Red Hat has announced the GA release of Red Hat Enterprise Linux AI 1.2. RHEL AI was announced earlier this year as Red Hat's AI solution for [1]a foundation model platform to develop / test / run Granite GenAI models . Not to be confused with the RHEL operating system itself, RHEL AI is all about building large language models for enterprise software with Granite LLMs and InstructLab tooling.

With today's release of RHEL AI 1.2, Red Hat is introducing initial support for AMD Instinct accelerators as a "technology preview" level feature. RHEL AI now supports the full AMD ROCm open-source compute stack. The RHEL AI 1.2 tech preview support for ROCm covers AMD Instinct MI210 GPUs for inference tasks and then the AMD Instinct MI300X GPUs for both training and inference AI workloads.

Great seeing RHEL AI finally get onboard with AMD GPU support albeit limited to the recent Instinct hardware only. Up to now RHEL AI had supported CPU based execution and, of course, NVIDIA hardware. RHEL AI 1.2 also add auto-detection of available hardware accelerators to ease in the support.

Red Hat Enterprise Linux AI 1.2 also adds tech preview level availability on Microsoft Azure and Google Cloud Platform. Some of the other changes include Lenovo ThinkSystem SR675 v3 server support, training checkpoint and resume handling, and enhanced training with PyTorch FSDP.

More details on RHEL AI 1.2 via [2]the Red Hat blog .



[1] https://www.phoronix.com/news/Red-Hat-RHEL-AI

[2] https://www.redhat.com/en/blog/whats-new-red-hat-enterprise-linux-ai-12