Dell and IBM subsidiary Red Hat are packaging a PowerEdge R760xa server with Red Hat Enterprise Linux AI (RHEL AI) to run large language models for customers embracing generative AI.
The 2RU R760xa server has up to 2 x 5th gen Xeon 64-core CPUs, PCIe gen 5 interconnects, supports Nvidia A10/A30/A40, L4/L40/L40S, A100 and H100 GPUs, and 8 x NVMe or 6 x SATA storage drive bays. It uses Nvidia’s Omniverse OVX 3.0 platform and is part of Dell’s AI Factory offering. RHEL AI is standard Red Hat Enterprise Linux enhanced with AI features.
Red Hat exec Joe Fernandes, VP and GM for Generative AI Foundation Platforms, stated: “By collaborating with Dell Technologies to validate and empower RHEL AI on Dell PowerEdge servers, we are enabling customers with greater confidence and flexibility to harness the power of GenAI workloads across hybrid cloud environments and propel their business into the future.”
Dell Technologies SVP Arun Narayanan said: “Validating RHEL AI for AI workloads on Dell PowerEdge servers provides customers with greater confidence that the servers, GPUs and foundational platforms are tested and validated on an ongoing basis. This simplifies the GenAI user experience and accelerates the process to build and deploy critical AI workloads on a trusted software stack.”
RHEL AI, included as part of Red Hat OpenShift AI, has two key features. The first is IBM Research’s open source Granite LLM set, and the second is a pair of InstructLab items, model alignment tools based on the LAB (Large-scale Alignment for chatBots) methodology and the InstructLab project.
The latter refers to model development with a community approach aimed at enhancing GenAI LLMs, and was created by IBM and Red Hat. The two say it can “enhance an LLM using far less human-generated information and far fewer computing resources than are typically used to retrain a model. And it makes it possible for upstream contributions to continuously make the model better.”
Red Hat’s OpenShift AI is a hybrid cloud machine learning operations (MLOps) platform for running models and InstructLab at scale across distributed cluster environments.
RHEL AI on Dell PowerEdge servers will be available in Q3 2024.
