←Story detail

Trending story

NVIDIA Releases AICR Recipe Family for Vera Rubin NVL72

1 reports1 story updatesUpdated Nov 3 02:00

Understand this story

AI overview

On November 3, 2026, NVIDIA released the AI Cluster Runtime (AICR) recipe family for Vera Rubin NVL72, which uses reusable recipes to lock in tuning configurations, generate deployment packages, and validate running clusters, enabling Kubernetes environments to support Vera Rubin NVL72 from Day 1. This recipe series locks in configurations including the R615.23 open-source kernel module driver, DRA driver v0.5.0 (including ComputeDomains), and the 64k-page kernel of Ubuntu 26.04, and completed end-to-end validation on a reference cluster using GPU Operator v26.7.0, with cross-node NCCL all-reduce reaching 744.71 GB/s bus bandwidth at 16 GiB. Early chip validation also found that go-nvlib did not map the Rubin architecture value, causing the DRA driver's kubelet plugin to crash and restart; the relevant fix has entered the official release of DRA driver v0.5.0. There are currently no further developments.

AI-generated from public reports

Latest development · Report

NVIDIA AICR brings Day 1 support for Vera Rubin to Kubernetes

Developments

1 confirmed developments

  1. NVIDIA releases AICR recipe family for Vera Rubin NVL72

    Source:NVIDIA Developer · Generative AI

Reports and discussion timeline

Browse reports and discussion by time. Their evidence levels remain distinct.

Nov 3

  1. NVIDIA Developer · Generative AIFirst-party

    NVIDIA AICR brings Day 1 support for Vera Rubin to Kubernetes

    NVIDIA AI Cluster Runtime (AICR) uses reusable recipes to codify tuning configurations, generate deployment packages, and validate running clusters, helping Kubernetes environments support Vera Rubin NVL72 from Day 1. The Vera Rubin recipe series locks in configurations including the R615.23 open-source kernel module driver, DRA driver v0.5.0 (with ComputeDomains), and the 64k-page kernel of Ubuntu 26.04; end-to-end validation was completed on a reference cluster with GPU Operator v26.7.0, where cross-node NCCL all-reduce reached 744.71 GB/s bus bandwidth at 16 GiB. Early chip validation also found that go-nvlib did not map the Rubin architecture value, causing the DRA driver kubelet plugin to crash and restart; the fix has been included in the official release of DRA driver v0.5.0.

Related stories

  1. Related storiesNVIDIA Expands Cosmos Platform Training Tools and Data Blueprints
  2. Related storiesHao AI Lab open-sources the JetSpec++ model, code, and inference engine
  3. Related storiesMistral releases Large 4 weights on Hugging Face
  4. Related storiesNVIDIA Releases Open-Weight Speech Transcription Model Nemotron 3.5 Transcribe