for workloads requiring distributed computing across multiple GPUs, reducing bottlenecks and improving overall throughput.
Red Hat AI Inference Server, the enterprise-grade distribution of vLLM, will support AMD Instinct GPUs out-of-the-box. As the top commercial contributor to vLLM, Red Hat ensures compatibility when deploying vLLM on various hardware choices, including AMD Instinct GPUs.
Datacenter Modernization
Beyond AI acceleration, the partnership addresses datacenter modernization through Red Hat OpenShift Virtualization, validated for AMD EPYC processors.
This integration helps enterprises optimize application deployment on leading server platforms while maintaining a bridge to cloud-native futures.
Red Hat OpenShift Virtualization on AMD EPYC CPUs enables high infrastructure consolidation ratios, potentially leading to significantly lower total cost of ownership across hardware, software licensing, and energy dimensions.
This strategic approach allows IT teams to manage critical workloads while freeing resources for AI workloads.
The collaboration demonstrates both companies’ commitment to expanding customer choice in building, deploying, and managing AI workloads across hybrid cloud environments, ultimately democratizing access to advanced AI technologies for enterprises of all sizes.
Try in-depth sandbox malware analysis for your SOC team. Get ANY.RUN special offer only until May 31 -> Try Here
